master's thesis study of flow field in rectangular sump models...

83
Master's Thesis Study of Flow Field in Rectangular Sump Models and Performance Analysis of a Mixed Flow Pump Supervisor: Prof. Young-Ho LEE August 2014 Department of Mechanical Engineering Graduate School of Korea Maritime and Ocean University Yuxin ZHAO

Upload: others

Post on 08-Apr-2020

0 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

Masters Thesis

Study of Flow Field in Rectangular Sump

Models and Performance Analysis of a

Mixed Flow Pump

Supervisor Prof Young-Ho LEE

August 2014

Department of Mechanical Engineering

Graduate School of Korea Maritime and Ocean University

Yuxin ZHAO

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

We certify that we have read this thesis and that in our opinion it is satisfactory in

scope and quality as a thesis for the degree of Master of Mechanical Engineering

submitted by Yuxin ZHAO

THESIS COMMITTEE

Chairperson Prof Dr Jae-Hyun JEONG

Department of Mechanical Engineering

Graduate School of Korea Maritime and Ocean University

___________________________________

Co-Chairperson Prof Dr Hyung-Ho JEONG

Department of Mechanical Engineering

Graduate School of Korea Maritime and Ocean University

__________________________________

Supervisor Prof Dr Young-Ho LEE

Department of Mechanical Engineering

Graduate School of Korea Maritime and Ocean University

___________________________________

19 June 2014

Department of Mechanical Engineering

Graduate School of Korea Maritime and Ocean University

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

본 論文 조우흠 工學碩士

學位論文 로 認准함

원 공학박사 재현 ( )

원 공학박사 형호 ( )

원 공학박사 영호 ( )

2014 년 6 월 19

한 해양대학 대학원

기계공학과

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

i

TABLE OF CONTENTS

List of Tablesiii

List of Figuresiv

Abstractvi

Nomenclatureviii

Abbreviationsix

Chapter 1 Introduction1

11 Background1

12 Previous study1

13 Methodology of study3

Chapter 2 Basic theory of sump model4

21 Pump intake design5

211 Inlet bell diameter design 5

212 Recommended dimensions for a rectangular sump7

22 Model test10

221 The scale effects11

222 Principle of similarity12

223 Preliminary model test13

224 Test criteria19

225 Remedial measures for problem intakes20

Chapter 3 Methodology25

31 Numerical modeling25

32 Turbulence models26

321 k-ε turbulence model27

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

ii

322 Wilcox k-ω turbulence model28

323 Shear stress transport model29

33 Brief introduction of simulation process31

331 Pre-processing technique31

332 Solving the simulation32

333 Post processing32

Chapter 4 Description of Model Cases33

41 Creating the geometry33

411 Geometry of the scaled sump model33

412 Geometry of the full size pump sump model35

42 Mesh generation39

43 Numerical approach42

44 Experimental setup45

Chapter 5 Results and Discussion47

51 Results of the scaled sump model47

511 Numerical simulation vortex check47

512 PIV results analysis49

513 Sump with anti free surface vortex devices53

52 Results of the mixed flow pump sump model56

521 Performance analysis of the mixed flow pump sump model56

522 Flow characteristics analysis58

523 Cavitation phenomenon analysis62

Chapter 6 Conclusions60

Acknowledgement67

References68

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

iii

List of Tables

Table 21 Acceptable velocity ranges for inlet bell diameter 6

Table 22 Recommended dimensions for a rectangular sump8

Table 23 Permission criteria of ANSIHI20

Table 41 Designed specifications of scaled sump model33

Table 42 Parameters of anti free surface vortex devices35

Table 43 Design specifications of the mixed flow pump38

Table 44 Mesh information for the pump sump model42

Table 51 Swirl angle of model test results54

Table 52 Summary of swirl angle CFD calculation results54

Table 53 Mixed flow pump performance results56

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

iv

List of Figures

Fig 21 Recommended inlet bell design diameter6

Fig 22 Recommended intake structure layout7

Fig 23 Filler wall details for proper bay width7

Fig 24 Typical vortex in a sump14

Fig 25 Vortex classifications15

Fig 26 Typical swirl meter17

Fig 27 Schematic view of swirl angle numerical calculation 18

Fig 28 Example of Pitot tube 19

Fig 29 Measuring point at bell throat19

Fig 210 Examples of approach flow patterns21

Fig 211 Methods to reduce submerged vortices23

Fig 41 Geometry of scaled sump model33

Fig 42 Schematic diagram of sump with anti free surface vortex devices34

Fig 43 Overall 3D drawing of the sump model35

Fig 44 2D drawing of the sump sketch36

Fig 45 Shape of the mixed flow pump37

Fig 46 Sump model with the submerged AVD38

Fig 47 Submerged AVD dimensions39

Fig 48 Overall mesh generation for scaled sump models40

Fig 49 Overall sump fluid domain mesh for the pump sump model41

Fig 410 Grid details of the pump sump model41

Fig 411 Boundaries of the scaled sump43

Fig 412 Experimental setup for the scaled sump model46

Fig 51 Streamline pattern of free surface47

Fig 52 Vortex core region of the sump model48

Fig 53 Velocity vectors on different sections48

Fig 54 Velocity vectors of PIV results51

Fig 55 Velocity vectors on different sections (PIV)52

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

v

Fig 56 Velocity vectors on free surface of different cases53

Fig 57 Tangential velocity profile of different models55

Fig 58 Performance curve of the mixed flow pump with the sump 57

Fig 59 Streamline pattern around the intake structure58

Fig 510 Vortex core region of the pump sump model59

Fig 511 Vorticity distribution in the flow direction60

Fig 512 Vorticity distribution in channel width direction61

Fig 513 Cavitation region of the mixed flow pump62

Fig 514 Cavitation performance curves of the mixed flow pump63

Fig 515 Vapor volume fraction distribution on blade surface64

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

vi

Study of Flow Field in Rectangular Sump Models and

Performance Analysis of a Mixed Flow Pump

Yuxin ZHAO

Department of Mechanical Engineering

Graduate School of Korea Maritime and Ocean University

Abstract

It is an accepted fact that quite a number of problems faced in a pumping station are

related to the design of sump or intake rather than pump design Head-capacity

curves provided by the pump manufacturer are obtained on the condition of no

vortices flowing into the pump intake The efficiency and performance of pumping

stations depend not only on the performance of the selected pumps but also on the

proper design of the intake sumps A faulty design of pump sump can lead to the

occurrence of swirl and vortices which reduce the pump performance and induce

vibration and additional noise Therefore sump model test is necessary to check the

flow condition around intake structure Numerical simulation is a good facility for

reducing the time and cost involved throughout the design process In this study the

commercial software ANSYS CFX-130 has been used for the CFD analysis of the

sump models

In a scaled sump model for air entrainment simulation numerical analysis of single

phase and two-phase with SST turbulence model was carried out to predict vortex

(both free surface vortex and submerged vortex) occurrence and location The

effectiveness of curtain walls and square bars to eliminate the free surface vortex is

evaluated Meanwhile the experiment including PIV system was performed in

KMOU to investigate the flow conditions around pump intake structure

In a full size sump model overall numerical analysis for the sump model with a

mixed flow pump installed was carried out Hydraulic performances of the mixed

flow pump for head rise shaft power pump efficiencies versus flow rate changed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

vii

from 50 to 140 of the design flow rate were studied by the performance curves

A trident shaped anti vortex device (AVD) composed of three wall fillets and one

center splitter was installed under the pump intake The effectiveness of the AVD

for the suppression of submerged vortex was evaluated In addition numerical

simulation of cavitation phenomenon in the mixed flow pump has been performed

by calculating the full cavitation model with k-ε turbulence model

According to the results details of the location size and strength of vortices were

predicted in the numerical simulation For the scaled sump model although there is

a little difference between the single phase and two-phase simulation all the results

predict the free surface vortex and submerged vortex formation and location The

CFD simulations of flow condition show good agreement with the PIV results The

effectiveness of curtain walls installed to inhibit the free surface vortex is confirmed

in the CFD results while the square bars installed in the channel failed to suppress

the free surface vortex For the mixed flow pump with sump the best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

The effectiveness of AVDs installed for the submerged vortex is confirmed by

comparisons of the sump model with and without AVD results

Cavitation phenomenon analysis shows the pump operating at the BEP condition

has a considerable extent over the requirement of cavitation performance With

numerical simulation the inception of cavitation in the blade passage is observed on

the suction surface where the leading edges meet the tips and then as the inlet total

pressure decrease the cavitation zone is spread out over the suction surface as well

as the leading edge of the impeller blade

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

viii

Nomenclature

B Distance from the back wall to the pump inlet bell centerline [m]

C Distance between the inlet bell and floor [m]

D Inlet bell design outside diameter [m]

Fr Froude number (=Vb(gD)05) [-]

g Acceleration due to gravity [ms2]

ht Total head [m]

n Number of revolution per min [-]

Pflow Flow power [W]

Pshaft Shaft power [W]

Pν Vapor pressure [Pa]

Q Volume flow rate [m3s]

Q0 Design volume flow rate [m3s]

Re Reynolds number (=VbDν) [-]

S Minimum submergence depth [m]

Vb Velocity at bell mouth [ms]

α Angle of floor slope [deg]

η Pump efficiency []

θ Swirl angle [deg]

λ The scale ratio of model dimensions to prototype [-]

μ Viscosity [Pa s]

ν Kinematic viscosity [m2s]

ρ Specific water density [kgm3]

σ Surface tension [Nm]

τ Stress tensor [-]

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

ix

Abbreviations

AVD Anti vortex device

BEP Best efficiency point

CFD Computational fluid Dynamics

NPSH Net positive suction head

NPSHR Required net positive suction head

PIV Particle Image Velocimetry

RANS Reynolds Averaged Navier- Stokes

RPM Number of revolutions per minute

SST Shear Stress Transport

VOF Volume of Fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

1

Chapter 1 Introduction

11 Background

It is an accepted fact that quite a number of problems faced in a pumping station are

related to the design of sump or intake rather than pump design As an important

part of the pumping station sump is designed to provide uniform swirl and

entrained air free flow to the pump intake Undesirable flow conditions (entrained

air non-uniform flow distribution vortices etc) can reduce the pump efficiency

induce vibration noise and cavitation even lead to excessive bearing loads of

impeller and structure damage Due to quite complex nature of the flow although

the flow conditions causing pump problems are well established there are no

specific solutions to eliminate them There are some design guidelines for the

specific geometrical and hydraulic constraints of the pump sump for any project

the model study is the only tool for solving potential problems in new designs and

rectifying problems observed in existing installations However the scaled model

tests are expensive and time-consuming so alternative Computational Fluid

Dynamics (CFD) methods for evaluating sump performance have been developed

With rapid progress in CFD numerical simulation is regarded as an effective tool in

solving fluid problems in pump sumps

12 Previous study

Various researches show the flow conditions around the sump intake structure are

quite complicated Shyam et al [1] used the commercial code CFX to carry out

two-phase flow simulation to capture air entrainment The air entrainments and its

location were well captured and numerical results were in accordance with the

experimental investigation Matsui J[2] and Okamura T[3] et al conducted a serial

research of the benchmark according to the Turbomachinery Society of Japan (TSJ)

standard including various commercial codes comparison performed in the flow

simulation and comparison analysis between the numerical and experimental results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

2

The purpose of the benchmark is to determine the accuracy and reliability of the

CFD codes used in universities and the industrial field It is also intended to identify

and investigate the formation of air-entrained and submerged vortices near the

pump Tanweer et al[4] have performed a numerical analysis of a multiple intake

pump sump to confirm that the geometry of sump affects the flow within sump

greatly Rajendran et al[5] have made a numerical analysis for the flow

characteristics of a sump model with pump intake and good agreements were

achieved by comparing the numerical results with the experiments Iwano et al[6]

have introduced a numerical method for the submerged vortex by analyzing the

flow in the pump sump with and without baffle plates Lee et al[7] have conducted

the CFD analysis of a multi-intake pump sump model to check the flow uniformity

by predicting the location number and vorticity of the vortex For the CFD

application in mixed flow pump design and performance analysis Oh H W [8]

conducted a mixed flow pump design optimization with specific parameters and

then compared the numerical results with the experiments The procedure presented

can be used as a practical design and analysis guide for the general turbomachinery

Kim et al[9] carried out a mixed flow pump performance analysis with numerical

simulation and improved the hydraulic efficiency by the modification of the

geometry

As an undesirable phenomenon caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage The cavitation

phenomenon has been studied by various investigators Van et al[10] have

investigated the cavitation inception behavior of a mixed-flow pump impellers

model and the calculated results matched well with the experiments Okamura et

al[11] have evaluated two numerical cavitation prediction methods used in the

pump industry manufacturers and attempts were made to improve the cavitation

performance Li et al[12] have conducted the numerical analysis on the basis of the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

3

development liquidvapor interface tracking method to predict the cavitation

characteristics with a centrifugal pump impeller model

Moreover there are a number of correlated studies [13]~[17]

13 Methodology of study

This study will investigate the flow phenomena of two sump models

A scaled sump model is adopted to carry out numerical simulation and PIV

experiment The flow conditions are predicted and the effectiveness of devices for

free surface vortex is evaluated by using numerical code a single intake sump

model with a mixed flow pump installed is modeled and the hydraulic performance

characteristics for the head rise shaft power and pump efficiencies versus flow rate

are studied In addition flow phenomena (vortices cavitation etc) are analyzed

and the effectiveness of an anti-vortex device (AVD) for the submerged vortex is

evaluated

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

4

Chapter 2 Basic theory of sump model

Specific hydraulic phenomena have been confirmed that they can lead to certain

common operational problems The common of the main problems are summarized

as follows

a Free surface vortices

b Submerged vortices

c Swirl of flow entering the pump

d Non-uniform distribution of velocity at the pump impeller

e Entrained air or gas bubbles

These problems encountered in the pump sump will affect the pump

performance and significantly increase the operational and maintenance costs In

order to identify sources of particular problems and find practical solution for

it the usual approach is to conduct the laboratory experiments on a scaled

physical model Typical design objectives are to ensure that a pump station is

designed according to best practice and conforms to the requirements set out in the

American National Standard for Pump Intake Design[18] These standards restrict

the degree to which the above-mentioned undesirable flow patterns may be present

The negative impact of each of these hydraulic conditions varies with pump specific

speed and size Application of the standard to design a sump does not generate a

problem free sump but provides a basis for the initial design

As there are no specific guidelines or criteria for design of trouble free intakes the

most common solution to potential problems in new design and rectification of

problems observed in existing designs is to construct a scaled model in a laboratory

observe and investigate the flow characteristics and propose modifications to the

intake geometry There are additional devices in the form of floor splitters or cones

back wall splitters fillets surface beams guide vanes etc aimed at controlling the

vortex and swirl formation Considerable prior experience and ingenuity are

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

5

required to solve the problems by this method and the cost and time required are

also significant

The intake structure refers to the channel leading from the water source which

may be a river or reservoir and all installations downstream including the pump

column or the intake pipe (the suction tube portion of a vertical intake) the

approach channel (upstream of pump bay) and the pump bay (bounded by the floor

the back wall side walls dividers walls separating adjacent pump column) Usually

the upstream end of the pump column has an inlet attachment called the suction bell

The function of the intake is to supply an evenly distributed flow of water to the

pump suction bell

Several types of pump intake basins are available and ANSIHI 98 provides

design requirements for each Intake structures can be categorized as being

clear liquids or solids-bearing liquids For clear liquids intakes are further

classified into rectangular formed circular and trench types as well as suction

tanks and cans For solid-bearing liquids trench type and rectangular wet wells are

usually considered

In this paper only rectangular intake structure for clear liquid is studied

21 Pump intake design

For pumps to achieve their optimum hydraulic performance across all operating

conditions the flow at the impeller must meet specific hydraulic conditions The

ideal flow entering the pump inlet should be of a uniform velocity distribution

without rotation and stable over time When designing a sump to achieve favorable

inflow to the pump or suction pipe bell various sump dimensions relative to the size

of the bell are required Using these standards of the distances to reduce the

probability of the occurence of strong submerged vortices and free-surface vortices

211 Inlet bell diameter design

The geometry is generally defined in terms of the pump inlet diameter as shown in

Table 21 and Fig 21 The bell diameter can be estimated based on an inlet pipe

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

6

Table 21 Acceptable velocity ranges for inlet bell diameter(ANSIHI 98)

Flow Rate

Qls

Recommended Inlet Bell Design

Velocityms

Acceptable Velocity

Range ms

Q<315 V=17 06leVle27

315 leQ<1260 V=17 09leVle24

Q ge1260 V=17 12leVle21

Fig 21 Recommended inlet bell design diameter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

7

212 Recommended dimensions for a rectangular sump

As the selected bell diameter has been determined the proportions of the inlet

structure can be estimated from Fig 22 and Fig 23 Table 22 shows the

recommended dimensions for a rectangular sump

Fig 22 Recommended intake structure layout

Fig 23 Filler wall details for proper bay width

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

8

Table 22 Recommended dimensions for a rectangular sump

Dimension Variable

Description Recommended value

A Distance from the pump inlet bell

centerline to the entrance A=5D minimum

a Length of constricted bay section near

the pump inlet 25D minimum

B Distance from the back wall to the

pump inlet bell centerline 075D

C Distance between the inlet bell and

floor 03D~05D

D Inlet bell outside diameter refer to section 211

H Minimum liquid depth H=S+C

h Minimum height of constricted bay

section near the pump inlet bell hgtH or 25D

S Minimum pump inlet bell submergence S=D(10+23Fr)

W Pump inlet bay entrance width 2D minimum

w Constricted bay width near the pump

inlet bell w=2D

X Pump inlet bay length 5D minimum

Y Distance from pump inlet bell

centerline to the through-flow traveling screen

4D minimum

Z1 Distance from sump inlet bell centerline to diverging walls

5D minimum

Z2 Distance from inlet bell centerline to

sloping floor 5D minimum

α Angle of floor slope -10deg~ +10deg

β Angle of wall convergence 0deg~ +10deg

empty Angle of convergence from constricted

area to by walls 10deg maximum

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

9

FD=Vb(gD)05 (21)

where

FD Froude number (dimensionless)

Vb velocity at suction inlet=FlowArea based on D

D Outside diameter of bell or pipe inlet

g gravitational acceleration

There is the design sequence summary for rectangular intake structures

middot Consider the flow patterns and boundary geometry of the body of liquid

from which the pump station is to receive flow Compare with the approach

flow condition and determine if a hydraulic physical model study is

required

middot Determine the number and size of pumps required to satisfy the range of

operating conditions likely to be encountered

middot Identify pump inlet bell diameter Refer to the section 211

middot Determine the bell-floor clearance 05D is a good preliminary design value

middot Determine the required bell submergence using the equation 21

middot Determine the minimum allowable liquid depth in the intake structure from

the sum of the floor clearance and the required bell submergence

middot Check bottom elevation near the entrance to the structure and determine if it

is necessary to slope the floor upstream of the bay entrance If the resulting

depth at the entrance to the intake structure is shallow then check to ensure

that gravity-driven flow is not restricted by the entrance condition

middot Check the pump bay velocity for the maximum single-pump flow and

minimum liquid depth with the bay width set to 2D If bay velocity exceeds

05 ms then increase the bay width to reduce the maximum velocity below

05 ms

middot If it is necessary to increase the pump bay width to greater than 2D then

decrease bay width in the vicinity of the pumps according to Fig 23

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

10

middot Compare cross-flow velocity (at maximum system flow) to average pump

bay velocity If cross-flow value exceeds 50 of the bay velocity a

physical hydraulic model study is necessary

middot Determine the length of the structure and dividing walls giving

consideration to minimum allowable distances to a sloping floor screening

equipment and length of dividing walls if dual flow traveling screens or

drum screens are to be used a physical hydraulic model study is required

middot If the final selected pump bell diameter and inlet velocity is within the

range given in section 211 then the sump dimensions (developed based on

the inlet bell design diameter) need not be changed and will comply with

these standards

22 Model test

The approach flow pattern in a pump intake facility is difficult to predict by

traditional mathematics or empirical formulae For large structures or those that

differ significantly from proven designs a model study is the only means to ensure

success A model study is used to identify adverse hydraulic conditions and

derive remedial measures for approach flow patterns generated by structures

upstream of the pump impeller Current model studies are not intended to

investigate flow patterns induced by the pump itself or the flow patterns within the

pump The objective of a model study is to ensure that the pump intake structure

generates favorable flow conditions at the inlet to the pump Evaluation for the

requirement of model test if

a Sump or piping geometry that deviates from this design standard

b Non-uniform or non-symmetric approach flow to the pump sump

exist

c The pumps have flows greater than 2520 ls per pump or the total

station flow with all pumps running would be greater than 6310 ls

d Situations that need ten times cost than the model study

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

11

Adverse hydraulic conditions that can affect pump performance include free and

submerged vortices swirl approaching the pump impeller flow separation at the

pump bell and a non-uniform axial velocity distribution at the suction

The purpose of model test is to examine these adverse hydraulic conditions and to

investigate measure against them

221 The scale effects

One possible difficulty in scale modeling is that it is impossible to reduce all

pertinent forces by the same factor With Froude-scale models the inertial and

gravitational forces are reduced similarly but the viscous and surface tension forces

cannot be simultaneously reduced The extra influence of these forces is known as

a ldquoscale effectrdquo A number of investigations have indicated that the scale effects for

the model are avoided if the Reynolds number based on the inlet flow and

submergence or intake diameter exceeds approximately 3x 104 and the Weber

number exceeds 150 [19]

In most engineering applications involving closed-conduit and open-channel flow

the Reynolds number limits are far exceeded and the flows are fully turbulent

In a sump model Froude number (Fr) and Reynolds number (Re) are the most

important non-dimensional parameters Some reduction in Re should be

introduced to the sump model However flow patterns are generally very similar

at high Re

No specific geometric scale rate is recommended but reasonable large

geometric scale to minimize viscous and surface tension scale effects and reproduce

the flow pattern

Re= uDν (22)

We= u2ρDσ (23)

where

u average axial velocity (ms)

ν kinematic viscosity of the liquid (m2s)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

12

ρ liquid density (kgm3)

σ surface tension of liquidair interface

According to the ANSIHI 98 the viscous scale effects on vortex may be negligible

if the resulting dimensionless numbers must meet these values Re>6times104 and

We>240

To allow practical visual observations of flow patterns accurate measurements of

swirl and velocity distribution and sufficient dimensional control the model scale

shall yield a bay width of at least 300mm a minimum liquid depth of at least

150mm and a pump throat or suction diameter of at least 80mm

222 Principle of similarity

In the similitude analysis the geometric and flow similarity requires the model and

the flow to be identical as the real model and flow This is to ensure the result

obtained from the model study is well presented in order to predict full scale

behavior Model involving a free surface are operated using Froude similarity since

the flow process is controlled by gravity and inertial forces

For similarity of the flow patterns the Froude number shall be equal in model and

prototype

Fr =

=

(24)

and as the scale ratio λ was selected in section 221 the following equation is

obtained

V = V times (

) = V times λ or Q = Q times (

) = Q times λ (25)

where Fr is the Froude number λ is the scale ratio of model dimensions to prototype

The test of 15 times Froude scaled flow should be conducted to ascertain the

potential scale affect on free surface vortices

Models of closed conduit piping system leading to a pump suction are not operated

based on Froude similitude However flow patterns are generally very similar at

high Re A minimum value of 1times105 for the Re at the pump suction is

recommended to ensure the flow patterns are correctly scaled

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

13

223 Preliminary model test

The initial design shall be tested first to identify any hydraulic problems

1) Vortex Formation Mechanisms

There are various investigations that provide insight into the fundamental processes

leading to the development of vortices both in experiment and numerical simulation

Shin et al [20] demonstrated that two basic mechanisms lead to inlet vortex

formation The first mechanism involves the development of an inlet vortex due to

the amplification of ambient vorticity in the approach flow as vortex lines

are convicted into the inlet The second mechanism involves the development of a

trailing vortex in the vicinity of the intake as a result of the variation in

circulation along the inlet For this second case a vortex can develop in a flow that

is irrotational upstream and the vortex development therefore does not depend

on the presence of ambient vorticity Shin et al investigation on kinematic

parameters indicate that the strength of an inlet-vortex or trailing vortex system

increases with decreasing distance from the surface However for an inlet in

an upstream irrotational flow two counter rotating vortices can still trail from

the rear of the inlet

2) Classification of vortex type

As illustrated in Fig 24 vortices in the vicinity of pump intakes may be adjacent to

the channel bottom or a channel wall (ie submerged vortices) or they may appear

adjacent to the free surface (ie free surface vortex)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

14

(a) Schematic diagram of vortices in a sump

(b) Experimental results of different vortex

Fig 24 Typical vortex in a sump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

15

For the model visual observation dye was injected around the intake structure near

the surface and submergence as Fig 24 (b) shows And the model test results of

vortex was classified as Fig 25

(a) Free Surface Vortices

(b) Submerged Vortices

Fig 25 Vortex classifications

There are a number of researches focused on the numerical theory to detect the

inception of a visible vortex As the diameters of vortices are usually much smaller

compared with the computing grid size to compensate for the lack of vortex

resolution in numerical simulation a stretching vortex model is applied to the local

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

16

flow field around semi-analytically identified vortex positions [21][22] The brief

explanations are as follows

a) Free surface vortex

The criteria expressed by equation 26 and 27 is applied to determine the

visible inception of a free surface vortex

≫ 1 (26)

=

gt 1 (27)

where P is the pressure drop at the vortex core and Ph is the static head at the

vortex element and f is the drag force acting at a unit length of water surface

vortex dimple based on Stokes approximation and caused by the down flow

Further f is the force acting on the bubble in the direction from the high

pressure region to the low pressure region like buoyancy

b) Submerged vortex

Equation 28 is applied to determine the visible inception of a submerged vortex

infin =

infin

infin gt 1 (28)

where is the static pressure at the vortex core infin is the atmospheric pressure

and is the critical cavitation inception pressure and nearly equal to the

saturated vapor pressure

3) Swirl angle

Swirl angle predicts the intensity of flow rotation Fig 26 shows experiment

calculation method with swirl meter and attempts of the CFD calculation method

were conducted as Fig 27 A four blade zero-pitch swirl meter which is supported

by a low friction bearing will be installed at about four suction pipe diameter (d)

downstream from the pump suction to measure swirl angle of pump approaching

flow For the swirl meter the tip to tip blade diameter is 075d and the length in

flow direction is 06 and one of the four blade is painted yellow as a reference to

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

17

count revolutions As flow swirl is usually unsteady the observation time of swirl

meter reading should be a continuous period of time

Fig 26 Typical swirl meter

(29)

(210)

(211)

Where qV flow rotational speed at swirl meter (ms)

zV average axial velocity at swirl meter (ms)

n Number of revolution per minute (rpm)

d Diameter of pipe at the swirl meter (m)

q flow rate (m3s)

zV

Vqq 1tan -=

)(60

smnd

Vtimestimes

=p

q

)(4

2

smd

qVz

times=

p

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

18

For the swirl angle calculating method with CFD the key point is to obtain the

average tangential velocity so the swirl check circles were created as illustrated in

Fig 27 The swirl check circle is located at the section of 4d height with diameter

of 025d 05d and 075d respectively

Swirl check circle

Fig 27 Schematic view of swirl angle numerical calculation

23

1 7501 501 250

acute++

=aringaringaring

N

i

N

i

NNVNVNV

Vi

q (212)

where

qV average tangential velocity (ms)

V circumferential velocity of check point in check circle

respectively (ms)

N check point total numbers in single check circle

4) Velocity

Velocity measurement at bell throat by Pitot tube was implemented by ANSIHI

method Typical Pitot tube configuration is shown in Fig 28 Eight measuring

points are defined as Fig 29 Time averaged velocity at eight points was obtained

and these eight data were again averaged to give throat sectional spatial velocity

average

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

19

Fig 28 Example of Pitot tube Fig 29 Measuring points at bell throat

224 Test criteria

There is the criteria for satisfactory performance of the sump model test and

summarized in Table 23

1) Free surface and submerged vortices

Free surface vortices entering the pump must be less severe than vortices with

coherent core Dye core vortices may be acceptable only if they occur less than 10

of the time or only for infrequent pump operating conditions

Submerged vortex which starts from the sump walls induce vibration and noise

problems Dye core submerged vortices extending into the pump bell mouth is not

allowed

2) Swirl angles

Swirl angle is used to check the flow rotation in the suction pipe The calculation

method is referred to section 223

Short-term maximum and long-term average values must be less than 5 degrees

Maximum short-term swirl angles up to 7 degrees may be acceptable only if they

occur less than 10 of the time or only for infrequent pump operating conditions

3)Velocity

Each point time-averaged velocity shall not be larger than 10 of sectional average

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

20

Standard deviation of time-varying velocity at each point was obtained and its

value shall not be larger than 10 of the time-averaged velocity at each point

1) Time-averaged

Variation[] = 100acute-

a

a

V

VV aV Averaged at section

2) Time-varying

Standard deviation= )1(

)(1i

2

-

-aring=

n

xxn

i

Variation[] = (Standard deviation X ) x 100

X Averaged at point

Table 23 Permission criteria of ANSIHI

Vortex occurence Swirl angle

Free surface vortex Submerged vortex

Observation

time Above 10 minutes Above 10 minutes

Above 10

minutes

Permission

criteria

Dye-core

Vortex(F3)

below 110 total

time

Dye-core

Vortex(S2)

below 110 total

time

Below

5degConditional

intermittent

below 7deg

225 Remedial measures for problem intakes

Usually in the preliminary test the worst flow condition is simulated And based on

the results remedial measures for problem intakes are discussed in this part

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

21

1) Overall approach flow patterns

The characteristics of flow approaching an intake structure is one of the foremost

considerations for the designer It is difficult to predict the effects of a given set of

flow conditions upstream from an intake structure Fig 210 depicts several typical

approach flow patterns of pump sump with different operating conditions

Fig 210 Examples of approach flow patterns

Generally speaking for the intake problems especially for the multiple pumps

sump there are the main measures using dividing walls placed between the pumps

as a partitioned structure trash racks with elongated bars guide piers or a number

of structural columns at the bay entrance can provide some assistance in controlling

cross-flow baffles flow distributors and screen mesh etc are applied to improve

the flow distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

22

2) Problems around pump intake

a) Free surface vortices

There are a number of factors that have an effect on the formation of surface

vortices To achieve a higher degree of certainty that objectionable surface vortices

do not form modifications can be made to intake structure to allow operation at

practical depths of submergence and the minimum submergence is referred in Table

22 the use of suction umbrellas vertical curtain walls and horizontal gratings

b) Submerged vortices

In the vicinity of intake bell there are the most complicated flow patterns and flow

makes the most changes in direction while maintaining a constant acceleration into

the pump bell to prevent local flow separation turbulence and vortex formation

Fig 211 represents a sample of various devices to inhibit the different submerged

vortex

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

23

Fig 211 Methods to reduce submerged vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

24

c) Preswirl

The approach flow distributions determine whether flow preswirl exists and a

sufficiently laterally skewed approach flow cause rotating around the intake bell

The most effective way of reducing preswirl is to establish a relatively uniform

approach flow within each pump bay by using the baffling schemes controlling

cross-flow and expanding concentrated flow discussed in the above part

To prevent non-uniform velocity distribution swirls and vortices necessary

modifications are conducted such as fore bay length baffle wall distributors

curtain wall mesh screen guide vane floor splitter wall fillet etc As the above

measures applied modification test will be conducted to check all the flow

conditions and compare with the preliminary results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

25

Chapter 3 Methodology

CFD (Computational fluid dynamics) is the technique which can design cycles to

better performance and reduce costs and time In CFD we solve the governing

equations of given physics (may be differential form or integral form) using some

numerical techniques like Finite Difference Method (FDM) Finite Element Method

(FEM) or Finite Volume Method (FVM) Although CFD plays a great part in the

design of turbomachinery CFD is not a replacement for experimental or analytical

approach In this study the numerical analysis was performed by the commercial

code ANSYS CFX 130 All aspects of the CFD study including modelling

meshing solving and post-processing of the results

31 Numerical modeling

ANSYS CFX solves the unsteady Navier-Stokes equations in their conservation

form The instantaneous equation of mass (continuity) in the stationary frame is

expressed as equation

ρ

δ + nabla ∙ (ρU) = 0 (31)

And the instantaneous equation for momentum is expressed as shown in equation

ρ

δ + nabla ∙ (ρU⨂U) = minusnablap + nabla ∙ τ + S (32)

Where τ is the stress tensor ( including both normal and shear components of the

stress)

These instantaneous equations are averaged for turbulent flows leading to additional

terms that need to be solved While the Navier-Stokes equations describe both

laminar and turbulent flows without addition terms realistic flows involve length

scales much smaller than the smallest finite volume mesh A Direct Numerical

Simulation of these flows would require significantly more computing power than

what is available now or in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

26

Therefore much research has been done to predict the effects of turbulence by

using turbulence models These models account for the effects of turbulence

without the use of a very fine mesh or direct numerical simulation

These turbulence models modify the transport equations by adding averaged and

fluctuating components The transport equations are changed to equations (33) and

(34)

ρ

δ + nabla ∙ (ρU) = 0 (33)

ρ

+ nabla ∙ ρ U ⨂U = minusnablap + nabla ∙ τ minus ρu⨂u + S (34)

The mass equation is not changed but the momentum equation contains extra terms

which are the Reynolds stresses ρ u⨂u and the Reynolds flux ρ u ϕ These

Reynolds stresses need to be modeled by additional equations to obtain closure

Obtaining closure implies that there are a sufficient number of equations to solve for

all the unknowns including the Reynolds stresses and Reynolds fluxes

Various turbulence models provide various ways to obtain closure In this

investigation Shear Stress Transport (SST) model and k-ε model were utilized The

advantage of SST model is that combines the advantages of other turbulence models

(the k-ε Wilcox k-ω) The 3 turbulence models will be discussed briefly

32 Turbulence models

Turbulence models are necessary because we cannot afford big enough computers

to directly capture every scale of motion There are three modelling frameworks

Direct Numerical Simulation (DNS) Reynolds-Averaged Navier-Stokes

Equations(RANS) Large Eddy Simulation(LES)

As RANS method is applied in most productionseveral RANS turbulence models

differ in their usage of wall functions the number of additional variables solved for

and what these variables represent

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

27

Two equation turbulence models

Two-equation turbulence models are widely used in CFD as they give a good

compromise between computational power needed and accuracy The term lsquotwo

equationrsquo refers to the fact that these models solve for the velocity and length scales

using separate transport equations The turbulence velocity scale is obtained by

solving the transport equation The turbulent length scale is estimated from two

properties of the turbulence field namely the turbulent kinetic energy and the

dissipation rate The dissipation rate of the kinetic energy is obtained from its

transport equation The most widely used are k-ε and k-ω two equation models In

the next section the k-ε Wilcox k-ω BSL k-ω and SST models will be briefly

discussed

321 k-ε turbulence model

The k-epsilon model solves for two variables k the turbulent kinetic energy and

epsilon the rate of dissipation of kinetic energy Wall functions are used in this

model so the flow in the buffer region is not simulated The k-epsilon model is very

popular for industrial applications due to its good convergence rate and relatively

low memory requirements It does not very accurately compute flow fields that

exhibit adverse pressure gradients strong curvature to the flow or jet flow It does

perform well for external flow problems around complex geometries

The k-ε model introduces k (m2s2) as the turbulence kinetic energy and ε (m2s3) as

the turbulence eddy dissipation The continuity equation remains the same

ρ

+ nabla ∙ (ρ U) = 0 (35)

The momentum equation changes as shown by equation

δρ

δ + nabla ∙ (ρU⨂U) = minusnablapprime + nabla ∙ μ

nablaU + (nablaU) + S (36)

Where SM is the sum of body forces μeff is the effective viscosity accounting for

turbulence and prsquo is the modified pressure The k-ε model uses the concept of eddy

viscosity giving the equation for effective viscosity as shown by equation

μ

= μ + μ (37)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

28

μt is the turbulence viscosity is linked to the turbulence kinetic energy and

dissipation by the equation

μ = Cμρ

ε (38)

Where Cμ is a constant

The values for k and ε come from the differential transport equations for the

turbulence kinetic energy and the turbulence dissipation rate

The turbulence kinetic energy equation is given as equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus ρε (39)

The turbulence dissipation rate is given by equation

(ρε)

+ nabla ∙ (ρUε) = nabla ∙ μ +

μ

σε nablaε +

ε

(Cε (P + Pε ) minus Cε ρε) (310)

Where Cε1 Cε2 σk σε are constants

Pk is the turbulence production due to viscous forces and is modeled by the

equation

P = μ nablaU ∙ nablaU + nablaU minus

nabla ∙ U 3μ

nabla ∙ U + ρk (311)

If a buoyancy term is added to the previous equation if the full buoyancy model is

used

The buoyancy term Pkb is modeled as

P = minusμ

ρσρg ∙ nablaρ (312)

322 Wilcox k-ω turbulence model

The k-omega model is similar to k-epsilon instead however it solves for omega mdash

the specific rate of dissipation of kinetic energy It also uses wall functions and

therefore has comparable memory requirements It has more difficulty converging

and is quite sensitive to the initial guess at the solution Hence the k-epsilon model

is often used first to find an initial condition for solving the k-omega model The k-

omega model is useful in many cases where the k-epsilon model is not accurate

such as internal flows flows that exhibit strong curvature separated flows and jets

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

29

This model has an advantage over the k-ε model where it does not involve complex

linear damping functions for near wall calculations at low Reynolds numbers

The k-ω model assumes that the turbulence viscosity is related to the turbulence

kinetic energy k and the turbulent frequency ω by the equation

μ = ρ

ω (313)

The transport equation for k is given by the equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus βprimeρ k ω (314)

The transport equation for ω is shown as equation

(ρω)

δ + nabla ∙ (ρ U ω) = nabla ∙ μ +

μ

μω

nablaω + αω

P + Pω minus β ρ ω (315)

The production rate of turbulence (Pk) is calculated as shown previously in the k-ε

section

The recommended values for the constants in the above equations are

βprime = 009

α = 59

β = 0075

σ = 2

σω = 2

The Reynolds stress tensor ρ u⨂u is calculated by

minusρ u⨂u = μ nablaU + (nablaU) minus

part ρk + μ

nabla ∙ U (316)

323 Shear stress transport model

The disadvantage of the Wilcox model is the strong sensitivity to free-stream

conditions Therefore a blending of the k-ω model near the surface and the k-ε in

the outer region was made by Menter [23] which resulted in the formulation of the

BSL k-ω turbulence model It consists of a transformation of the k-ε model to a k-ω

formulation and subsequently adding the resulting equations The Wilcox model is

multiplied by a blending function F1 and the transformed k-ε by another function 1-

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

30

F1 F1 is a function of wall distance (being the value of one near the surface and zero

outside the boundary layer) Outside the boundary and on the edge of the boundary

layer the standard k-ε model is used [24]

SST model is a combination of the k-epsilon in the free stream and the k-omega

models near the walls It does not use wall functions and tends to be most accurate

when solving the flow near the wall The SST model does not always converge to

the solution quickly so the k-epsilon or k-omega models are often solved first to

give good initial conditions

The k-ω based Shear Stress Transport (SST) model of Menter was applied for

turbulence treatment The transport equations for the SST model are expressed

below where the turbulent kinetic energy lsquokrsquo and turbulent frequency or dissipation

per unit turbulent kinetic energy are computed by using the following relations

For Turbulence Kinetic Energy

kkPukt

kTkk Ntilde+Ntilde+-=Ntilde+

para

para)[()( nsnwb (317)

where Pk is the production limiter

For Specific Dissipation Rate

ww

swnsnbwaww

ww NtildeNtilde-+Ntilde+Ntilde+-=Ntilde+para

para

1)1(2])[()( 21

22 kFSut

T (318)

The first blending function F1 is calculated from

iumlthorn

iumlyacute

uuml

iumlicirc

iumliacute

igrave

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

uacuteuacuteucirc

ugrave

ecircecirceuml

eacute

dividedivideoslash

oumlccedilccedilegrave

aelig=

4

22

21

4500maxmintanh

yCD

k

yy

kF

kw

wrs

w

n

wb (319)

y is the distance to the nearest wall and v is the kinematic viscosity

divideoslash

oumlccedilegrave

aeligNtildeNtilde= -10

2 101

2max ww

rsww kCDk (320)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

31

And Kinematic eddy viscosity

)max( 21

1

SF

kT

wa

an = (321)

where F2 is a blending function which restricts the limiter to the wall boundary and

S is the invariant measure of the strain rate

uacuteuacute

ucirc

ugrave

ecircecirc

euml

eacute

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

dividedivideoslash

oumlccedilccedilegrave

aelig=

2

22

5002maxtanh

w

n

wb yy

kF (322)

33 Brief introduction of simulation process

All aspects of CFD study including modelling meshing solving and post-

processing of the results are briefly introduced in this chapter

331 Pre-processing technique

The speed and accuracy of simulation results depending fundamentally on meshing

and the ability to prepare model geometry correctly

The steps in pre-processing are explained as follows

1) Domain Definitions

The domain definition depends mainly on the physics involved in the problem and

objective of the problem

a) Geometry creation or geometry import

b) Geometry cleanup

i) Removing the parts not necessary for simulation under consideration

ii) Closing the gaps in the geometry

iii) Removing small surfaces curves to make meshing process simple

2)Grid generation

To solve the governing equations the numerical methods like Finite Difference

Finite Volume or Finite Element Method to divide the domain into small parts and

solve the equations on one part at a time are applied

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 2: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

We certify that we have read this thesis and that in our opinion it is satisfactory in

scope and quality as a thesis for the degree of Master of Mechanical Engineering

submitted by Yuxin ZHAO

THESIS COMMITTEE

Chairperson Prof Dr Jae-Hyun JEONG

Department of Mechanical Engineering

Graduate School of Korea Maritime and Ocean University

___________________________________

Co-Chairperson Prof Dr Hyung-Ho JEONG

Department of Mechanical Engineering

Graduate School of Korea Maritime and Ocean University

__________________________________

Supervisor Prof Dr Young-Ho LEE

Department of Mechanical Engineering

Graduate School of Korea Maritime and Ocean University

___________________________________

19 June 2014

Department of Mechanical Engineering

Graduate School of Korea Maritime and Ocean University

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

본 論文 조우흠 工學碩士

學位論文 로 認准함

원 공학박사 재현 ( )

원 공학박사 형호 ( )

원 공학박사 영호 ( )

2014 년 6 월 19

한 해양대학 대학원

기계공학과

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

i

TABLE OF CONTENTS

List of Tablesiii

List of Figuresiv

Abstractvi

Nomenclatureviii

Abbreviationsix

Chapter 1 Introduction1

11 Background1

12 Previous study1

13 Methodology of study3

Chapter 2 Basic theory of sump model4

21 Pump intake design5

211 Inlet bell diameter design 5

212 Recommended dimensions for a rectangular sump7

22 Model test10

221 The scale effects11

222 Principle of similarity12

223 Preliminary model test13

224 Test criteria19

225 Remedial measures for problem intakes20

Chapter 3 Methodology25

31 Numerical modeling25

32 Turbulence models26

321 k-ε turbulence model27

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

ii

322 Wilcox k-ω turbulence model28

323 Shear stress transport model29

33 Brief introduction of simulation process31

331 Pre-processing technique31

332 Solving the simulation32

333 Post processing32

Chapter 4 Description of Model Cases33

41 Creating the geometry33

411 Geometry of the scaled sump model33

412 Geometry of the full size pump sump model35

42 Mesh generation39

43 Numerical approach42

44 Experimental setup45

Chapter 5 Results and Discussion47

51 Results of the scaled sump model47

511 Numerical simulation vortex check47

512 PIV results analysis49

513 Sump with anti free surface vortex devices53

52 Results of the mixed flow pump sump model56

521 Performance analysis of the mixed flow pump sump model56

522 Flow characteristics analysis58

523 Cavitation phenomenon analysis62

Chapter 6 Conclusions60

Acknowledgement67

References68

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

iii

List of Tables

Table 21 Acceptable velocity ranges for inlet bell diameter 6

Table 22 Recommended dimensions for a rectangular sump8

Table 23 Permission criteria of ANSIHI20

Table 41 Designed specifications of scaled sump model33

Table 42 Parameters of anti free surface vortex devices35

Table 43 Design specifications of the mixed flow pump38

Table 44 Mesh information for the pump sump model42

Table 51 Swirl angle of model test results54

Table 52 Summary of swirl angle CFD calculation results54

Table 53 Mixed flow pump performance results56

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

iv

List of Figures

Fig 21 Recommended inlet bell design diameter6

Fig 22 Recommended intake structure layout7

Fig 23 Filler wall details for proper bay width7

Fig 24 Typical vortex in a sump14

Fig 25 Vortex classifications15

Fig 26 Typical swirl meter17

Fig 27 Schematic view of swirl angle numerical calculation 18

Fig 28 Example of Pitot tube 19

Fig 29 Measuring point at bell throat19

Fig 210 Examples of approach flow patterns21

Fig 211 Methods to reduce submerged vortices23

Fig 41 Geometry of scaled sump model33

Fig 42 Schematic diagram of sump with anti free surface vortex devices34

Fig 43 Overall 3D drawing of the sump model35

Fig 44 2D drawing of the sump sketch36

Fig 45 Shape of the mixed flow pump37

Fig 46 Sump model with the submerged AVD38

Fig 47 Submerged AVD dimensions39

Fig 48 Overall mesh generation for scaled sump models40

Fig 49 Overall sump fluid domain mesh for the pump sump model41

Fig 410 Grid details of the pump sump model41

Fig 411 Boundaries of the scaled sump43

Fig 412 Experimental setup for the scaled sump model46

Fig 51 Streamline pattern of free surface47

Fig 52 Vortex core region of the sump model48

Fig 53 Velocity vectors on different sections48

Fig 54 Velocity vectors of PIV results51

Fig 55 Velocity vectors on different sections (PIV)52

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

v

Fig 56 Velocity vectors on free surface of different cases53

Fig 57 Tangential velocity profile of different models55

Fig 58 Performance curve of the mixed flow pump with the sump 57

Fig 59 Streamline pattern around the intake structure58

Fig 510 Vortex core region of the pump sump model59

Fig 511 Vorticity distribution in the flow direction60

Fig 512 Vorticity distribution in channel width direction61

Fig 513 Cavitation region of the mixed flow pump62

Fig 514 Cavitation performance curves of the mixed flow pump63

Fig 515 Vapor volume fraction distribution on blade surface64

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

vi

Study of Flow Field in Rectangular Sump Models and

Performance Analysis of a Mixed Flow Pump

Yuxin ZHAO

Department of Mechanical Engineering

Graduate School of Korea Maritime and Ocean University

Abstract

It is an accepted fact that quite a number of problems faced in a pumping station are

related to the design of sump or intake rather than pump design Head-capacity

curves provided by the pump manufacturer are obtained on the condition of no

vortices flowing into the pump intake The efficiency and performance of pumping

stations depend not only on the performance of the selected pumps but also on the

proper design of the intake sumps A faulty design of pump sump can lead to the

occurrence of swirl and vortices which reduce the pump performance and induce

vibration and additional noise Therefore sump model test is necessary to check the

flow condition around intake structure Numerical simulation is a good facility for

reducing the time and cost involved throughout the design process In this study the

commercial software ANSYS CFX-130 has been used for the CFD analysis of the

sump models

In a scaled sump model for air entrainment simulation numerical analysis of single

phase and two-phase with SST turbulence model was carried out to predict vortex

(both free surface vortex and submerged vortex) occurrence and location The

effectiveness of curtain walls and square bars to eliminate the free surface vortex is

evaluated Meanwhile the experiment including PIV system was performed in

KMOU to investigate the flow conditions around pump intake structure

In a full size sump model overall numerical analysis for the sump model with a

mixed flow pump installed was carried out Hydraulic performances of the mixed

flow pump for head rise shaft power pump efficiencies versus flow rate changed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

vii

from 50 to 140 of the design flow rate were studied by the performance curves

A trident shaped anti vortex device (AVD) composed of three wall fillets and one

center splitter was installed under the pump intake The effectiveness of the AVD

for the suppression of submerged vortex was evaluated In addition numerical

simulation of cavitation phenomenon in the mixed flow pump has been performed

by calculating the full cavitation model with k-ε turbulence model

According to the results details of the location size and strength of vortices were

predicted in the numerical simulation For the scaled sump model although there is

a little difference between the single phase and two-phase simulation all the results

predict the free surface vortex and submerged vortex formation and location The

CFD simulations of flow condition show good agreement with the PIV results The

effectiveness of curtain walls installed to inhibit the free surface vortex is confirmed

in the CFD results while the square bars installed in the channel failed to suppress

the free surface vortex For the mixed flow pump with sump the best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

The effectiveness of AVDs installed for the submerged vortex is confirmed by

comparisons of the sump model with and without AVD results

Cavitation phenomenon analysis shows the pump operating at the BEP condition

has a considerable extent over the requirement of cavitation performance With

numerical simulation the inception of cavitation in the blade passage is observed on

the suction surface where the leading edges meet the tips and then as the inlet total

pressure decrease the cavitation zone is spread out over the suction surface as well

as the leading edge of the impeller blade

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

viii

Nomenclature

B Distance from the back wall to the pump inlet bell centerline [m]

C Distance between the inlet bell and floor [m]

D Inlet bell design outside diameter [m]

Fr Froude number (=Vb(gD)05) [-]

g Acceleration due to gravity [ms2]

ht Total head [m]

n Number of revolution per min [-]

Pflow Flow power [W]

Pshaft Shaft power [W]

Pν Vapor pressure [Pa]

Q Volume flow rate [m3s]

Q0 Design volume flow rate [m3s]

Re Reynolds number (=VbDν) [-]

S Minimum submergence depth [m]

Vb Velocity at bell mouth [ms]

α Angle of floor slope [deg]

η Pump efficiency []

θ Swirl angle [deg]

λ The scale ratio of model dimensions to prototype [-]

μ Viscosity [Pa s]

ν Kinematic viscosity [m2s]

ρ Specific water density [kgm3]

σ Surface tension [Nm]

τ Stress tensor [-]

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

ix

Abbreviations

AVD Anti vortex device

BEP Best efficiency point

CFD Computational fluid Dynamics

NPSH Net positive suction head

NPSHR Required net positive suction head

PIV Particle Image Velocimetry

RANS Reynolds Averaged Navier- Stokes

RPM Number of revolutions per minute

SST Shear Stress Transport

VOF Volume of Fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

1

Chapter 1 Introduction

11 Background

It is an accepted fact that quite a number of problems faced in a pumping station are

related to the design of sump or intake rather than pump design As an important

part of the pumping station sump is designed to provide uniform swirl and

entrained air free flow to the pump intake Undesirable flow conditions (entrained

air non-uniform flow distribution vortices etc) can reduce the pump efficiency

induce vibration noise and cavitation even lead to excessive bearing loads of

impeller and structure damage Due to quite complex nature of the flow although

the flow conditions causing pump problems are well established there are no

specific solutions to eliminate them There are some design guidelines for the

specific geometrical and hydraulic constraints of the pump sump for any project

the model study is the only tool for solving potential problems in new designs and

rectifying problems observed in existing installations However the scaled model

tests are expensive and time-consuming so alternative Computational Fluid

Dynamics (CFD) methods for evaluating sump performance have been developed

With rapid progress in CFD numerical simulation is regarded as an effective tool in

solving fluid problems in pump sumps

12 Previous study

Various researches show the flow conditions around the sump intake structure are

quite complicated Shyam et al [1] used the commercial code CFX to carry out

two-phase flow simulation to capture air entrainment The air entrainments and its

location were well captured and numerical results were in accordance with the

experimental investigation Matsui J[2] and Okamura T[3] et al conducted a serial

research of the benchmark according to the Turbomachinery Society of Japan (TSJ)

standard including various commercial codes comparison performed in the flow

simulation and comparison analysis between the numerical and experimental results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

2

The purpose of the benchmark is to determine the accuracy and reliability of the

CFD codes used in universities and the industrial field It is also intended to identify

and investigate the formation of air-entrained and submerged vortices near the

pump Tanweer et al[4] have performed a numerical analysis of a multiple intake

pump sump to confirm that the geometry of sump affects the flow within sump

greatly Rajendran et al[5] have made a numerical analysis for the flow

characteristics of a sump model with pump intake and good agreements were

achieved by comparing the numerical results with the experiments Iwano et al[6]

have introduced a numerical method for the submerged vortex by analyzing the

flow in the pump sump with and without baffle plates Lee et al[7] have conducted

the CFD analysis of a multi-intake pump sump model to check the flow uniformity

by predicting the location number and vorticity of the vortex For the CFD

application in mixed flow pump design and performance analysis Oh H W [8]

conducted a mixed flow pump design optimization with specific parameters and

then compared the numerical results with the experiments The procedure presented

can be used as a practical design and analysis guide for the general turbomachinery

Kim et al[9] carried out a mixed flow pump performance analysis with numerical

simulation and improved the hydraulic efficiency by the modification of the

geometry

As an undesirable phenomenon caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage The cavitation

phenomenon has been studied by various investigators Van et al[10] have

investigated the cavitation inception behavior of a mixed-flow pump impellers

model and the calculated results matched well with the experiments Okamura et

al[11] have evaluated two numerical cavitation prediction methods used in the

pump industry manufacturers and attempts were made to improve the cavitation

performance Li et al[12] have conducted the numerical analysis on the basis of the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

3

development liquidvapor interface tracking method to predict the cavitation

characteristics with a centrifugal pump impeller model

Moreover there are a number of correlated studies [13]~[17]

13 Methodology of study

This study will investigate the flow phenomena of two sump models

A scaled sump model is adopted to carry out numerical simulation and PIV

experiment The flow conditions are predicted and the effectiveness of devices for

free surface vortex is evaluated by using numerical code a single intake sump

model with a mixed flow pump installed is modeled and the hydraulic performance

characteristics for the head rise shaft power and pump efficiencies versus flow rate

are studied In addition flow phenomena (vortices cavitation etc) are analyzed

and the effectiveness of an anti-vortex device (AVD) for the submerged vortex is

evaluated

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

4

Chapter 2 Basic theory of sump model

Specific hydraulic phenomena have been confirmed that they can lead to certain

common operational problems The common of the main problems are summarized

as follows

a Free surface vortices

b Submerged vortices

c Swirl of flow entering the pump

d Non-uniform distribution of velocity at the pump impeller

e Entrained air or gas bubbles

These problems encountered in the pump sump will affect the pump

performance and significantly increase the operational and maintenance costs In

order to identify sources of particular problems and find practical solution for

it the usual approach is to conduct the laboratory experiments on a scaled

physical model Typical design objectives are to ensure that a pump station is

designed according to best practice and conforms to the requirements set out in the

American National Standard for Pump Intake Design[18] These standards restrict

the degree to which the above-mentioned undesirable flow patterns may be present

The negative impact of each of these hydraulic conditions varies with pump specific

speed and size Application of the standard to design a sump does not generate a

problem free sump but provides a basis for the initial design

As there are no specific guidelines or criteria for design of trouble free intakes the

most common solution to potential problems in new design and rectification of

problems observed in existing designs is to construct a scaled model in a laboratory

observe and investigate the flow characteristics and propose modifications to the

intake geometry There are additional devices in the form of floor splitters or cones

back wall splitters fillets surface beams guide vanes etc aimed at controlling the

vortex and swirl formation Considerable prior experience and ingenuity are

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

5

required to solve the problems by this method and the cost and time required are

also significant

The intake structure refers to the channel leading from the water source which

may be a river or reservoir and all installations downstream including the pump

column or the intake pipe (the suction tube portion of a vertical intake) the

approach channel (upstream of pump bay) and the pump bay (bounded by the floor

the back wall side walls dividers walls separating adjacent pump column) Usually

the upstream end of the pump column has an inlet attachment called the suction bell

The function of the intake is to supply an evenly distributed flow of water to the

pump suction bell

Several types of pump intake basins are available and ANSIHI 98 provides

design requirements for each Intake structures can be categorized as being

clear liquids or solids-bearing liquids For clear liquids intakes are further

classified into rectangular formed circular and trench types as well as suction

tanks and cans For solid-bearing liquids trench type and rectangular wet wells are

usually considered

In this paper only rectangular intake structure for clear liquid is studied

21 Pump intake design

For pumps to achieve their optimum hydraulic performance across all operating

conditions the flow at the impeller must meet specific hydraulic conditions The

ideal flow entering the pump inlet should be of a uniform velocity distribution

without rotation and stable over time When designing a sump to achieve favorable

inflow to the pump or suction pipe bell various sump dimensions relative to the size

of the bell are required Using these standards of the distances to reduce the

probability of the occurence of strong submerged vortices and free-surface vortices

211 Inlet bell diameter design

The geometry is generally defined in terms of the pump inlet diameter as shown in

Table 21 and Fig 21 The bell diameter can be estimated based on an inlet pipe

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

6

Table 21 Acceptable velocity ranges for inlet bell diameter(ANSIHI 98)

Flow Rate

Qls

Recommended Inlet Bell Design

Velocityms

Acceptable Velocity

Range ms

Q<315 V=17 06leVle27

315 leQ<1260 V=17 09leVle24

Q ge1260 V=17 12leVle21

Fig 21 Recommended inlet bell design diameter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

7

212 Recommended dimensions for a rectangular sump

As the selected bell diameter has been determined the proportions of the inlet

structure can be estimated from Fig 22 and Fig 23 Table 22 shows the

recommended dimensions for a rectangular sump

Fig 22 Recommended intake structure layout

Fig 23 Filler wall details for proper bay width

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

8

Table 22 Recommended dimensions for a rectangular sump

Dimension Variable

Description Recommended value

A Distance from the pump inlet bell

centerline to the entrance A=5D minimum

a Length of constricted bay section near

the pump inlet 25D minimum

B Distance from the back wall to the

pump inlet bell centerline 075D

C Distance between the inlet bell and

floor 03D~05D

D Inlet bell outside diameter refer to section 211

H Minimum liquid depth H=S+C

h Minimum height of constricted bay

section near the pump inlet bell hgtH or 25D

S Minimum pump inlet bell submergence S=D(10+23Fr)

W Pump inlet bay entrance width 2D minimum

w Constricted bay width near the pump

inlet bell w=2D

X Pump inlet bay length 5D minimum

Y Distance from pump inlet bell

centerline to the through-flow traveling screen

4D minimum

Z1 Distance from sump inlet bell centerline to diverging walls

5D minimum

Z2 Distance from inlet bell centerline to

sloping floor 5D minimum

α Angle of floor slope -10deg~ +10deg

β Angle of wall convergence 0deg~ +10deg

empty Angle of convergence from constricted

area to by walls 10deg maximum

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

9

FD=Vb(gD)05 (21)

where

FD Froude number (dimensionless)

Vb velocity at suction inlet=FlowArea based on D

D Outside diameter of bell or pipe inlet

g gravitational acceleration

There is the design sequence summary for rectangular intake structures

middot Consider the flow patterns and boundary geometry of the body of liquid

from which the pump station is to receive flow Compare with the approach

flow condition and determine if a hydraulic physical model study is

required

middot Determine the number and size of pumps required to satisfy the range of

operating conditions likely to be encountered

middot Identify pump inlet bell diameter Refer to the section 211

middot Determine the bell-floor clearance 05D is a good preliminary design value

middot Determine the required bell submergence using the equation 21

middot Determine the minimum allowable liquid depth in the intake structure from

the sum of the floor clearance and the required bell submergence

middot Check bottom elevation near the entrance to the structure and determine if it

is necessary to slope the floor upstream of the bay entrance If the resulting

depth at the entrance to the intake structure is shallow then check to ensure

that gravity-driven flow is not restricted by the entrance condition

middot Check the pump bay velocity for the maximum single-pump flow and

minimum liquid depth with the bay width set to 2D If bay velocity exceeds

05 ms then increase the bay width to reduce the maximum velocity below

05 ms

middot If it is necessary to increase the pump bay width to greater than 2D then

decrease bay width in the vicinity of the pumps according to Fig 23

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

10

middot Compare cross-flow velocity (at maximum system flow) to average pump

bay velocity If cross-flow value exceeds 50 of the bay velocity a

physical hydraulic model study is necessary

middot Determine the length of the structure and dividing walls giving

consideration to minimum allowable distances to a sloping floor screening

equipment and length of dividing walls if dual flow traveling screens or

drum screens are to be used a physical hydraulic model study is required

middot If the final selected pump bell diameter and inlet velocity is within the

range given in section 211 then the sump dimensions (developed based on

the inlet bell design diameter) need not be changed and will comply with

these standards

22 Model test

The approach flow pattern in a pump intake facility is difficult to predict by

traditional mathematics or empirical formulae For large structures or those that

differ significantly from proven designs a model study is the only means to ensure

success A model study is used to identify adverse hydraulic conditions and

derive remedial measures for approach flow patterns generated by structures

upstream of the pump impeller Current model studies are not intended to

investigate flow patterns induced by the pump itself or the flow patterns within the

pump The objective of a model study is to ensure that the pump intake structure

generates favorable flow conditions at the inlet to the pump Evaluation for the

requirement of model test if

a Sump or piping geometry that deviates from this design standard

b Non-uniform or non-symmetric approach flow to the pump sump

exist

c The pumps have flows greater than 2520 ls per pump or the total

station flow with all pumps running would be greater than 6310 ls

d Situations that need ten times cost than the model study

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

11

Adverse hydraulic conditions that can affect pump performance include free and

submerged vortices swirl approaching the pump impeller flow separation at the

pump bell and a non-uniform axial velocity distribution at the suction

The purpose of model test is to examine these adverse hydraulic conditions and to

investigate measure against them

221 The scale effects

One possible difficulty in scale modeling is that it is impossible to reduce all

pertinent forces by the same factor With Froude-scale models the inertial and

gravitational forces are reduced similarly but the viscous and surface tension forces

cannot be simultaneously reduced The extra influence of these forces is known as

a ldquoscale effectrdquo A number of investigations have indicated that the scale effects for

the model are avoided if the Reynolds number based on the inlet flow and

submergence or intake diameter exceeds approximately 3x 104 and the Weber

number exceeds 150 [19]

In most engineering applications involving closed-conduit and open-channel flow

the Reynolds number limits are far exceeded and the flows are fully turbulent

In a sump model Froude number (Fr) and Reynolds number (Re) are the most

important non-dimensional parameters Some reduction in Re should be

introduced to the sump model However flow patterns are generally very similar

at high Re

No specific geometric scale rate is recommended but reasonable large

geometric scale to minimize viscous and surface tension scale effects and reproduce

the flow pattern

Re= uDν (22)

We= u2ρDσ (23)

where

u average axial velocity (ms)

ν kinematic viscosity of the liquid (m2s)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

12

ρ liquid density (kgm3)

σ surface tension of liquidair interface

According to the ANSIHI 98 the viscous scale effects on vortex may be negligible

if the resulting dimensionless numbers must meet these values Re>6times104 and

We>240

To allow practical visual observations of flow patterns accurate measurements of

swirl and velocity distribution and sufficient dimensional control the model scale

shall yield a bay width of at least 300mm a minimum liquid depth of at least

150mm and a pump throat or suction diameter of at least 80mm

222 Principle of similarity

In the similitude analysis the geometric and flow similarity requires the model and

the flow to be identical as the real model and flow This is to ensure the result

obtained from the model study is well presented in order to predict full scale

behavior Model involving a free surface are operated using Froude similarity since

the flow process is controlled by gravity and inertial forces

For similarity of the flow patterns the Froude number shall be equal in model and

prototype

Fr =

=

(24)

and as the scale ratio λ was selected in section 221 the following equation is

obtained

V = V times (

) = V times λ or Q = Q times (

) = Q times λ (25)

where Fr is the Froude number λ is the scale ratio of model dimensions to prototype

The test of 15 times Froude scaled flow should be conducted to ascertain the

potential scale affect on free surface vortices

Models of closed conduit piping system leading to a pump suction are not operated

based on Froude similitude However flow patterns are generally very similar at

high Re A minimum value of 1times105 for the Re at the pump suction is

recommended to ensure the flow patterns are correctly scaled

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

13

223 Preliminary model test

The initial design shall be tested first to identify any hydraulic problems

1) Vortex Formation Mechanisms

There are various investigations that provide insight into the fundamental processes

leading to the development of vortices both in experiment and numerical simulation

Shin et al [20] demonstrated that two basic mechanisms lead to inlet vortex

formation The first mechanism involves the development of an inlet vortex due to

the amplification of ambient vorticity in the approach flow as vortex lines

are convicted into the inlet The second mechanism involves the development of a

trailing vortex in the vicinity of the intake as a result of the variation in

circulation along the inlet For this second case a vortex can develop in a flow that

is irrotational upstream and the vortex development therefore does not depend

on the presence of ambient vorticity Shin et al investigation on kinematic

parameters indicate that the strength of an inlet-vortex or trailing vortex system

increases with decreasing distance from the surface However for an inlet in

an upstream irrotational flow two counter rotating vortices can still trail from

the rear of the inlet

2) Classification of vortex type

As illustrated in Fig 24 vortices in the vicinity of pump intakes may be adjacent to

the channel bottom or a channel wall (ie submerged vortices) or they may appear

adjacent to the free surface (ie free surface vortex)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

14

(a) Schematic diagram of vortices in a sump

(b) Experimental results of different vortex

Fig 24 Typical vortex in a sump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

15

For the model visual observation dye was injected around the intake structure near

the surface and submergence as Fig 24 (b) shows And the model test results of

vortex was classified as Fig 25

(a) Free Surface Vortices

(b) Submerged Vortices

Fig 25 Vortex classifications

There are a number of researches focused on the numerical theory to detect the

inception of a visible vortex As the diameters of vortices are usually much smaller

compared with the computing grid size to compensate for the lack of vortex

resolution in numerical simulation a stretching vortex model is applied to the local

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

16

flow field around semi-analytically identified vortex positions [21][22] The brief

explanations are as follows

a) Free surface vortex

The criteria expressed by equation 26 and 27 is applied to determine the

visible inception of a free surface vortex

≫ 1 (26)

=

gt 1 (27)

where P is the pressure drop at the vortex core and Ph is the static head at the

vortex element and f is the drag force acting at a unit length of water surface

vortex dimple based on Stokes approximation and caused by the down flow

Further f is the force acting on the bubble in the direction from the high

pressure region to the low pressure region like buoyancy

b) Submerged vortex

Equation 28 is applied to determine the visible inception of a submerged vortex

infin =

infin

infin gt 1 (28)

where is the static pressure at the vortex core infin is the atmospheric pressure

and is the critical cavitation inception pressure and nearly equal to the

saturated vapor pressure

3) Swirl angle

Swirl angle predicts the intensity of flow rotation Fig 26 shows experiment

calculation method with swirl meter and attempts of the CFD calculation method

were conducted as Fig 27 A four blade zero-pitch swirl meter which is supported

by a low friction bearing will be installed at about four suction pipe diameter (d)

downstream from the pump suction to measure swirl angle of pump approaching

flow For the swirl meter the tip to tip blade diameter is 075d and the length in

flow direction is 06 and one of the four blade is painted yellow as a reference to

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

17

count revolutions As flow swirl is usually unsteady the observation time of swirl

meter reading should be a continuous period of time

Fig 26 Typical swirl meter

(29)

(210)

(211)

Where qV flow rotational speed at swirl meter (ms)

zV average axial velocity at swirl meter (ms)

n Number of revolution per minute (rpm)

d Diameter of pipe at the swirl meter (m)

q flow rate (m3s)

zV

Vqq 1tan -=

)(60

smnd

Vtimestimes

=p

q

)(4

2

smd

qVz

times=

p

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

18

For the swirl angle calculating method with CFD the key point is to obtain the

average tangential velocity so the swirl check circles were created as illustrated in

Fig 27 The swirl check circle is located at the section of 4d height with diameter

of 025d 05d and 075d respectively

Swirl check circle

Fig 27 Schematic view of swirl angle numerical calculation

23

1 7501 501 250

acute++

=aringaringaring

N

i

N

i

NNVNVNV

Vi

q (212)

where

qV average tangential velocity (ms)

V circumferential velocity of check point in check circle

respectively (ms)

N check point total numbers in single check circle

4) Velocity

Velocity measurement at bell throat by Pitot tube was implemented by ANSIHI

method Typical Pitot tube configuration is shown in Fig 28 Eight measuring

points are defined as Fig 29 Time averaged velocity at eight points was obtained

and these eight data were again averaged to give throat sectional spatial velocity

average

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

19

Fig 28 Example of Pitot tube Fig 29 Measuring points at bell throat

224 Test criteria

There is the criteria for satisfactory performance of the sump model test and

summarized in Table 23

1) Free surface and submerged vortices

Free surface vortices entering the pump must be less severe than vortices with

coherent core Dye core vortices may be acceptable only if they occur less than 10

of the time or only for infrequent pump operating conditions

Submerged vortex which starts from the sump walls induce vibration and noise

problems Dye core submerged vortices extending into the pump bell mouth is not

allowed

2) Swirl angles

Swirl angle is used to check the flow rotation in the suction pipe The calculation

method is referred to section 223

Short-term maximum and long-term average values must be less than 5 degrees

Maximum short-term swirl angles up to 7 degrees may be acceptable only if they

occur less than 10 of the time or only for infrequent pump operating conditions

3)Velocity

Each point time-averaged velocity shall not be larger than 10 of sectional average

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

20

Standard deviation of time-varying velocity at each point was obtained and its

value shall not be larger than 10 of the time-averaged velocity at each point

1) Time-averaged

Variation[] = 100acute-

a

a

V

VV aV Averaged at section

2) Time-varying

Standard deviation= )1(

)(1i

2

-

-aring=

n

xxn

i

Variation[] = (Standard deviation X ) x 100

X Averaged at point

Table 23 Permission criteria of ANSIHI

Vortex occurence Swirl angle

Free surface vortex Submerged vortex

Observation

time Above 10 minutes Above 10 minutes

Above 10

minutes

Permission

criteria

Dye-core

Vortex(F3)

below 110 total

time

Dye-core

Vortex(S2)

below 110 total

time

Below

5degConditional

intermittent

below 7deg

225 Remedial measures for problem intakes

Usually in the preliminary test the worst flow condition is simulated And based on

the results remedial measures for problem intakes are discussed in this part

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

21

1) Overall approach flow patterns

The characteristics of flow approaching an intake structure is one of the foremost

considerations for the designer It is difficult to predict the effects of a given set of

flow conditions upstream from an intake structure Fig 210 depicts several typical

approach flow patterns of pump sump with different operating conditions

Fig 210 Examples of approach flow patterns

Generally speaking for the intake problems especially for the multiple pumps

sump there are the main measures using dividing walls placed between the pumps

as a partitioned structure trash racks with elongated bars guide piers or a number

of structural columns at the bay entrance can provide some assistance in controlling

cross-flow baffles flow distributors and screen mesh etc are applied to improve

the flow distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

22

2) Problems around pump intake

a) Free surface vortices

There are a number of factors that have an effect on the formation of surface

vortices To achieve a higher degree of certainty that objectionable surface vortices

do not form modifications can be made to intake structure to allow operation at

practical depths of submergence and the minimum submergence is referred in Table

22 the use of suction umbrellas vertical curtain walls and horizontal gratings

b) Submerged vortices

In the vicinity of intake bell there are the most complicated flow patterns and flow

makes the most changes in direction while maintaining a constant acceleration into

the pump bell to prevent local flow separation turbulence and vortex formation

Fig 211 represents a sample of various devices to inhibit the different submerged

vortex

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

23

Fig 211 Methods to reduce submerged vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

24

c) Preswirl

The approach flow distributions determine whether flow preswirl exists and a

sufficiently laterally skewed approach flow cause rotating around the intake bell

The most effective way of reducing preswirl is to establish a relatively uniform

approach flow within each pump bay by using the baffling schemes controlling

cross-flow and expanding concentrated flow discussed in the above part

To prevent non-uniform velocity distribution swirls and vortices necessary

modifications are conducted such as fore bay length baffle wall distributors

curtain wall mesh screen guide vane floor splitter wall fillet etc As the above

measures applied modification test will be conducted to check all the flow

conditions and compare with the preliminary results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

25

Chapter 3 Methodology

CFD (Computational fluid dynamics) is the technique which can design cycles to

better performance and reduce costs and time In CFD we solve the governing

equations of given physics (may be differential form or integral form) using some

numerical techniques like Finite Difference Method (FDM) Finite Element Method

(FEM) or Finite Volume Method (FVM) Although CFD plays a great part in the

design of turbomachinery CFD is not a replacement for experimental or analytical

approach In this study the numerical analysis was performed by the commercial

code ANSYS CFX 130 All aspects of the CFD study including modelling

meshing solving and post-processing of the results

31 Numerical modeling

ANSYS CFX solves the unsteady Navier-Stokes equations in their conservation

form The instantaneous equation of mass (continuity) in the stationary frame is

expressed as equation

ρ

δ + nabla ∙ (ρU) = 0 (31)

And the instantaneous equation for momentum is expressed as shown in equation

ρ

δ + nabla ∙ (ρU⨂U) = minusnablap + nabla ∙ τ + S (32)

Where τ is the stress tensor ( including both normal and shear components of the

stress)

These instantaneous equations are averaged for turbulent flows leading to additional

terms that need to be solved While the Navier-Stokes equations describe both

laminar and turbulent flows without addition terms realistic flows involve length

scales much smaller than the smallest finite volume mesh A Direct Numerical

Simulation of these flows would require significantly more computing power than

what is available now or in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

26

Therefore much research has been done to predict the effects of turbulence by

using turbulence models These models account for the effects of turbulence

without the use of a very fine mesh or direct numerical simulation

These turbulence models modify the transport equations by adding averaged and

fluctuating components The transport equations are changed to equations (33) and

(34)

ρ

δ + nabla ∙ (ρU) = 0 (33)

ρ

+ nabla ∙ ρ U ⨂U = minusnablap + nabla ∙ τ minus ρu⨂u + S (34)

The mass equation is not changed but the momentum equation contains extra terms

which are the Reynolds stresses ρ u⨂u and the Reynolds flux ρ u ϕ These

Reynolds stresses need to be modeled by additional equations to obtain closure

Obtaining closure implies that there are a sufficient number of equations to solve for

all the unknowns including the Reynolds stresses and Reynolds fluxes

Various turbulence models provide various ways to obtain closure In this

investigation Shear Stress Transport (SST) model and k-ε model were utilized The

advantage of SST model is that combines the advantages of other turbulence models

(the k-ε Wilcox k-ω) The 3 turbulence models will be discussed briefly

32 Turbulence models

Turbulence models are necessary because we cannot afford big enough computers

to directly capture every scale of motion There are three modelling frameworks

Direct Numerical Simulation (DNS) Reynolds-Averaged Navier-Stokes

Equations(RANS) Large Eddy Simulation(LES)

As RANS method is applied in most productionseveral RANS turbulence models

differ in their usage of wall functions the number of additional variables solved for

and what these variables represent

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

27

Two equation turbulence models

Two-equation turbulence models are widely used in CFD as they give a good

compromise between computational power needed and accuracy The term lsquotwo

equationrsquo refers to the fact that these models solve for the velocity and length scales

using separate transport equations The turbulence velocity scale is obtained by

solving the transport equation The turbulent length scale is estimated from two

properties of the turbulence field namely the turbulent kinetic energy and the

dissipation rate The dissipation rate of the kinetic energy is obtained from its

transport equation The most widely used are k-ε and k-ω two equation models In

the next section the k-ε Wilcox k-ω BSL k-ω and SST models will be briefly

discussed

321 k-ε turbulence model

The k-epsilon model solves for two variables k the turbulent kinetic energy and

epsilon the rate of dissipation of kinetic energy Wall functions are used in this

model so the flow in the buffer region is not simulated The k-epsilon model is very

popular for industrial applications due to its good convergence rate and relatively

low memory requirements It does not very accurately compute flow fields that

exhibit adverse pressure gradients strong curvature to the flow or jet flow It does

perform well for external flow problems around complex geometries

The k-ε model introduces k (m2s2) as the turbulence kinetic energy and ε (m2s3) as

the turbulence eddy dissipation The continuity equation remains the same

ρ

+ nabla ∙ (ρ U) = 0 (35)

The momentum equation changes as shown by equation

δρ

δ + nabla ∙ (ρU⨂U) = minusnablapprime + nabla ∙ μ

nablaU + (nablaU) + S (36)

Where SM is the sum of body forces μeff is the effective viscosity accounting for

turbulence and prsquo is the modified pressure The k-ε model uses the concept of eddy

viscosity giving the equation for effective viscosity as shown by equation

μ

= μ + μ (37)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

28

μt is the turbulence viscosity is linked to the turbulence kinetic energy and

dissipation by the equation

μ = Cμρ

ε (38)

Where Cμ is a constant

The values for k and ε come from the differential transport equations for the

turbulence kinetic energy and the turbulence dissipation rate

The turbulence kinetic energy equation is given as equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus ρε (39)

The turbulence dissipation rate is given by equation

(ρε)

+ nabla ∙ (ρUε) = nabla ∙ μ +

μ

σε nablaε +

ε

(Cε (P + Pε ) minus Cε ρε) (310)

Where Cε1 Cε2 σk σε are constants

Pk is the turbulence production due to viscous forces and is modeled by the

equation

P = μ nablaU ∙ nablaU + nablaU minus

nabla ∙ U 3μ

nabla ∙ U + ρk (311)

If a buoyancy term is added to the previous equation if the full buoyancy model is

used

The buoyancy term Pkb is modeled as

P = minusμ

ρσρg ∙ nablaρ (312)

322 Wilcox k-ω turbulence model

The k-omega model is similar to k-epsilon instead however it solves for omega mdash

the specific rate of dissipation of kinetic energy It also uses wall functions and

therefore has comparable memory requirements It has more difficulty converging

and is quite sensitive to the initial guess at the solution Hence the k-epsilon model

is often used first to find an initial condition for solving the k-omega model The k-

omega model is useful in many cases where the k-epsilon model is not accurate

such as internal flows flows that exhibit strong curvature separated flows and jets

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

29

This model has an advantage over the k-ε model where it does not involve complex

linear damping functions for near wall calculations at low Reynolds numbers

The k-ω model assumes that the turbulence viscosity is related to the turbulence

kinetic energy k and the turbulent frequency ω by the equation

μ = ρ

ω (313)

The transport equation for k is given by the equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus βprimeρ k ω (314)

The transport equation for ω is shown as equation

(ρω)

δ + nabla ∙ (ρ U ω) = nabla ∙ μ +

μ

μω

nablaω + αω

P + Pω minus β ρ ω (315)

The production rate of turbulence (Pk) is calculated as shown previously in the k-ε

section

The recommended values for the constants in the above equations are

βprime = 009

α = 59

β = 0075

σ = 2

σω = 2

The Reynolds stress tensor ρ u⨂u is calculated by

minusρ u⨂u = μ nablaU + (nablaU) minus

part ρk + μ

nabla ∙ U (316)

323 Shear stress transport model

The disadvantage of the Wilcox model is the strong sensitivity to free-stream

conditions Therefore a blending of the k-ω model near the surface and the k-ε in

the outer region was made by Menter [23] which resulted in the formulation of the

BSL k-ω turbulence model It consists of a transformation of the k-ε model to a k-ω

formulation and subsequently adding the resulting equations The Wilcox model is

multiplied by a blending function F1 and the transformed k-ε by another function 1-

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

30

F1 F1 is a function of wall distance (being the value of one near the surface and zero

outside the boundary layer) Outside the boundary and on the edge of the boundary

layer the standard k-ε model is used [24]

SST model is a combination of the k-epsilon in the free stream and the k-omega

models near the walls It does not use wall functions and tends to be most accurate

when solving the flow near the wall The SST model does not always converge to

the solution quickly so the k-epsilon or k-omega models are often solved first to

give good initial conditions

The k-ω based Shear Stress Transport (SST) model of Menter was applied for

turbulence treatment The transport equations for the SST model are expressed

below where the turbulent kinetic energy lsquokrsquo and turbulent frequency or dissipation

per unit turbulent kinetic energy are computed by using the following relations

For Turbulence Kinetic Energy

kkPukt

kTkk Ntilde+Ntilde+-=Ntilde+

para

para)[()( nsnwb (317)

where Pk is the production limiter

For Specific Dissipation Rate

ww

swnsnbwaww

ww NtildeNtilde-+Ntilde+Ntilde+-=Ntilde+para

para

1)1(2])[()( 21

22 kFSut

T (318)

The first blending function F1 is calculated from

iumlthorn

iumlyacute

uuml

iumlicirc

iumliacute

igrave

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

uacuteuacuteucirc

ugrave

ecircecirceuml

eacute

dividedivideoslash

oumlccedilccedilegrave

aelig=

4

22

21

4500maxmintanh

yCD

k

yy

kF

kw

wrs

w

n

wb (319)

y is the distance to the nearest wall and v is the kinematic viscosity

divideoslash

oumlccedilegrave

aeligNtildeNtilde= -10

2 101

2max ww

rsww kCDk (320)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

31

And Kinematic eddy viscosity

)max( 21

1

SF

kT

wa

an = (321)

where F2 is a blending function which restricts the limiter to the wall boundary and

S is the invariant measure of the strain rate

uacuteuacute

ucirc

ugrave

ecircecirc

euml

eacute

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

dividedivideoslash

oumlccedilccedilegrave

aelig=

2

22

5002maxtanh

w

n

wb yy

kF (322)

33 Brief introduction of simulation process

All aspects of CFD study including modelling meshing solving and post-

processing of the results are briefly introduced in this chapter

331 Pre-processing technique

The speed and accuracy of simulation results depending fundamentally on meshing

and the ability to prepare model geometry correctly

The steps in pre-processing are explained as follows

1) Domain Definitions

The domain definition depends mainly on the physics involved in the problem and

objective of the problem

a) Geometry creation or geometry import

b) Geometry cleanup

i) Removing the parts not necessary for simulation under consideration

ii) Closing the gaps in the geometry

iii) Removing small surfaces curves to make meshing process simple

2)Grid generation

To solve the governing equations the numerical methods like Finite Difference

Finite Volume or Finite Element Method to divide the domain into small parts and

solve the equations on one part at a time are applied

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 3: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

본 論文 조우흠 工學碩士

學位論文 로 認准함

원 공학박사 재현 ( )

원 공학박사 형호 ( )

원 공학박사 영호 ( )

2014 년 6 월 19

한 해양대학 대학원

기계공학과

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

i

TABLE OF CONTENTS

List of Tablesiii

List of Figuresiv

Abstractvi

Nomenclatureviii

Abbreviationsix

Chapter 1 Introduction1

11 Background1

12 Previous study1

13 Methodology of study3

Chapter 2 Basic theory of sump model4

21 Pump intake design5

211 Inlet bell diameter design 5

212 Recommended dimensions for a rectangular sump7

22 Model test10

221 The scale effects11

222 Principle of similarity12

223 Preliminary model test13

224 Test criteria19

225 Remedial measures for problem intakes20

Chapter 3 Methodology25

31 Numerical modeling25

32 Turbulence models26

321 k-ε turbulence model27

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

ii

322 Wilcox k-ω turbulence model28

323 Shear stress transport model29

33 Brief introduction of simulation process31

331 Pre-processing technique31

332 Solving the simulation32

333 Post processing32

Chapter 4 Description of Model Cases33

41 Creating the geometry33

411 Geometry of the scaled sump model33

412 Geometry of the full size pump sump model35

42 Mesh generation39

43 Numerical approach42

44 Experimental setup45

Chapter 5 Results and Discussion47

51 Results of the scaled sump model47

511 Numerical simulation vortex check47

512 PIV results analysis49

513 Sump with anti free surface vortex devices53

52 Results of the mixed flow pump sump model56

521 Performance analysis of the mixed flow pump sump model56

522 Flow characteristics analysis58

523 Cavitation phenomenon analysis62

Chapter 6 Conclusions60

Acknowledgement67

References68

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

iii

List of Tables

Table 21 Acceptable velocity ranges for inlet bell diameter 6

Table 22 Recommended dimensions for a rectangular sump8

Table 23 Permission criteria of ANSIHI20

Table 41 Designed specifications of scaled sump model33

Table 42 Parameters of anti free surface vortex devices35

Table 43 Design specifications of the mixed flow pump38

Table 44 Mesh information for the pump sump model42

Table 51 Swirl angle of model test results54

Table 52 Summary of swirl angle CFD calculation results54

Table 53 Mixed flow pump performance results56

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

iv

List of Figures

Fig 21 Recommended inlet bell design diameter6

Fig 22 Recommended intake structure layout7

Fig 23 Filler wall details for proper bay width7

Fig 24 Typical vortex in a sump14

Fig 25 Vortex classifications15

Fig 26 Typical swirl meter17

Fig 27 Schematic view of swirl angle numerical calculation 18

Fig 28 Example of Pitot tube 19

Fig 29 Measuring point at bell throat19

Fig 210 Examples of approach flow patterns21

Fig 211 Methods to reduce submerged vortices23

Fig 41 Geometry of scaled sump model33

Fig 42 Schematic diagram of sump with anti free surface vortex devices34

Fig 43 Overall 3D drawing of the sump model35

Fig 44 2D drawing of the sump sketch36

Fig 45 Shape of the mixed flow pump37

Fig 46 Sump model with the submerged AVD38

Fig 47 Submerged AVD dimensions39

Fig 48 Overall mesh generation for scaled sump models40

Fig 49 Overall sump fluid domain mesh for the pump sump model41

Fig 410 Grid details of the pump sump model41

Fig 411 Boundaries of the scaled sump43

Fig 412 Experimental setup for the scaled sump model46

Fig 51 Streamline pattern of free surface47

Fig 52 Vortex core region of the sump model48

Fig 53 Velocity vectors on different sections48

Fig 54 Velocity vectors of PIV results51

Fig 55 Velocity vectors on different sections (PIV)52

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

v

Fig 56 Velocity vectors on free surface of different cases53

Fig 57 Tangential velocity profile of different models55

Fig 58 Performance curve of the mixed flow pump with the sump 57

Fig 59 Streamline pattern around the intake structure58

Fig 510 Vortex core region of the pump sump model59

Fig 511 Vorticity distribution in the flow direction60

Fig 512 Vorticity distribution in channel width direction61

Fig 513 Cavitation region of the mixed flow pump62

Fig 514 Cavitation performance curves of the mixed flow pump63

Fig 515 Vapor volume fraction distribution on blade surface64

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

vi

Study of Flow Field in Rectangular Sump Models and

Performance Analysis of a Mixed Flow Pump

Yuxin ZHAO

Department of Mechanical Engineering

Graduate School of Korea Maritime and Ocean University

Abstract

It is an accepted fact that quite a number of problems faced in a pumping station are

related to the design of sump or intake rather than pump design Head-capacity

curves provided by the pump manufacturer are obtained on the condition of no

vortices flowing into the pump intake The efficiency and performance of pumping

stations depend not only on the performance of the selected pumps but also on the

proper design of the intake sumps A faulty design of pump sump can lead to the

occurrence of swirl and vortices which reduce the pump performance and induce

vibration and additional noise Therefore sump model test is necessary to check the

flow condition around intake structure Numerical simulation is a good facility for

reducing the time and cost involved throughout the design process In this study the

commercial software ANSYS CFX-130 has been used for the CFD analysis of the

sump models

In a scaled sump model for air entrainment simulation numerical analysis of single

phase and two-phase with SST turbulence model was carried out to predict vortex

(both free surface vortex and submerged vortex) occurrence and location The

effectiveness of curtain walls and square bars to eliminate the free surface vortex is

evaluated Meanwhile the experiment including PIV system was performed in

KMOU to investigate the flow conditions around pump intake structure

In a full size sump model overall numerical analysis for the sump model with a

mixed flow pump installed was carried out Hydraulic performances of the mixed

flow pump for head rise shaft power pump efficiencies versus flow rate changed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

vii

from 50 to 140 of the design flow rate were studied by the performance curves

A trident shaped anti vortex device (AVD) composed of three wall fillets and one

center splitter was installed under the pump intake The effectiveness of the AVD

for the suppression of submerged vortex was evaluated In addition numerical

simulation of cavitation phenomenon in the mixed flow pump has been performed

by calculating the full cavitation model with k-ε turbulence model

According to the results details of the location size and strength of vortices were

predicted in the numerical simulation For the scaled sump model although there is

a little difference between the single phase and two-phase simulation all the results

predict the free surface vortex and submerged vortex formation and location The

CFD simulations of flow condition show good agreement with the PIV results The

effectiveness of curtain walls installed to inhibit the free surface vortex is confirmed

in the CFD results while the square bars installed in the channel failed to suppress

the free surface vortex For the mixed flow pump with sump the best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

The effectiveness of AVDs installed for the submerged vortex is confirmed by

comparisons of the sump model with and without AVD results

Cavitation phenomenon analysis shows the pump operating at the BEP condition

has a considerable extent over the requirement of cavitation performance With

numerical simulation the inception of cavitation in the blade passage is observed on

the suction surface where the leading edges meet the tips and then as the inlet total

pressure decrease the cavitation zone is spread out over the suction surface as well

as the leading edge of the impeller blade

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

viii

Nomenclature

B Distance from the back wall to the pump inlet bell centerline [m]

C Distance between the inlet bell and floor [m]

D Inlet bell design outside diameter [m]

Fr Froude number (=Vb(gD)05) [-]

g Acceleration due to gravity [ms2]

ht Total head [m]

n Number of revolution per min [-]

Pflow Flow power [W]

Pshaft Shaft power [W]

Pν Vapor pressure [Pa]

Q Volume flow rate [m3s]

Q0 Design volume flow rate [m3s]

Re Reynolds number (=VbDν) [-]

S Minimum submergence depth [m]

Vb Velocity at bell mouth [ms]

α Angle of floor slope [deg]

η Pump efficiency []

θ Swirl angle [deg]

λ The scale ratio of model dimensions to prototype [-]

μ Viscosity [Pa s]

ν Kinematic viscosity [m2s]

ρ Specific water density [kgm3]

σ Surface tension [Nm]

τ Stress tensor [-]

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

ix

Abbreviations

AVD Anti vortex device

BEP Best efficiency point

CFD Computational fluid Dynamics

NPSH Net positive suction head

NPSHR Required net positive suction head

PIV Particle Image Velocimetry

RANS Reynolds Averaged Navier- Stokes

RPM Number of revolutions per minute

SST Shear Stress Transport

VOF Volume of Fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

1

Chapter 1 Introduction

11 Background

It is an accepted fact that quite a number of problems faced in a pumping station are

related to the design of sump or intake rather than pump design As an important

part of the pumping station sump is designed to provide uniform swirl and

entrained air free flow to the pump intake Undesirable flow conditions (entrained

air non-uniform flow distribution vortices etc) can reduce the pump efficiency

induce vibration noise and cavitation even lead to excessive bearing loads of

impeller and structure damage Due to quite complex nature of the flow although

the flow conditions causing pump problems are well established there are no

specific solutions to eliminate them There are some design guidelines for the

specific geometrical and hydraulic constraints of the pump sump for any project

the model study is the only tool for solving potential problems in new designs and

rectifying problems observed in existing installations However the scaled model

tests are expensive and time-consuming so alternative Computational Fluid

Dynamics (CFD) methods for evaluating sump performance have been developed

With rapid progress in CFD numerical simulation is regarded as an effective tool in

solving fluid problems in pump sumps

12 Previous study

Various researches show the flow conditions around the sump intake structure are

quite complicated Shyam et al [1] used the commercial code CFX to carry out

two-phase flow simulation to capture air entrainment The air entrainments and its

location were well captured and numerical results were in accordance with the

experimental investigation Matsui J[2] and Okamura T[3] et al conducted a serial

research of the benchmark according to the Turbomachinery Society of Japan (TSJ)

standard including various commercial codes comparison performed in the flow

simulation and comparison analysis between the numerical and experimental results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

2

The purpose of the benchmark is to determine the accuracy and reliability of the

CFD codes used in universities and the industrial field It is also intended to identify

and investigate the formation of air-entrained and submerged vortices near the

pump Tanweer et al[4] have performed a numerical analysis of a multiple intake

pump sump to confirm that the geometry of sump affects the flow within sump

greatly Rajendran et al[5] have made a numerical analysis for the flow

characteristics of a sump model with pump intake and good agreements were

achieved by comparing the numerical results with the experiments Iwano et al[6]

have introduced a numerical method for the submerged vortex by analyzing the

flow in the pump sump with and without baffle plates Lee et al[7] have conducted

the CFD analysis of a multi-intake pump sump model to check the flow uniformity

by predicting the location number and vorticity of the vortex For the CFD

application in mixed flow pump design and performance analysis Oh H W [8]

conducted a mixed flow pump design optimization with specific parameters and

then compared the numerical results with the experiments The procedure presented

can be used as a practical design and analysis guide for the general turbomachinery

Kim et al[9] carried out a mixed flow pump performance analysis with numerical

simulation and improved the hydraulic efficiency by the modification of the

geometry

As an undesirable phenomenon caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage The cavitation

phenomenon has been studied by various investigators Van et al[10] have

investigated the cavitation inception behavior of a mixed-flow pump impellers

model and the calculated results matched well with the experiments Okamura et

al[11] have evaluated two numerical cavitation prediction methods used in the

pump industry manufacturers and attempts were made to improve the cavitation

performance Li et al[12] have conducted the numerical analysis on the basis of the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

3

development liquidvapor interface tracking method to predict the cavitation

characteristics with a centrifugal pump impeller model

Moreover there are a number of correlated studies [13]~[17]

13 Methodology of study

This study will investigate the flow phenomena of two sump models

A scaled sump model is adopted to carry out numerical simulation and PIV

experiment The flow conditions are predicted and the effectiveness of devices for

free surface vortex is evaluated by using numerical code a single intake sump

model with a mixed flow pump installed is modeled and the hydraulic performance

characteristics for the head rise shaft power and pump efficiencies versus flow rate

are studied In addition flow phenomena (vortices cavitation etc) are analyzed

and the effectiveness of an anti-vortex device (AVD) for the submerged vortex is

evaluated

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

4

Chapter 2 Basic theory of sump model

Specific hydraulic phenomena have been confirmed that they can lead to certain

common operational problems The common of the main problems are summarized

as follows

a Free surface vortices

b Submerged vortices

c Swirl of flow entering the pump

d Non-uniform distribution of velocity at the pump impeller

e Entrained air or gas bubbles

These problems encountered in the pump sump will affect the pump

performance and significantly increase the operational and maintenance costs In

order to identify sources of particular problems and find practical solution for

it the usual approach is to conduct the laboratory experiments on a scaled

physical model Typical design objectives are to ensure that a pump station is

designed according to best practice and conforms to the requirements set out in the

American National Standard for Pump Intake Design[18] These standards restrict

the degree to which the above-mentioned undesirable flow patterns may be present

The negative impact of each of these hydraulic conditions varies with pump specific

speed and size Application of the standard to design a sump does not generate a

problem free sump but provides a basis for the initial design

As there are no specific guidelines or criteria for design of trouble free intakes the

most common solution to potential problems in new design and rectification of

problems observed in existing designs is to construct a scaled model in a laboratory

observe and investigate the flow characteristics and propose modifications to the

intake geometry There are additional devices in the form of floor splitters or cones

back wall splitters fillets surface beams guide vanes etc aimed at controlling the

vortex and swirl formation Considerable prior experience and ingenuity are

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

5

required to solve the problems by this method and the cost and time required are

also significant

The intake structure refers to the channel leading from the water source which

may be a river or reservoir and all installations downstream including the pump

column or the intake pipe (the suction tube portion of a vertical intake) the

approach channel (upstream of pump bay) and the pump bay (bounded by the floor

the back wall side walls dividers walls separating adjacent pump column) Usually

the upstream end of the pump column has an inlet attachment called the suction bell

The function of the intake is to supply an evenly distributed flow of water to the

pump suction bell

Several types of pump intake basins are available and ANSIHI 98 provides

design requirements for each Intake structures can be categorized as being

clear liquids or solids-bearing liquids For clear liquids intakes are further

classified into rectangular formed circular and trench types as well as suction

tanks and cans For solid-bearing liquids trench type and rectangular wet wells are

usually considered

In this paper only rectangular intake structure for clear liquid is studied

21 Pump intake design

For pumps to achieve their optimum hydraulic performance across all operating

conditions the flow at the impeller must meet specific hydraulic conditions The

ideal flow entering the pump inlet should be of a uniform velocity distribution

without rotation and stable over time When designing a sump to achieve favorable

inflow to the pump or suction pipe bell various sump dimensions relative to the size

of the bell are required Using these standards of the distances to reduce the

probability of the occurence of strong submerged vortices and free-surface vortices

211 Inlet bell diameter design

The geometry is generally defined in terms of the pump inlet diameter as shown in

Table 21 and Fig 21 The bell diameter can be estimated based on an inlet pipe

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

6

Table 21 Acceptable velocity ranges for inlet bell diameter(ANSIHI 98)

Flow Rate

Qls

Recommended Inlet Bell Design

Velocityms

Acceptable Velocity

Range ms

Q<315 V=17 06leVle27

315 leQ<1260 V=17 09leVle24

Q ge1260 V=17 12leVle21

Fig 21 Recommended inlet bell design diameter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

7

212 Recommended dimensions for a rectangular sump

As the selected bell diameter has been determined the proportions of the inlet

structure can be estimated from Fig 22 and Fig 23 Table 22 shows the

recommended dimensions for a rectangular sump

Fig 22 Recommended intake structure layout

Fig 23 Filler wall details for proper bay width

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

8

Table 22 Recommended dimensions for a rectangular sump

Dimension Variable

Description Recommended value

A Distance from the pump inlet bell

centerline to the entrance A=5D minimum

a Length of constricted bay section near

the pump inlet 25D minimum

B Distance from the back wall to the

pump inlet bell centerline 075D

C Distance between the inlet bell and

floor 03D~05D

D Inlet bell outside diameter refer to section 211

H Minimum liquid depth H=S+C

h Minimum height of constricted bay

section near the pump inlet bell hgtH or 25D

S Minimum pump inlet bell submergence S=D(10+23Fr)

W Pump inlet bay entrance width 2D minimum

w Constricted bay width near the pump

inlet bell w=2D

X Pump inlet bay length 5D minimum

Y Distance from pump inlet bell

centerline to the through-flow traveling screen

4D minimum

Z1 Distance from sump inlet bell centerline to diverging walls

5D minimum

Z2 Distance from inlet bell centerline to

sloping floor 5D minimum

α Angle of floor slope -10deg~ +10deg

β Angle of wall convergence 0deg~ +10deg

empty Angle of convergence from constricted

area to by walls 10deg maximum

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

9

FD=Vb(gD)05 (21)

where

FD Froude number (dimensionless)

Vb velocity at suction inlet=FlowArea based on D

D Outside diameter of bell or pipe inlet

g gravitational acceleration

There is the design sequence summary for rectangular intake structures

middot Consider the flow patterns and boundary geometry of the body of liquid

from which the pump station is to receive flow Compare with the approach

flow condition and determine if a hydraulic physical model study is

required

middot Determine the number and size of pumps required to satisfy the range of

operating conditions likely to be encountered

middot Identify pump inlet bell diameter Refer to the section 211

middot Determine the bell-floor clearance 05D is a good preliminary design value

middot Determine the required bell submergence using the equation 21

middot Determine the minimum allowable liquid depth in the intake structure from

the sum of the floor clearance and the required bell submergence

middot Check bottom elevation near the entrance to the structure and determine if it

is necessary to slope the floor upstream of the bay entrance If the resulting

depth at the entrance to the intake structure is shallow then check to ensure

that gravity-driven flow is not restricted by the entrance condition

middot Check the pump bay velocity for the maximum single-pump flow and

minimum liquid depth with the bay width set to 2D If bay velocity exceeds

05 ms then increase the bay width to reduce the maximum velocity below

05 ms

middot If it is necessary to increase the pump bay width to greater than 2D then

decrease bay width in the vicinity of the pumps according to Fig 23

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

10

middot Compare cross-flow velocity (at maximum system flow) to average pump

bay velocity If cross-flow value exceeds 50 of the bay velocity a

physical hydraulic model study is necessary

middot Determine the length of the structure and dividing walls giving

consideration to minimum allowable distances to a sloping floor screening

equipment and length of dividing walls if dual flow traveling screens or

drum screens are to be used a physical hydraulic model study is required

middot If the final selected pump bell diameter and inlet velocity is within the

range given in section 211 then the sump dimensions (developed based on

the inlet bell design diameter) need not be changed and will comply with

these standards

22 Model test

The approach flow pattern in a pump intake facility is difficult to predict by

traditional mathematics or empirical formulae For large structures or those that

differ significantly from proven designs a model study is the only means to ensure

success A model study is used to identify adverse hydraulic conditions and

derive remedial measures for approach flow patterns generated by structures

upstream of the pump impeller Current model studies are not intended to

investigate flow patterns induced by the pump itself or the flow patterns within the

pump The objective of a model study is to ensure that the pump intake structure

generates favorable flow conditions at the inlet to the pump Evaluation for the

requirement of model test if

a Sump or piping geometry that deviates from this design standard

b Non-uniform or non-symmetric approach flow to the pump sump

exist

c The pumps have flows greater than 2520 ls per pump or the total

station flow with all pumps running would be greater than 6310 ls

d Situations that need ten times cost than the model study

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

11

Adverse hydraulic conditions that can affect pump performance include free and

submerged vortices swirl approaching the pump impeller flow separation at the

pump bell and a non-uniform axial velocity distribution at the suction

The purpose of model test is to examine these adverse hydraulic conditions and to

investigate measure against them

221 The scale effects

One possible difficulty in scale modeling is that it is impossible to reduce all

pertinent forces by the same factor With Froude-scale models the inertial and

gravitational forces are reduced similarly but the viscous and surface tension forces

cannot be simultaneously reduced The extra influence of these forces is known as

a ldquoscale effectrdquo A number of investigations have indicated that the scale effects for

the model are avoided if the Reynolds number based on the inlet flow and

submergence or intake diameter exceeds approximately 3x 104 and the Weber

number exceeds 150 [19]

In most engineering applications involving closed-conduit and open-channel flow

the Reynolds number limits are far exceeded and the flows are fully turbulent

In a sump model Froude number (Fr) and Reynolds number (Re) are the most

important non-dimensional parameters Some reduction in Re should be

introduced to the sump model However flow patterns are generally very similar

at high Re

No specific geometric scale rate is recommended but reasonable large

geometric scale to minimize viscous and surface tension scale effects and reproduce

the flow pattern

Re= uDν (22)

We= u2ρDσ (23)

where

u average axial velocity (ms)

ν kinematic viscosity of the liquid (m2s)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

12

ρ liquid density (kgm3)

σ surface tension of liquidair interface

According to the ANSIHI 98 the viscous scale effects on vortex may be negligible

if the resulting dimensionless numbers must meet these values Re>6times104 and

We>240

To allow practical visual observations of flow patterns accurate measurements of

swirl and velocity distribution and sufficient dimensional control the model scale

shall yield a bay width of at least 300mm a minimum liquid depth of at least

150mm and a pump throat or suction diameter of at least 80mm

222 Principle of similarity

In the similitude analysis the geometric and flow similarity requires the model and

the flow to be identical as the real model and flow This is to ensure the result

obtained from the model study is well presented in order to predict full scale

behavior Model involving a free surface are operated using Froude similarity since

the flow process is controlled by gravity and inertial forces

For similarity of the flow patterns the Froude number shall be equal in model and

prototype

Fr =

=

(24)

and as the scale ratio λ was selected in section 221 the following equation is

obtained

V = V times (

) = V times λ or Q = Q times (

) = Q times λ (25)

where Fr is the Froude number λ is the scale ratio of model dimensions to prototype

The test of 15 times Froude scaled flow should be conducted to ascertain the

potential scale affect on free surface vortices

Models of closed conduit piping system leading to a pump suction are not operated

based on Froude similitude However flow patterns are generally very similar at

high Re A minimum value of 1times105 for the Re at the pump suction is

recommended to ensure the flow patterns are correctly scaled

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

13

223 Preliminary model test

The initial design shall be tested first to identify any hydraulic problems

1) Vortex Formation Mechanisms

There are various investigations that provide insight into the fundamental processes

leading to the development of vortices both in experiment and numerical simulation

Shin et al [20] demonstrated that two basic mechanisms lead to inlet vortex

formation The first mechanism involves the development of an inlet vortex due to

the amplification of ambient vorticity in the approach flow as vortex lines

are convicted into the inlet The second mechanism involves the development of a

trailing vortex in the vicinity of the intake as a result of the variation in

circulation along the inlet For this second case a vortex can develop in a flow that

is irrotational upstream and the vortex development therefore does not depend

on the presence of ambient vorticity Shin et al investigation on kinematic

parameters indicate that the strength of an inlet-vortex or trailing vortex system

increases with decreasing distance from the surface However for an inlet in

an upstream irrotational flow two counter rotating vortices can still trail from

the rear of the inlet

2) Classification of vortex type

As illustrated in Fig 24 vortices in the vicinity of pump intakes may be adjacent to

the channel bottom or a channel wall (ie submerged vortices) or they may appear

adjacent to the free surface (ie free surface vortex)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

14

(a) Schematic diagram of vortices in a sump

(b) Experimental results of different vortex

Fig 24 Typical vortex in a sump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

15

For the model visual observation dye was injected around the intake structure near

the surface and submergence as Fig 24 (b) shows And the model test results of

vortex was classified as Fig 25

(a) Free Surface Vortices

(b) Submerged Vortices

Fig 25 Vortex classifications

There are a number of researches focused on the numerical theory to detect the

inception of a visible vortex As the diameters of vortices are usually much smaller

compared with the computing grid size to compensate for the lack of vortex

resolution in numerical simulation a stretching vortex model is applied to the local

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

16

flow field around semi-analytically identified vortex positions [21][22] The brief

explanations are as follows

a) Free surface vortex

The criteria expressed by equation 26 and 27 is applied to determine the

visible inception of a free surface vortex

≫ 1 (26)

=

gt 1 (27)

where P is the pressure drop at the vortex core and Ph is the static head at the

vortex element and f is the drag force acting at a unit length of water surface

vortex dimple based on Stokes approximation and caused by the down flow

Further f is the force acting on the bubble in the direction from the high

pressure region to the low pressure region like buoyancy

b) Submerged vortex

Equation 28 is applied to determine the visible inception of a submerged vortex

infin =

infin

infin gt 1 (28)

where is the static pressure at the vortex core infin is the atmospheric pressure

and is the critical cavitation inception pressure and nearly equal to the

saturated vapor pressure

3) Swirl angle

Swirl angle predicts the intensity of flow rotation Fig 26 shows experiment

calculation method with swirl meter and attempts of the CFD calculation method

were conducted as Fig 27 A four blade zero-pitch swirl meter which is supported

by a low friction bearing will be installed at about four suction pipe diameter (d)

downstream from the pump suction to measure swirl angle of pump approaching

flow For the swirl meter the tip to tip blade diameter is 075d and the length in

flow direction is 06 and one of the four blade is painted yellow as a reference to

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

17

count revolutions As flow swirl is usually unsteady the observation time of swirl

meter reading should be a continuous period of time

Fig 26 Typical swirl meter

(29)

(210)

(211)

Where qV flow rotational speed at swirl meter (ms)

zV average axial velocity at swirl meter (ms)

n Number of revolution per minute (rpm)

d Diameter of pipe at the swirl meter (m)

q flow rate (m3s)

zV

Vqq 1tan -=

)(60

smnd

Vtimestimes

=p

q

)(4

2

smd

qVz

times=

p

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

18

For the swirl angle calculating method with CFD the key point is to obtain the

average tangential velocity so the swirl check circles were created as illustrated in

Fig 27 The swirl check circle is located at the section of 4d height with diameter

of 025d 05d and 075d respectively

Swirl check circle

Fig 27 Schematic view of swirl angle numerical calculation

23

1 7501 501 250

acute++

=aringaringaring

N

i

N

i

NNVNVNV

Vi

q (212)

where

qV average tangential velocity (ms)

V circumferential velocity of check point in check circle

respectively (ms)

N check point total numbers in single check circle

4) Velocity

Velocity measurement at bell throat by Pitot tube was implemented by ANSIHI

method Typical Pitot tube configuration is shown in Fig 28 Eight measuring

points are defined as Fig 29 Time averaged velocity at eight points was obtained

and these eight data were again averaged to give throat sectional spatial velocity

average

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

19

Fig 28 Example of Pitot tube Fig 29 Measuring points at bell throat

224 Test criteria

There is the criteria for satisfactory performance of the sump model test and

summarized in Table 23

1) Free surface and submerged vortices

Free surface vortices entering the pump must be less severe than vortices with

coherent core Dye core vortices may be acceptable only if they occur less than 10

of the time or only for infrequent pump operating conditions

Submerged vortex which starts from the sump walls induce vibration and noise

problems Dye core submerged vortices extending into the pump bell mouth is not

allowed

2) Swirl angles

Swirl angle is used to check the flow rotation in the suction pipe The calculation

method is referred to section 223

Short-term maximum and long-term average values must be less than 5 degrees

Maximum short-term swirl angles up to 7 degrees may be acceptable only if they

occur less than 10 of the time or only for infrequent pump operating conditions

3)Velocity

Each point time-averaged velocity shall not be larger than 10 of sectional average

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

20

Standard deviation of time-varying velocity at each point was obtained and its

value shall not be larger than 10 of the time-averaged velocity at each point

1) Time-averaged

Variation[] = 100acute-

a

a

V

VV aV Averaged at section

2) Time-varying

Standard deviation= )1(

)(1i

2

-

-aring=

n

xxn

i

Variation[] = (Standard deviation X ) x 100

X Averaged at point

Table 23 Permission criteria of ANSIHI

Vortex occurence Swirl angle

Free surface vortex Submerged vortex

Observation

time Above 10 minutes Above 10 minutes

Above 10

minutes

Permission

criteria

Dye-core

Vortex(F3)

below 110 total

time

Dye-core

Vortex(S2)

below 110 total

time

Below

5degConditional

intermittent

below 7deg

225 Remedial measures for problem intakes

Usually in the preliminary test the worst flow condition is simulated And based on

the results remedial measures for problem intakes are discussed in this part

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

21

1) Overall approach flow patterns

The characteristics of flow approaching an intake structure is one of the foremost

considerations for the designer It is difficult to predict the effects of a given set of

flow conditions upstream from an intake structure Fig 210 depicts several typical

approach flow patterns of pump sump with different operating conditions

Fig 210 Examples of approach flow patterns

Generally speaking for the intake problems especially for the multiple pumps

sump there are the main measures using dividing walls placed between the pumps

as a partitioned structure trash racks with elongated bars guide piers or a number

of structural columns at the bay entrance can provide some assistance in controlling

cross-flow baffles flow distributors and screen mesh etc are applied to improve

the flow distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

22

2) Problems around pump intake

a) Free surface vortices

There are a number of factors that have an effect on the formation of surface

vortices To achieve a higher degree of certainty that objectionable surface vortices

do not form modifications can be made to intake structure to allow operation at

practical depths of submergence and the minimum submergence is referred in Table

22 the use of suction umbrellas vertical curtain walls and horizontal gratings

b) Submerged vortices

In the vicinity of intake bell there are the most complicated flow patterns and flow

makes the most changes in direction while maintaining a constant acceleration into

the pump bell to prevent local flow separation turbulence and vortex formation

Fig 211 represents a sample of various devices to inhibit the different submerged

vortex

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

23

Fig 211 Methods to reduce submerged vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

24

c) Preswirl

The approach flow distributions determine whether flow preswirl exists and a

sufficiently laterally skewed approach flow cause rotating around the intake bell

The most effective way of reducing preswirl is to establish a relatively uniform

approach flow within each pump bay by using the baffling schemes controlling

cross-flow and expanding concentrated flow discussed in the above part

To prevent non-uniform velocity distribution swirls and vortices necessary

modifications are conducted such as fore bay length baffle wall distributors

curtain wall mesh screen guide vane floor splitter wall fillet etc As the above

measures applied modification test will be conducted to check all the flow

conditions and compare with the preliminary results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

25

Chapter 3 Methodology

CFD (Computational fluid dynamics) is the technique which can design cycles to

better performance and reduce costs and time In CFD we solve the governing

equations of given physics (may be differential form or integral form) using some

numerical techniques like Finite Difference Method (FDM) Finite Element Method

(FEM) or Finite Volume Method (FVM) Although CFD plays a great part in the

design of turbomachinery CFD is not a replacement for experimental or analytical

approach In this study the numerical analysis was performed by the commercial

code ANSYS CFX 130 All aspects of the CFD study including modelling

meshing solving and post-processing of the results

31 Numerical modeling

ANSYS CFX solves the unsteady Navier-Stokes equations in their conservation

form The instantaneous equation of mass (continuity) in the stationary frame is

expressed as equation

ρ

δ + nabla ∙ (ρU) = 0 (31)

And the instantaneous equation for momentum is expressed as shown in equation

ρ

δ + nabla ∙ (ρU⨂U) = minusnablap + nabla ∙ τ + S (32)

Where τ is the stress tensor ( including both normal and shear components of the

stress)

These instantaneous equations are averaged for turbulent flows leading to additional

terms that need to be solved While the Navier-Stokes equations describe both

laminar and turbulent flows without addition terms realistic flows involve length

scales much smaller than the smallest finite volume mesh A Direct Numerical

Simulation of these flows would require significantly more computing power than

what is available now or in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

26

Therefore much research has been done to predict the effects of turbulence by

using turbulence models These models account for the effects of turbulence

without the use of a very fine mesh or direct numerical simulation

These turbulence models modify the transport equations by adding averaged and

fluctuating components The transport equations are changed to equations (33) and

(34)

ρ

δ + nabla ∙ (ρU) = 0 (33)

ρ

+ nabla ∙ ρ U ⨂U = minusnablap + nabla ∙ τ minus ρu⨂u + S (34)

The mass equation is not changed but the momentum equation contains extra terms

which are the Reynolds stresses ρ u⨂u and the Reynolds flux ρ u ϕ These

Reynolds stresses need to be modeled by additional equations to obtain closure

Obtaining closure implies that there are a sufficient number of equations to solve for

all the unknowns including the Reynolds stresses and Reynolds fluxes

Various turbulence models provide various ways to obtain closure In this

investigation Shear Stress Transport (SST) model and k-ε model were utilized The

advantage of SST model is that combines the advantages of other turbulence models

(the k-ε Wilcox k-ω) The 3 turbulence models will be discussed briefly

32 Turbulence models

Turbulence models are necessary because we cannot afford big enough computers

to directly capture every scale of motion There are three modelling frameworks

Direct Numerical Simulation (DNS) Reynolds-Averaged Navier-Stokes

Equations(RANS) Large Eddy Simulation(LES)

As RANS method is applied in most productionseveral RANS turbulence models

differ in their usage of wall functions the number of additional variables solved for

and what these variables represent

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

27

Two equation turbulence models

Two-equation turbulence models are widely used in CFD as they give a good

compromise between computational power needed and accuracy The term lsquotwo

equationrsquo refers to the fact that these models solve for the velocity and length scales

using separate transport equations The turbulence velocity scale is obtained by

solving the transport equation The turbulent length scale is estimated from two

properties of the turbulence field namely the turbulent kinetic energy and the

dissipation rate The dissipation rate of the kinetic energy is obtained from its

transport equation The most widely used are k-ε and k-ω two equation models In

the next section the k-ε Wilcox k-ω BSL k-ω and SST models will be briefly

discussed

321 k-ε turbulence model

The k-epsilon model solves for two variables k the turbulent kinetic energy and

epsilon the rate of dissipation of kinetic energy Wall functions are used in this

model so the flow in the buffer region is not simulated The k-epsilon model is very

popular for industrial applications due to its good convergence rate and relatively

low memory requirements It does not very accurately compute flow fields that

exhibit adverse pressure gradients strong curvature to the flow or jet flow It does

perform well for external flow problems around complex geometries

The k-ε model introduces k (m2s2) as the turbulence kinetic energy and ε (m2s3) as

the turbulence eddy dissipation The continuity equation remains the same

ρ

+ nabla ∙ (ρ U) = 0 (35)

The momentum equation changes as shown by equation

δρ

δ + nabla ∙ (ρU⨂U) = minusnablapprime + nabla ∙ μ

nablaU + (nablaU) + S (36)

Where SM is the sum of body forces μeff is the effective viscosity accounting for

turbulence and prsquo is the modified pressure The k-ε model uses the concept of eddy

viscosity giving the equation for effective viscosity as shown by equation

μ

= μ + μ (37)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

28

μt is the turbulence viscosity is linked to the turbulence kinetic energy and

dissipation by the equation

μ = Cμρ

ε (38)

Where Cμ is a constant

The values for k and ε come from the differential transport equations for the

turbulence kinetic energy and the turbulence dissipation rate

The turbulence kinetic energy equation is given as equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus ρε (39)

The turbulence dissipation rate is given by equation

(ρε)

+ nabla ∙ (ρUε) = nabla ∙ μ +

μ

σε nablaε +

ε

(Cε (P + Pε ) minus Cε ρε) (310)

Where Cε1 Cε2 σk σε are constants

Pk is the turbulence production due to viscous forces and is modeled by the

equation

P = μ nablaU ∙ nablaU + nablaU minus

nabla ∙ U 3μ

nabla ∙ U + ρk (311)

If a buoyancy term is added to the previous equation if the full buoyancy model is

used

The buoyancy term Pkb is modeled as

P = minusμ

ρσρg ∙ nablaρ (312)

322 Wilcox k-ω turbulence model

The k-omega model is similar to k-epsilon instead however it solves for omega mdash

the specific rate of dissipation of kinetic energy It also uses wall functions and

therefore has comparable memory requirements It has more difficulty converging

and is quite sensitive to the initial guess at the solution Hence the k-epsilon model

is often used first to find an initial condition for solving the k-omega model The k-

omega model is useful in many cases where the k-epsilon model is not accurate

such as internal flows flows that exhibit strong curvature separated flows and jets

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

29

This model has an advantage over the k-ε model where it does not involve complex

linear damping functions for near wall calculations at low Reynolds numbers

The k-ω model assumes that the turbulence viscosity is related to the turbulence

kinetic energy k and the turbulent frequency ω by the equation

μ = ρ

ω (313)

The transport equation for k is given by the equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus βprimeρ k ω (314)

The transport equation for ω is shown as equation

(ρω)

δ + nabla ∙ (ρ U ω) = nabla ∙ μ +

μ

μω

nablaω + αω

P + Pω minus β ρ ω (315)

The production rate of turbulence (Pk) is calculated as shown previously in the k-ε

section

The recommended values for the constants in the above equations are

βprime = 009

α = 59

β = 0075

σ = 2

σω = 2

The Reynolds stress tensor ρ u⨂u is calculated by

minusρ u⨂u = μ nablaU + (nablaU) minus

part ρk + μ

nabla ∙ U (316)

323 Shear stress transport model

The disadvantage of the Wilcox model is the strong sensitivity to free-stream

conditions Therefore a blending of the k-ω model near the surface and the k-ε in

the outer region was made by Menter [23] which resulted in the formulation of the

BSL k-ω turbulence model It consists of a transformation of the k-ε model to a k-ω

formulation and subsequently adding the resulting equations The Wilcox model is

multiplied by a blending function F1 and the transformed k-ε by another function 1-

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

30

F1 F1 is a function of wall distance (being the value of one near the surface and zero

outside the boundary layer) Outside the boundary and on the edge of the boundary

layer the standard k-ε model is used [24]

SST model is a combination of the k-epsilon in the free stream and the k-omega

models near the walls It does not use wall functions and tends to be most accurate

when solving the flow near the wall The SST model does not always converge to

the solution quickly so the k-epsilon or k-omega models are often solved first to

give good initial conditions

The k-ω based Shear Stress Transport (SST) model of Menter was applied for

turbulence treatment The transport equations for the SST model are expressed

below where the turbulent kinetic energy lsquokrsquo and turbulent frequency or dissipation

per unit turbulent kinetic energy are computed by using the following relations

For Turbulence Kinetic Energy

kkPukt

kTkk Ntilde+Ntilde+-=Ntilde+

para

para)[()( nsnwb (317)

where Pk is the production limiter

For Specific Dissipation Rate

ww

swnsnbwaww

ww NtildeNtilde-+Ntilde+Ntilde+-=Ntilde+para

para

1)1(2])[()( 21

22 kFSut

T (318)

The first blending function F1 is calculated from

iumlthorn

iumlyacute

uuml

iumlicirc

iumliacute

igrave

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

uacuteuacuteucirc

ugrave

ecircecirceuml

eacute

dividedivideoslash

oumlccedilccedilegrave

aelig=

4

22

21

4500maxmintanh

yCD

k

yy

kF

kw

wrs

w

n

wb (319)

y is the distance to the nearest wall and v is the kinematic viscosity

divideoslash

oumlccedilegrave

aeligNtildeNtilde= -10

2 101

2max ww

rsww kCDk (320)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

31

And Kinematic eddy viscosity

)max( 21

1

SF

kT

wa

an = (321)

where F2 is a blending function which restricts the limiter to the wall boundary and

S is the invariant measure of the strain rate

uacuteuacute

ucirc

ugrave

ecircecirc

euml

eacute

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

dividedivideoslash

oumlccedilccedilegrave

aelig=

2

22

5002maxtanh

w

n

wb yy

kF (322)

33 Brief introduction of simulation process

All aspects of CFD study including modelling meshing solving and post-

processing of the results are briefly introduced in this chapter

331 Pre-processing technique

The speed and accuracy of simulation results depending fundamentally on meshing

and the ability to prepare model geometry correctly

The steps in pre-processing are explained as follows

1) Domain Definitions

The domain definition depends mainly on the physics involved in the problem and

objective of the problem

a) Geometry creation or geometry import

b) Geometry cleanup

i) Removing the parts not necessary for simulation under consideration

ii) Closing the gaps in the geometry

iii) Removing small surfaces curves to make meshing process simple

2)Grid generation

To solve the governing equations the numerical methods like Finite Difference

Finite Volume or Finite Element Method to divide the domain into small parts and

solve the equations on one part at a time are applied

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 4: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

i

TABLE OF CONTENTS

List of Tablesiii

List of Figuresiv

Abstractvi

Nomenclatureviii

Abbreviationsix

Chapter 1 Introduction1

11 Background1

12 Previous study1

13 Methodology of study3

Chapter 2 Basic theory of sump model4

21 Pump intake design5

211 Inlet bell diameter design 5

212 Recommended dimensions for a rectangular sump7

22 Model test10

221 The scale effects11

222 Principle of similarity12

223 Preliminary model test13

224 Test criteria19

225 Remedial measures for problem intakes20

Chapter 3 Methodology25

31 Numerical modeling25

32 Turbulence models26

321 k-ε turbulence model27

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

ii

322 Wilcox k-ω turbulence model28

323 Shear stress transport model29

33 Brief introduction of simulation process31

331 Pre-processing technique31

332 Solving the simulation32

333 Post processing32

Chapter 4 Description of Model Cases33

41 Creating the geometry33

411 Geometry of the scaled sump model33

412 Geometry of the full size pump sump model35

42 Mesh generation39

43 Numerical approach42

44 Experimental setup45

Chapter 5 Results and Discussion47

51 Results of the scaled sump model47

511 Numerical simulation vortex check47

512 PIV results analysis49

513 Sump with anti free surface vortex devices53

52 Results of the mixed flow pump sump model56

521 Performance analysis of the mixed flow pump sump model56

522 Flow characteristics analysis58

523 Cavitation phenomenon analysis62

Chapter 6 Conclusions60

Acknowledgement67

References68

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

iii

List of Tables

Table 21 Acceptable velocity ranges for inlet bell diameter 6

Table 22 Recommended dimensions for a rectangular sump8

Table 23 Permission criteria of ANSIHI20

Table 41 Designed specifications of scaled sump model33

Table 42 Parameters of anti free surface vortex devices35

Table 43 Design specifications of the mixed flow pump38

Table 44 Mesh information for the pump sump model42

Table 51 Swirl angle of model test results54

Table 52 Summary of swirl angle CFD calculation results54

Table 53 Mixed flow pump performance results56

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

iv

List of Figures

Fig 21 Recommended inlet bell design diameter6

Fig 22 Recommended intake structure layout7

Fig 23 Filler wall details for proper bay width7

Fig 24 Typical vortex in a sump14

Fig 25 Vortex classifications15

Fig 26 Typical swirl meter17

Fig 27 Schematic view of swirl angle numerical calculation 18

Fig 28 Example of Pitot tube 19

Fig 29 Measuring point at bell throat19

Fig 210 Examples of approach flow patterns21

Fig 211 Methods to reduce submerged vortices23

Fig 41 Geometry of scaled sump model33

Fig 42 Schematic diagram of sump with anti free surface vortex devices34

Fig 43 Overall 3D drawing of the sump model35

Fig 44 2D drawing of the sump sketch36

Fig 45 Shape of the mixed flow pump37

Fig 46 Sump model with the submerged AVD38

Fig 47 Submerged AVD dimensions39

Fig 48 Overall mesh generation for scaled sump models40

Fig 49 Overall sump fluid domain mesh for the pump sump model41

Fig 410 Grid details of the pump sump model41

Fig 411 Boundaries of the scaled sump43

Fig 412 Experimental setup for the scaled sump model46

Fig 51 Streamline pattern of free surface47

Fig 52 Vortex core region of the sump model48

Fig 53 Velocity vectors on different sections48

Fig 54 Velocity vectors of PIV results51

Fig 55 Velocity vectors on different sections (PIV)52

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

v

Fig 56 Velocity vectors on free surface of different cases53

Fig 57 Tangential velocity profile of different models55

Fig 58 Performance curve of the mixed flow pump with the sump 57

Fig 59 Streamline pattern around the intake structure58

Fig 510 Vortex core region of the pump sump model59

Fig 511 Vorticity distribution in the flow direction60

Fig 512 Vorticity distribution in channel width direction61

Fig 513 Cavitation region of the mixed flow pump62

Fig 514 Cavitation performance curves of the mixed flow pump63

Fig 515 Vapor volume fraction distribution on blade surface64

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

vi

Study of Flow Field in Rectangular Sump Models and

Performance Analysis of a Mixed Flow Pump

Yuxin ZHAO

Department of Mechanical Engineering

Graduate School of Korea Maritime and Ocean University

Abstract

It is an accepted fact that quite a number of problems faced in a pumping station are

related to the design of sump or intake rather than pump design Head-capacity

curves provided by the pump manufacturer are obtained on the condition of no

vortices flowing into the pump intake The efficiency and performance of pumping

stations depend not only on the performance of the selected pumps but also on the

proper design of the intake sumps A faulty design of pump sump can lead to the

occurrence of swirl and vortices which reduce the pump performance and induce

vibration and additional noise Therefore sump model test is necessary to check the

flow condition around intake structure Numerical simulation is a good facility for

reducing the time and cost involved throughout the design process In this study the

commercial software ANSYS CFX-130 has been used for the CFD analysis of the

sump models

In a scaled sump model for air entrainment simulation numerical analysis of single

phase and two-phase with SST turbulence model was carried out to predict vortex

(both free surface vortex and submerged vortex) occurrence and location The

effectiveness of curtain walls and square bars to eliminate the free surface vortex is

evaluated Meanwhile the experiment including PIV system was performed in

KMOU to investigate the flow conditions around pump intake structure

In a full size sump model overall numerical analysis for the sump model with a

mixed flow pump installed was carried out Hydraulic performances of the mixed

flow pump for head rise shaft power pump efficiencies versus flow rate changed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

vii

from 50 to 140 of the design flow rate were studied by the performance curves

A trident shaped anti vortex device (AVD) composed of three wall fillets and one

center splitter was installed under the pump intake The effectiveness of the AVD

for the suppression of submerged vortex was evaluated In addition numerical

simulation of cavitation phenomenon in the mixed flow pump has been performed

by calculating the full cavitation model with k-ε turbulence model

According to the results details of the location size and strength of vortices were

predicted in the numerical simulation For the scaled sump model although there is

a little difference between the single phase and two-phase simulation all the results

predict the free surface vortex and submerged vortex formation and location The

CFD simulations of flow condition show good agreement with the PIV results The

effectiveness of curtain walls installed to inhibit the free surface vortex is confirmed

in the CFD results while the square bars installed in the channel failed to suppress

the free surface vortex For the mixed flow pump with sump the best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

The effectiveness of AVDs installed for the submerged vortex is confirmed by

comparisons of the sump model with and without AVD results

Cavitation phenomenon analysis shows the pump operating at the BEP condition

has a considerable extent over the requirement of cavitation performance With

numerical simulation the inception of cavitation in the blade passage is observed on

the suction surface where the leading edges meet the tips and then as the inlet total

pressure decrease the cavitation zone is spread out over the suction surface as well

as the leading edge of the impeller blade

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

viii

Nomenclature

B Distance from the back wall to the pump inlet bell centerline [m]

C Distance between the inlet bell and floor [m]

D Inlet bell design outside diameter [m]

Fr Froude number (=Vb(gD)05) [-]

g Acceleration due to gravity [ms2]

ht Total head [m]

n Number of revolution per min [-]

Pflow Flow power [W]

Pshaft Shaft power [W]

Pν Vapor pressure [Pa]

Q Volume flow rate [m3s]

Q0 Design volume flow rate [m3s]

Re Reynolds number (=VbDν) [-]

S Minimum submergence depth [m]

Vb Velocity at bell mouth [ms]

α Angle of floor slope [deg]

η Pump efficiency []

θ Swirl angle [deg]

λ The scale ratio of model dimensions to prototype [-]

μ Viscosity [Pa s]

ν Kinematic viscosity [m2s]

ρ Specific water density [kgm3]

σ Surface tension [Nm]

τ Stress tensor [-]

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

ix

Abbreviations

AVD Anti vortex device

BEP Best efficiency point

CFD Computational fluid Dynamics

NPSH Net positive suction head

NPSHR Required net positive suction head

PIV Particle Image Velocimetry

RANS Reynolds Averaged Navier- Stokes

RPM Number of revolutions per minute

SST Shear Stress Transport

VOF Volume of Fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

1

Chapter 1 Introduction

11 Background

It is an accepted fact that quite a number of problems faced in a pumping station are

related to the design of sump or intake rather than pump design As an important

part of the pumping station sump is designed to provide uniform swirl and

entrained air free flow to the pump intake Undesirable flow conditions (entrained

air non-uniform flow distribution vortices etc) can reduce the pump efficiency

induce vibration noise and cavitation even lead to excessive bearing loads of

impeller and structure damage Due to quite complex nature of the flow although

the flow conditions causing pump problems are well established there are no

specific solutions to eliminate them There are some design guidelines for the

specific geometrical and hydraulic constraints of the pump sump for any project

the model study is the only tool for solving potential problems in new designs and

rectifying problems observed in existing installations However the scaled model

tests are expensive and time-consuming so alternative Computational Fluid

Dynamics (CFD) methods for evaluating sump performance have been developed

With rapid progress in CFD numerical simulation is regarded as an effective tool in

solving fluid problems in pump sumps

12 Previous study

Various researches show the flow conditions around the sump intake structure are

quite complicated Shyam et al [1] used the commercial code CFX to carry out

two-phase flow simulation to capture air entrainment The air entrainments and its

location were well captured and numerical results were in accordance with the

experimental investigation Matsui J[2] and Okamura T[3] et al conducted a serial

research of the benchmark according to the Turbomachinery Society of Japan (TSJ)

standard including various commercial codes comparison performed in the flow

simulation and comparison analysis between the numerical and experimental results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

2

The purpose of the benchmark is to determine the accuracy and reliability of the

CFD codes used in universities and the industrial field It is also intended to identify

and investigate the formation of air-entrained and submerged vortices near the

pump Tanweer et al[4] have performed a numerical analysis of a multiple intake

pump sump to confirm that the geometry of sump affects the flow within sump

greatly Rajendran et al[5] have made a numerical analysis for the flow

characteristics of a sump model with pump intake and good agreements were

achieved by comparing the numerical results with the experiments Iwano et al[6]

have introduced a numerical method for the submerged vortex by analyzing the

flow in the pump sump with and without baffle plates Lee et al[7] have conducted

the CFD analysis of a multi-intake pump sump model to check the flow uniformity

by predicting the location number and vorticity of the vortex For the CFD

application in mixed flow pump design and performance analysis Oh H W [8]

conducted a mixed flow pump design optimization with specific parameters and

then compared the numerical results with the experiments The procedure presented

can be used as a practical design and analysis guide for the general turbomachinery

Kim et al[9] carried out a mixed flow pump performance analysis with numerical

simulation and improved the hydraulic efficiency by the modification of the

geometry

As an undesirable phenomenon caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage The cavitation

phenomenon has been studied by various investigators Van et al[10] have

investigated the cavitation inception behavior of a mixed-flow pump impellers

model and the calculated results matched well with the experiments Okamura et

al[11] have evaluated two numerical cavitation prediction methods used in the

pump industry manufacturers and attempts were made to improve the cavitation

performance Li et al[12] have conducted the numerical analysis on the basis of the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

3

development liquidvapor interface tracking method to predict the cavitation

characteristics with a centrifugal pump impeller model

Moreover there are a number of correlated studies [13]~[17]

13 Methodology of study

This study will investigate the flow phenomena of two sump models

A scaled sump model is adopted to carry out numerical simulation and PIV

experiment The flow conditions are predicted and the effectiveness of devices for

free surface vortex is evaluated by using numerical code a single intake sump

model with a mixed flow pump installed is modeled and the hydraulic performance

characteristics for the head rise shaft power and pump efficiencies versus flow rate

are studied In addition flow phenomena (vortices cavitation etc) are analyzed

and the effectiveness of an anti-vortex device (AVD) for the submerged vortex is

evaluated

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

4

Chapter 2 Basic theory of sump model

Specific hydraulic phenomena have been confirmed that they can lead to certain

common operational problems The common of the main problems are summarized

as follows

a Free surface vortices

b Submerged vortices

c Swirl of flow entering the pump

d Non-uniform distribution of velocity at the pump impeller

e Entrained air or gas bubbles

These problems encountered in the pump sump will affect the pump

performance and significantly increase the operational and maintenance costs In

order to identify sources of particular problems and find practical solution for

it the usual approach is to conduct the laboratory experiments on a scaled

physical model Typical design objectives are to ensure that a pump station is

designed according to best practice and conforms to the requirements set out in the

American National Standard for Pump Intake Design[18] These standards restrict

the degree to which the above-mentioned undesirable flow patterns may be present

The negative impact of each of these hydraulic conditions varies with pump specific

speed and size Application of the standard to design a sump does not generate a

problem free sump but provides a basis for the initial design

As there are no specific guidelines or criteria for design of trouble free intakes the

most common solution to potential problems in new design and rectification of

problems observed in existing designs is to construct a scaled model in a laboratory

observe and investigate the flow characteristics and propose modifications to the

intake geometry There are additional devices in the form of floor splitters or cones

back wall splitters fillets surface beams guide vanes etc aimed at controlling the

vortex and swirl formation Considerable prior experience and ingenuity are

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

5

required to solve the problems by this method and the cost and time required are

also significant

The intake structure refers to the channel leading from the water source which

may be a river or reservoir and all installations downstream including the pump

column or the intake pipe (the suction tube portion of a vertical intake) the

approach channel (upstream of pump bay) and the pump bay (bounded by the floor

the back wall side walls dividers walls separating adjacent pump column) Usually

the upstream end of the pump column has an inlet attachment called the suction bell

The function of the intake is to supply an evenly distributed flow of water to the

pump suction bell

Several types of pump intake basins are available and ANSIHI 98 provides

design requirements for each Intake structures can be categorized as being

clear liquids or solids-bearing liquids For clear liquids intakes are further

classified into rectangular formed circular and trench types as well as suction

tanks and cans For solid-bearing liquids trench type and rectangular wet wells are

usually considered

In this paper only rectangular intake structure for clear liquid is studied

21 Pump intake design

For pumps to achieve their optimum hydraulic performance across all operating

conditions the flow at the impeller must meet specific hydraulic conditions The

ideal flow entering the pump inlet should be of a uniform velocity distribution

without rotation and stable over time When designing a sump to achieve favorable

inflow to the pump or suction pipe bell various sump dimensions relative to the size

of the bell are required Using these standards of the distances to reduce the

probability of the occurence of strong submerged vortices and free-surface vortices

211 Inlet bell diameter design

The geometry is generally defined in terms of the pump inlet diameter as shown in

Table 21 and Fig 21 The bell diameter can be estimated based on an inlet pipe

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

6

Table 21 Acceptable velocity ranges for inlet bell diameter(ANSIHI 98)

Flow Rate

Qls

Recommended Inlet Bell Design

Velocityms

Acceptable Velocity

Range ms

Q<315 V=17 06leVle27

315 leQ<1260 V=17 09leVle24

Q ge1260 V=17 12leVle21

Fig 21 Recommended inlet bell design diameter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

7

212 Recommended dimensions for a rectangular sump

As the selected bell diameter has been determined the proportions of the inlet

structure can be estimated from Fig 22 and Fig 23 Table 22 shows the

recommended dimensions for a rectangular sump

Fig 22 Recommended intake structure layout

Fig 23 Filler wall details for proper bay width

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

8

Table 22 Recommended dimensions for a rectangular sump

Dimension Variable

Description Recommended value

A Distance from the pump inlet bell

centerline to the entrance A=5D minimum

a Length of constricted bay section near

the pump inlet 25D minimum

B Distance from the back wall to the

pump inlet bell centerline 075D

C Distance between the inlet bell and

floor 03D~05D

D Inlet bell outside diameter refer to section 211

H Minimum liquid depth H=S+C

h Minimum height of constricted bay

section near the pump inlet bell hgtH or 25D

S Minimum pump inlet bell submergence S=D(10+23Fr)

W Pump inlet bay entrance width 2D minimum

w Constricted bay width near the pump

inlet bell w=2D

X Pump inlet bay length 5D minimum

Y Distance from pump inlet bell

centerline to the through-flow traveling screen

4D minimum

Z1 Distance from sump inlet bell centerline to diverging walls

5D minimum

Z2 Distance from inlet bell centerline to

sloping floor 5D minimum

α Angle of floor slope -10deg~ +10deg

β Angle of wall convergence 0deg~ +10deg

empty Angle of convergence from constricted

area to by walls 10deg maximum

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

9

FD=Vb(gD)05 (21)

where

FD Froude number (dimensionless)

Vb velocity at suction inlet=FlowArea based on D

D Outside diameter of bell or pipe inlet

g gravitational acceleration

There is the design sequence summary for rectangular intake structures

middot Consider the flow patterns and boundary geometry of the body of liquid

from which the pump station is to receive flow Compare with the approach

flow condition and determine if a hydraulic physical model study is

required

middot Determine the number and size of pumps required to satisfy the range of

operating conditions likely to be encountered

middot Identify pump inlet bell diameter Refer to the section 211

middot Determine the bell-floor clearance 05D is a good preliminary design value

middot Determine the required bell submergence using the equation 21

middot Determine the minimum allowable liquid depth in the intake structure from

the sum of the floor clearance and the required bell submergence

middot Check bottom elevation near the entrance to the structure and determine if it

is necessary to slope the floor upstream of the bay entrance If the resulting

depth at the entrance to the intake structure is shallow then check to ensure

that gravity-driven flow is not restricted by the entrance condition

middot Check the pump bay velocity for the maximum single-pump flow and

minimum liquid depth with the bay width set to 2D If bay velocity exceeds

05 ms then increase the bay width to reduce the maximum velocity below

05 ms

middot If it is necessary to increase the pump bay width to greater than 2D then

decrease bay width in the vicinity of the pumps according to Fig 23

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

10

middot Compare cross-flow velocity (at maximum system flow) to average pump

bay velocity If cross-flow value exceeds 50 of the bay velocity a

physical hydraulic model study is necessary

middot Determine the length of the structure and dividing walls giving

consideration to minimum allowable distances to a sloping floor screening

equipment and length of dividing walls if dual flow traveling screens or

drum screens are to be used a physical hydraulic model study is required

middot If the final selected pump bell diameter and inlet velocity is within the

range given in section 211 then the sump dimensions (developed based on

the inlet bell design diameter) need not be changed and will comply with

these standards

22 Model test

The approach flow pattern in a pump intake facility is difficult to predict by

traditional mathematics or empirical formulae For large structures or those that

differ significantly from proven designs a model study is the only means to ensure

success A model study is used to identify adverse hydraulic conditions and

derive remedial measures for approach flow patterns generated by structures

upstream of the pump impeller Current model studies are not intended to

investigate flow patterns induced by the pump itself or the flow patterns within the

pump The objective of a model study is to ensure that the pump intake structure

generates favorable flow conditions at the inlet to the pump Evaluation for the

requirement of model test if

a Sump or piping geometry that deviates from this design standard

b Non-uniform or non-symmetric approach flow to the pump sump

exist

c The pumps have flows greater than 2520 ls per pump or the total

station flow with all pumps running would be greater than 6310 ls

d Situations that need ten times cost than the model study

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

11

Adverse hydraulic conditions that can affect pump performance include free and

submerged vortices swirl approaching the pump impeller flow separation at the

pump bell and a non-uniform axial velocity distribution at the suction

The purpose of model test is to examine these adverse hydraulic conditions and to

investigate measure against them

221 The scale effects

One possible difficulty in scale modeling is that it is impossible to reduce all

pertinent forces by the same factor With Froude-scale models the inertial and

gravitational forces are reduced similarly but the viscous and surface tension forces

cannot be simultaneously reduced The extra influence of these forces is known as

a ldquoscale effectrdquo A number of investigations have indicated that the scale effects for

the model are avoided if the Reynolds number based on the inlet flow and

submergence or intake diameter exceeds approximately 3x 104 and the Weber

number exceeds 150 [19]

In most engineering applications involving closed-conduit and open-channel flow

the Reynolds number limits are far exceeded and the flows are fully turbulent

In a sump model Froude number (Fr) and Reynolds number (Re) are the most

important non-dimensional parameters Some reduction in Re should be

introduced to the sump model However flow patterns are generally very similar

at high Re

No specific geometric scale rate is recommended but reasonable large

geometric scale to minimize viscous and surface tension scale effects and reproduce

the flow pattern

Re= uDν (22)

We= u2ρDσ (23)

where

u average axial velocity (ms)

ν kinematic viscosity of the liquid (m2s)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

12

ρ liquid density (kgm3)

σ surface tension of liquidair interface

According to the ANSIHI 98 the viscous scale effects on vortex may be negligible

if the resulting dimensionless numbers must meet these values Re>6times104 and

We>240

To allow practical visual observations of flow patterns accurate measurements of

swirl and velocity distribution and sufficient dimensional control the model scale

shall yield a bay width of at least 300mm a minimum liquid depth of at least

150mm and a pump throat or suction diameter of at least 80mm

222 Principle of similarity

In the similitude analysis the geometric and flow similarity requires the model and

the flow to be identical as the real model and flow This is to ensure the result

obtained from the model study is well presented in order to predict full scale

behavior Model involving a free surface are operated using Froude similarity since

the flow process is controlled by gravity and inertial forces

For similarity of the flow patterns the Froude number shall be equal in model and

prototype

Fr =

=

(24)

and as the scale ratio λ was selected in section 221 the following equation is

obtained

V = V times (

) = V times λ or Q = Q times (

) = Q times λ (25)

where Fr is the Froude number λ is the scale ratio of model dimensions to prototype

The test of 15 times Froude scaled flow should be conducted to ascertain the

potential scale affect on free surface vortices

Models of closed conduit piping system leading to a pump suction are not operated

based on Froude similitude However flow patterns are generally very similar at

high Re A minimum value of 1times105 for the Re at the pump suction is

recommended to ensure the flow patterns are correctly scaled

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

13

223 Preliminary model test

The initial design shall be tested first to identify any hydraulic problems

1) Vortex Formation Mechanisms

There are various investigations that provide insight into the fundamental processes

leading to the development of vortices both in experiment and numerical simulation

Shin et al [20] demonstrated that two basic mechanisms lead to inlet vortex

formation The first mechanism involves the development of an inlet vortex due to

the amplification of ambient vorticity in the approach flow as vortex lines

are convicted into the inlet The second mechanism involves the development of a

trailing vortex in the vicinity of the intake as a result of the variation in

circulation along the inlet For this second case a vortex can develop in a flow that

is irrotational upstream and the vortex development therefore does not depend

on the presence of ambient vorticity Shin et al investigation on kinematic

parameters indicate that the strength of an inlet-vortex or trailing vortex system

increases with decreasing distance from the surface However for an inlet in

an upstream irrotational flow two counter rotating vortices can still trail from

the rear of the inlet

2) Classification of vortex type

As illustrated in Fig 24 vortices in the vicinity of pump intakes may be adjacent to

the channel bottom or a channel wall (ie submerged vortices) or they may appear

adjacent to the free surface (ie free surface vortex)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

14

(a) Schematic diagram of vortices in a sump

(b) Experimental results of different vortex

Fig 24 Typical vortex in a sump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

15

For the model visual observation dye was injected around the intake structure near

the surface and submergence as Fig 24 (b) shows And the model test results of

vortex was classified as Fig 25

(a) Free Surface Vortices

(b) Submerged Vortices

Fig 25 Vortex classifications

There are a number of researches focused on the numerical theory to detect the

inception of a visible vortex As the diameters of vortices are usually much smaller

compared with the computing grid size to compensate for the lack of vortex

resolution in numerical simulation a stretching vortex model is applied to the local

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

16

flow field around semi-analytically identified vortex positions [21][22] The brief

explanations are as follows

a) Free surface vortex

The criteria expressed by equation 26 and 27 is applied to determine the

visible inception of a free surface vortex

≫ 1 (26)

=

gt 1 (27)

where P is the pressure drop at the vortex core and Ph is the static head at the

vortex element and f is the drag force acting at a unit length of water surface

vortex dimple based on Stokes approximation and caused by the down flow

Further f is the force acting on the bubble in the direction from the high

pressure region to the low pressure region like buoyancy

b) Submerged vortex

Equation 28 is applied to determine the visible inception of a submerged vortex

infin =

infin

infin gt 1 (28)

where is the static pressure at the vortex core infin is the atmospheric pressure

and is the critical cavitation inception pressure and nearly equal to the

saturated vapor pressure

3) Swirl angle

Swirl angle predicts the intensity of flow rotation Fig 26 shows experiment

calculation method with swirl meter and attempts of the CFD calculation method

were conducted as Fig 27 A four blade zero-pitch swirl meter which is supported

by a low friction bearing will be installed at about four suction pipe diameter (d)

downstream from the pump suction to measure swirl angle of pump approaching

flow For the swirl meter the tip to tip blade diameter is 075d and the length in

flow direction is 06 and one of the four blade is painted yellow as a reference to

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

17

count revolutions As flow swirl is usually unsteady the observation time of swirl

meter reading should be a continuous period of time

Fig 26 Typical swirl meter

(29)

(210)

(211)

Where qV flow rotational speed at swirl meter (ms)

zV average axial velocity at swirl meter (ms)

n Number of revolution per minute (rpm)

d Diameter of pipe at the swirl meter (m)

q flow rate (m3s)

zV

Vqq 1tan -=

)(60

smnd

Vtimestimes

=p

q

)(4

2

smd

qVz

times=

p

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

18

For the swirl angle calculating method with CFD the key point is to obtain the

average tangential velocity so the swirl check circles were created as illustrated in

Fig 27 The swirl check circle is located at the section of 4d height with diameter

of 025d 05d and 075d respectively

Swirl check circle

Fig 27 Schematic view of swirl angle numerical calculation

23

1 7501 501 250

acute++

=aringaringaring

N

i

N

i

NNVNVNV

Vi

q (212)

where

qV average tangential velocity (ms)

V circumferential velocity of check point in check circle

respectively (ms)

N check point total numbers in single check circle

4) Velocity

Velocity measurement at bell throat by Pitot tube was implemented by ANSIHI

method Typical Pitot tube configuration is shown in Fig 28 Eight measuring

points are defined as Fig 29 Time averaged velocity at eight points was obtained

and these eight data were again averaged to give throat sectional spatial velocity

average

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

19

Fig 28 Example of Pitot tube Fig 29 Measuring points at bell throat

224 Test criteria

There is the criteria for satisfactory performance of the sump model test and

summarized in Table 23

1) Free surface and submerged vortices

Free surface vortices entering the pump must be less severe than vortices with

coherent core Dye core vortices may be acceptable only if they occur less than 10

of the time or only for infrequent pump operating conditions

Submerged vortex which starts from the sump walls induce vibration and noise

problems Dye core submerged vortices extending into the pump bell mouth is not

allowed

2) Swirl angles

Swirl angle is used to check the flow rotation in the suction pipe The calculation

method is referred to section 223

Short-term maximum and long-term average values must be less than 5 degrees

Maximum short-term swirl angles up to 7 degrees may be acceptable only if they

occur less than 10 of the time or only for infrequent pump operating conditions

3)Velocity

Each point time-averaged velocity shall not be larger than 10 of sectional average

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

20

Standard deviation of time-varying velocity at each point was obtained and its

value shall not be larger than 10 of the time-averaged velocity at each point

1) Time-averaged

Variation[] = 100acute-

a

a

V

VV aV Averaged at section

2) Time-varying

Standard deviation= )1(

)(1i

2

-

-aring=

n

xxn

i

Variation[] = (Standard deviation X ) x 100

X Averaged at point

Table 23 Permission criteria of ANSIHI

Vortex occurence Swirl angle

Free surface vortex Submerged vortex

Observation

time Above 10 minutes Above 10 minutes

Above 10

minutes

Permission

criteria

Dye-core

Vortex(F3)

below 110 total

time

Dye-core

Vortex(S2)

below 110 total

time

Below

5degConditional

intermittent

below 7deg

225 Remedial measures for problem intakes

Usually in the preliminary test the worst flow condition is simulated And based on

the results remedial measures for problem intakes are discussed in this part

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

21

1) Overall approach flow patterns

The characteristics of flow approaching an intake structure is one of the foremost

considerations for the designer It is difficult to predict the effects of a given set of

flow conditions upstream from an intake structure Fig 210 depicts several typical

approach flow patterns of pump sump with different operating conditions

Fig 210 Examples of approach flow patterns

Generally speaking for the intake problems especially for the multiple pumps

sump there are the main measures using dividing walls placed between the pumps

as a partitioned structure trash racks with elongated bars guide piers or a number

of structural columns at the bay entrance can provide some assistance in controlling

cross-flow baffles flow distributors and screen mesh etc are applied to improve

the flow distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

22

2) Problems around pump intake

a) Free surface vortices

There are a number of factors that have an effect on the formation of surface

vortices To achieve a higher degree of certainty that objectionable surface vortices

do not form modifications can be made to intake structure to allow operation at

practical depths of submergence and the minimum submergence is referred in Table

22 the use of suction umbrellas vertical curtain walls and horizontal gratings

b) Submerged vortices

In the vicinity of intake bell there are the most complicated flow patterns and flow

makes the most changes in direction while maintaining a constant acceleration into

the pump bell to prevent local flow separation turbulence and vortex formation

Fig 211 represents a sample of various devices to inhibit the different submerged

vortex

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

23

Fig 211 Methods to reduce submerged vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

24

c) Preswirl

The approach flow distributions determine whether flow preswirl exists and a

sufficiently laterally skewed approach flow cause rotating around the intake bell

The most effective way of reducing preswirl is to establish a relatively uniform

approach flow within each pump bay by using the baffling schemes controlling

cross-flow and expanding concentrated flow discussed in the above part

To prevent non-uniform velocity distribution swirls and vortices necessary

modifications are conducted such as fore bay length baffle wall distributors

curtain wall mesh screen guide vane floor splitter wall fillet etc As the above

measures applied modification test will be conducted to check all the flow

conditions and compare with the preliminary results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

25

Chapter 3 Methodology

CFD (Computational fluid dynamics) is the technique which can design cycles to

better performance and reduce costs and time In CFD we solve the governing

equations of given physics (may be differential form or integral form) using some

numerical techniques like Finite Difference Method (FDM) Finite Element Method

(FEM) or Finite Volume Method (FVM) Although CFD plays a great part in the

design of turbomachinery CFD is not a replacement for experimental or analytical

approach In this study the numerical analysis was performed by the commercial

code ANSYS CFX 130 All aspects of the CFD study including modelling

meshing solving and post-processing of the results

31 Numerical modeling

ANSYS CFX solves the unsteady Navier-Stokes equations in their conservation

form The instantaneous equation of mass (continuity) in the stationary frame is

expressed as equation

ρ

δ + nabla ∙ (ρU) = 0 (31)

And the instantaneous equation for momentum is expressed as shown in equation

ρ

δ + nabla ∙ (ρU⨂U) = minusnablap + nabla ∙ τ + S (32)

Where τ is the stress tensor ( including both normal and shear components of the

stress)

These instantaneous equations are averaged for turbulent flows leading to additional

terms that need to be solved While the Navier-Stokes equations describe both

laminar and turbulent flows without addition terms realistic flows involve length

scales much smaller than the smallest finite volume mesh A Direct Numerical

Simulation of these flows would require significantly more computing power than

what is available now or in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

26

Therefore much research has been done to predict the effects of turbulence by

using turbulence models These models account for the effects of turbulence

without the use of a very fine mesh or direct numerical simulation

These turbulence models modify the transport equations by adding averaged and

fluctuating components The transport equations are changed to equations (33) and

(34)

ρ

δ + nabla ∙ (ρU) = 0 (33)

ρ

+ nabla ∙ ρ U ⨂U = minusnablap + nabla ∙ τ minus ρu⨂u + S (34)

The mass equation is not changed but the momentum equation contains extra terms

which are the Reynolds stresses ρ u⨂u and the Reynolds flux ρ u ϕ These

Reynolds stresses need to be modeled by additional equations to obtain closure

Obtaining closure implies that there are a sufficient number of equations to solve for

all the unknowns including the Reynolds stresses and Reynolds fluxes

Various turbulence models provide various ways to obtain closure In this

investigation Shear Stress Transport (SST) model and k-ε model were utilized The

advantage of SST model is that combines the advantages of other turbulence models

(the k-ε Wilcox k-ω) The 3 turbulence models will be discussed briefly

32 Turbulence models

Turbulence models are necessary because we cannot afford big enough computers

to directly capture every scale of motion There are three modelling frameworks

Direct Numerical Simulation (DNS) Reynolds-Averaged Navier-Stokes

Equations(RANS) Large Eddy Simulation(LES)

As RANS method is applied in most productionseveral RANS turbulence models

differ in their usage of wall functions the number of additional variables solved for

and what these variables represent

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

27

Two equation turbulence models

Two-equation turbulence models are widely used in CFD as they give a good

compromise between computational power needed and accuracy The term lsquotwo

equationrsquo refers to the fact that these models solve for the velocity and length scales

using separate transport equations The turbulence velocity scale is obtained by

solving the transport equation The turbulent length scale is estimated from two

properties of the turbulence field namely the turbulent kinetic energy and the

dissipation rate The dissipation rate of the kinetic energy is obtained from its

transport equation The most widely used are k-ε and k-ω two equation models In

the next section the k-ε Wilcox k-ω BSL k-ω and SST models will be briefly

discussed

321 k-ε turbulence model

The k-epsilon model solves for two variables k the turbulent kinetic energy and

epsilon the rate of dissipation of kinetic energy Wall functions are used in this

model so the flow in the buffer region is not simulated The k-epsilon model is very

popular for industrial applications due to its good convergence rate and relatively

low memory requirements It does not very accurately compute flow fields that

exhibit adverse pressure gradients strong curvature to the flow or jet flow It does

perform well for external flow problems around complex geometries

The k-ε model introduces k (m2s2) as the turbulence kinetic energy and ε (m2s3) as

the turbulence eddy dissipation The continuity equation remains the same

ρ

+ nabla ∙ (ρ U) = 0 (35)

The momentum equation changes as shown by equation

δρ

δ + nabla ∙ (ρU⨂U) = minusnablapprime + nabla ∙ μ

nablaU + (nablaU) + S (36)

Where SM is the sum of body forces μeff is the effective viscosity accounting for

turbulence and prsquo is the modified pressure The k-ε model uses the concept of eddy

viscosity giving the equation for effective viscosity as shown by equation

μ

= μ + μ (37)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

28

μt is the turbulence viscosity is linked to the turbulence kinetic energy and

dissipation by the equation

μ = Cμρ

ε (38)

Where Cμ is a constant

The values for k and ε come from the differential transport equations for the

turbulence kinetic energy and the turbulence dissipation rate

The turbulence kinetic energy equation is given as equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus ρε (39)

The turbulence dissipation rate is given by equation

(ρε)

+ nabla ∙ (ρUε) = nabla ∙ μ +

μ

σε nablaε +

ε

(Cε (P + Pε ) minus Cε ρε) (310)

Where Cε1 Cε2 σk σε are constants

Pk is the turbulence production due to viscous forces and is modeled by the

equation

P = μ nablaU ∙ nablaU + nablaU minus

nabla ∙ U 3μ

nabla ∙ U + ρk (311)

If a buoyancy term is added to the previous equation if the full buoyancy model is

used

The buoyancy term Pkb is modeled as

P = minusμ

ρσρg ∙ nablaρ (312)

322 Wilcox k-ω turbulence model

The k-omega model is similar to k-epsilon instead however it solves for omega mdash

the specific rate of dissipation of kinetic energy It also uses wall functions and

therefore has comparable memory requirements It has more difficulty converging

and is quite sensitive to the initial guess at the solution Hence the k-epsilon model

is often used first to find an initial condition for solving the k-omega model The k-

omega model is useful in many cases where the k-epsilon model is not accurate

such as internal flows flows that exhibit strong curvature separated flows and jets

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

29

This model has an advantage over the k-ε model where it does not involve complex

linear damping functions for near wall calculations at low Reynolds numbers

The k-ω model assumes that the turbulence viscosity is related to the turbulence

kinetic energy k and the turbulent frequency ω by the equation

μ = ρ

ω (313)

The transport equation for k is given by the equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus βprimeρ k ω (314)

The transport equation for ω is shown as equation

(ρω)

δ + nabla ∙ (ρ U ω) = nabla ∙ μ +

μ

μω

nablaω + αω

P + Pω minus β ρ ω (315)

The production rate of turbulence (Pk) is calculated as shown previously in the k-ε

section

The recommended values for the constants in the above equations are

βprime = 009

α = 59

β = 0075

σ = 2

σω = 2

The Reynolds stress tensor ρ u⨂u is calculated by

minusρ u⨂u = μ nablaU + (nablaU) minus

part ρk + μ

nabla ∙ U (316)

323 Shear stress transport model

The disadvantage of the Wilcox model is the strong sensitivity to free-stream

conditions Therefore a blending of the k-ω model near the surface and the k-ε in

the outer region was made by Menter [23] which resulted in the formulation of the

BSL k-ω turbulence model It consists of a transformation of the k-ε model to a k-ω

formulation and subsequently adding the resulting equations The Wilcox model is

multiplied by a blending function F1 and the transformed k-ε by another function 1-

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

30

F1 F1 is a function of wall distance (being the value of one near the surface and zero

outside the boundary layer) Outside the boundary and on the edge of the boundary

layer the standard k-ε model is used [24]

SST model is a combination of the k-epsilon in the free stream and the k-omega

models near the walls It does not use wall functions and tends to be most accurate

when solving the flow near the wall The SST model does not always converge to

the solution quickly so the k-epsilon or k-omega models are often solved first to

give good initial conditions

The k-ω based Shear Stress Transport (SST) model of Menter was applied for

turbulence treatment The transport equations for the SST model are expressed

below where the turbulent kinetic energy lsquokrsquo and turbulent frequency or dissipation

per unit turbulent kinetic energy are computed by using the following relations

For Turbulence Kinetic Energy

kkPukt

kTkk Ntilde+Ntilde+-=Ntilde+

para

para)[()( nsnwb (317)

where Pk is the production limiter

For Specific Dissipation Rate

ww

swnsnbwaww

ww NtildeNtilde-+Ntilde+Ntilde+-=Ntilde+para

para

1)1(2])[()( 21

22 kFSut

T (318)

The first blending function F1 is calculated from

iumlthorn

iumlyacute

uuml

iumlicirc

iumliacute

igrave

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

uacuteuacuteucirc

ugrave

ecircecirceuml

eacute

dividedivideoslash

oumlccedilccedilegrave

aelig=

4

22

21

4500maxmintanh

yCD

k

yy

kF

kw

wrs

w

n

wb (319)

y is the distance to the nearest wall and v is the kinematic viscosity

divideoslash

oumlccedilegrave

aeligNtildeNtilde= -10

2 101

2max ww

rsww kCDk (320)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

31

And Kinematic eddy viscosity

)max( 21

1

SF

kT

wa

an = (321)

where F2 is a blending function which restricts the limiter to the wall boundary and

S is the invariant measure of the strain rate

uacuteuacute

ucirc

ugrave

ecircecirc

euml

eacute

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

dividedivideoslash

oumlccedilccedilegrave

aelig=

2

22

5002maxtanh

w

n

wb yy

kF (322)

33 Brief introduction of simulation process

All aspects of CFD study including modelling meshing solving and post-

processing of the results are briefly introduced in this chapter

331 Pre-processing technique

The speed and accuracy of simulation results depending fundamentally on meshing

and the ability to prepare model geometry correctly

The steps in pre-processing are explained as follows

1) Domain Definitions

The domain definition depends mainly on the physics involved in the problem and

objective of the problem

a) Geometry creation or geometry import

b) Geometry cleanup

i) Removing the parts not necessary for simulation under consideration

ii) Closing the gaps in the geometry

iii) Removing small surfaces curves to make meshing process simple

2)Grid generation

To solve the governing equations the numerical methods like Finite Difference

Finite Volume or Finite Element Method to divide the domain into small parts and

solve the equations on one part at a time are applied

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 5: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

ii

322 Wilcox k-ω turbulence model28

323 Shear stress transport model29

33 Brief introduction of simulation process31

331 Pre-processing technique31

332 Solving the simulation32

333 Post processing32

Chapter 4 Description of Model Cases33

41 Creating the geometry33

411 Geometry of the scaled sump model33

412 Geometry of the full size pump sump model35

42 Mesh generation39

43 Numerical approach42

44 Experimental setup45

Chapter 5 Results and Discussion47

51 Results of the scaled sump model47

511 Numerical simulation vortex check47

512 PIV results analysis49

513 Sump with anti free surface vortex devices53

52 Results of the mixed flow pump sump model56

521 Performance analysis of the mixed flow pump sump model56

522 Flow characteristics analysis58

523 Cavitation phenomenon analysis62

Chapter 6 Conclusions60

Acknowledgement67

References68

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

iii

List of Tables

Table 21 Acceptable velocity ranges for inlet bell diameter 6

Table 22 Recommended dimensions for a rectangular sump8

Table 23 Permission criteria of ANSIHI20

Table 41 Designed specifications of scaled sump model33

Table 42 Parameters of anti free surface vortex devices35

Table 43 Design specifications of the mixed flow pump38

Table 44 Mesh information for the pump sump model42

Table 51 Swirl angle of model test results54

Table 52 Summary of swirl angle CFD calculation results54

Table 53 Mixed flow pump performance results56

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

iv

List of Figures

Fig 21 Recommended inlet bell design diameter6

Fig 22 Recommended intake structure layout7

Fig 23 Filler wall details for proper bay width7

Fig 24 Typical vortex in a sump14

Fig 25 Vortex classifications15

Fig 26 Typical swirl meter17

Fig 27 Schematic view of swirl angle numerical calculation 18

Fig 28 Example of Pitot tube 19

Fig 29 Measuring point at bell throat19

Fig 210 Examples of approach flow patterns21

Fig 211 Methods to reduce submerged vortices23

Fig 41 Geometry of scaled sump model33

Fig 42 Schematic diagram of sump with anti free surface vortex devices34

Fig 43 Overall 3D drawing of the sump model35

Fig 44 2D drawing of the sump sketch36

Fig 45 Shape of the mixed flow pump37

Fig 46 Sump model with the submerged AVD38

Fig 47 Submerged AVD dimensions39

Fig 48 Overall mesh generation for scaled sump models40

Fig 49 Overall sump fluid domain mesh for the pump sump model41

Fig 410 Grid details of the pump sump model41

Fig 411 Boundaries of the scaled sump43

Fig 412 Experimental setup for the scaled sump model46

Fig 51 Streamline pattern of free surface47

Fig 52 Vortex core region of the sump model48

Fig 53 Velocity vectors on different sections48

Fig 54 Velocity vectors of PIV results51

Fig 55 Velocity vectors on different sections (PIV)52

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

v

Fig 56 Velocity vectors on free surface of different cases53

Fig 57 Tangential velocity profile of different models55

Fig 58 Performance curve of the mixed flow pump with the sump 57

Fig 59 Streamline pattern around the intake structure58

Fig 510 Vortex core region of the pump sump model59

Fig 511 Vorticity distribution in the flow direction60

Fig 512 Vorticity distribution in channel width direction61

Fig 513 Cavitation region of the mixed flow pump62

Fig 514 Cavitation performance curves of the mixed flow pump63

Fig 515 Vapor volume fraction distribution on blade surface64

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

vi

Study of Flow Field in Rectangular Sump Models and

Performance Analysis of a Mixed Flow Pump

Yuxin ZHAO

Department of Mechanical Engineering

Graduate School of Korea Maritime and Ocean University

Abstract

It is an accepted fact that quite a number of problems faced in a pumping station are

related to the design of sump or intake rather than pump design Head-capacity

curves provided by the pump manufacturer are obtained on the condition of no

vortices flowing into the pump intake The efficiency and performance of pumping

stations depend not only on the performance of the selected pumps but also on the

proper design of the intake sumps A faulty design of pump sump can lead to the

occurrence of swirl and vortices which reduce the pump performance and induce

vibration and additional noise Therefore sump model test is necessary to check the

flow condition around intake structure Numerical simulation is a good facility for

reducing the time and cost involved throughout the design process In this study the

commercial software ANSYS CFX-130 has been used for the CFD analysis of the

sump models

In a scaled sump model for air entrainment simulation numerical analysis of single

phase and two-phase with SST turbulence model was carried out to predict vortex

(both free surface vortex and submerged vortex) occurrence and location The

effectiveness of curtain walls and square bars to eliminate the free surface vortex is

evaluated Meanwhile the experiment including PIV system was performed in

KMOU to investigate the flow conditions around pump intake structure

In a full size sump model overall numerical analysis for the sump model with a

mixed flow pump installed was carried out Hydraulic performances of the mixed

flow pump for head rise shaft power pump efficiencies versus flow rate changed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

vii

from 50 to 140 of the design flow rate were studied by the performance curves

A trident shaped anti vortex device (AVD) composed of three wall fillets and one

center splitter was installed under the pump intake The effectiveness of the AVD

for the suppression of submerged vortex was evaluated In addition numerical

simulation of cavitation phenomenon in the mixed flow pump has been performed

by calculating the full cavitation model with k-ε turbulence model

According to the results details of the location size and strength of vortices were

predicted in the numerical simulation For the scaled sump model although there is

a little difference between the single phase and two-phase simulation all the results

predict the free surface vortex and submerged vortex formation and location The

CFD simulations of flow condition show good agreement with the PIV results The

effectiveness of curtain walls installed to inhibit the free surface vortex is confirmed

in the CFD results while the square bars installed in the channel failed to suppress

the free surface vortex For the mixed flow pump with sump the best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

The effectiveness of AVDs installed for the submerged vortex is confirmed by

comparisons of the sump model with and without AVD results

Cavitation phenomenon analysis shows the pump operating at the BEP condition

has a considerable extent over the requirement of cavitation performance With

numerical simulation the inception of cavitation in the blade passage is observed on

the suction surface where the leading edges meet the tips and then as the inlet total

pressure decrease the cavitation zone is spread out over the suction surface as well

as the leading edge of the impeller blade

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

viii

Nomenclature

B Distance from the back wall to the pump inlet bell centerline [m]

C Distance between the inlet bell and floor [m]

D Inlet bell design outside diameter [m]

Fr Froude number (=Vb(gD)05) [-]

g Acceleration due to gravity [ms2]

ht Total head [m]

n Number of revolution per min [-]

Pflow Flow power [W]

Pshaft Shaft power [W]

Pν Vapor pressure [Pa]

Q Volume flow rate [m3s]

Q0 Design volume flow rate [m3s]

Re Reynolds number (=VbDν) [-]

S Minimum submergence depth [m]

Vb Velocity at bell mouth [ms]

α Angle of floor slope [deg]

η Pump efficiency []

θ Swirl angle [deg]

λ The scale ratio of model dimensions to prototype [-]

μ Viscosity [Pa s]

ν Kinematic viscosity [m2s]

ρ Specific water density [kgm3]

σ Surface tension [Nm]

τ Stress tensor [-]

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

ix

Abbreviations

AVD Anti vortex device

BEP Best efficiency point

CFD Computational fluid Dynamics

NPSH Net positive suction head

NPSHR Required net positive suction head

PIV Particle Image Velocimetry

RANS Reynolds Averaged Navier- Stokes

RPM Number of revolutions per minute

SST Shear Stress Transport

VOF Volume of Fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

1

Chapter 1 Introduction

11 Background

It is an accepted fact that quite a number of problems faced in a pumping station are

related to the design of sump or intake rather than pump design As an important

part of the pumping station sump is designed to provide uniform swirl and

entrained air free flow to the pump intake Undesirable flow conditions (entrained

air non-uniform flow distribution vortices etc) can reduce the pump efficiency

induce vibration noise and cavitation even lead to excessive bearing loads of

impeller and structure damage Due to quite complex nature of the flow although

the flow conditions causing pump problems are well established there are no

specific solutions to eliminate them There are some design guidelines for the

specific geometrical and hydraulic constraints of the pump sump for any project

the model study is the only tool for solving potential problems in new designs and

rectifying problems observed in existing installations However the scaled model

tests are expensive and time-consuming so alternative Computational Fluid

Dynamics (CFD) methods for evaluating sump performance have been developed

With rapid progress in CFD numerical simulation is regarded as an effective tool in

solving fluid problems in pump sumps

12 Previous study

Various researches show the flow conditions around the sump intake structure are

quite complicated Shyam et al [1] used the commercial code CFX to carry out

two-phase flow simulation to capture air entrainment The air entrainments and its

location were well captured and numerical results were in accordance with the

experimental investigation Matsui J[2] and Okamura T[3] et al conducted a serial

research of the benchmark according to the Turbomachinery Society of Japan (TSJ)

standard including various commercial codes comparison performed in the flow

simulation and comparison analysis between the numerical and experimental results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

2

The purpose of the benchmark is to determine the accuracy and reliability of the

CFD codes used in universities and the industrial field It is also intended to identify

and investigate the formation of air-entrained and submerged vortices near the

pump Tanweer et al[4] have performed a numerical analysis of a multiple intake

pump sump to confirm that the geometry of sump affects the flow within sump

greatly Rajendran et al[5] have made a numerical analysis for the flow

characteristics of a sump model with pump intake and good agreements were

achieved by comparing the numerical results with the experiments Iwano et al[6]

have introduced a numerical method for the submerged vortex by analyzing the

flow in the pump sump with and without baffle plates Lee et al[7] have conducted

the CFD analysis of a multi-intake pump sump model to check the flow uniformity

by predicting the location number and vorticity of the vortex For the CFD

application in mixed flow pump design and performance analysis Oh H W [8]

conducted a mixed flow pump design optimization with specific parameters and

then compared the numerical results with the experiments The procedure presented

can be used as a practical design and analysis guide for the general turbomachinery

Kim et al[9] carried out a mixed flow pump performance analysis with numerical

simulation and improved the hydraulic efficiency by the modification of the

geometry

As an undesirable phenomenon caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage The cavitation

phenomenon has been studied by various investigators Van et al[10] have

investigated the cavitation inception behavior of a mixed-flow pump impellers

model and the calculated results matched well with the experiments Okamura et

al[11] have evaluated two numerical cavitation prediction methods used in the

pump industry manufacturers and attempts were made to improve the cavitation

performance Li et al[12] have conducted the numerical analysis on the basis of the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

3

development liquidvapor interface tracking method to predict the cavitation

characteristics with a centrifugal pump impeller model

Moreover there are a number of correlated studies [13]~[17]

13 Methodology of study

This study will investigate the flow phenomena of two sump models

A scaled sump model is adopted to carry out numerical simulation and PIV

experiment The flow conditions are predicted and the effectiveness of devices for

free surface vortex is evaluated by using numerical code a single intake sump

model with a mixed flow pump installed is modeled and the hydraulic performance

characteristics for the head rise shaft power and pump efficiencies versus flow rate

are studied In addition flow phenomena (vortices cavitation etc) are analyzed

and the effectiveness of an anti-vortex device (AVD) for the submerged vortex is

evaluated

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

4

Chapter 2 Basic theory of sump model

Specific hydraulic phenomena have been confirmed that they can lead to certain

common operational problems The common of the main problems are summarized

as follows

a Free surface vortices

b Submerged vortices

c Swirl of flow entering the pump

d Non-uniform distribution of velocity at the pump impeller

e Entrained air or gas bubbles

These problems encountered in the pump sump will affect the pump

performance and significantly increase the operational and maintenance costs In

order to identify sources of particular problems and find practical solution for

it the usual approach is to conduct the laboratory experiments on a scaled

physical model Typical design objectives are to ensure that a pump station is

designed according to best practice and conforms to the requirements set out in the

American National Standard for Pump Intake Design[18] These standards restrict

the degree to which the above-mentioned undesirable flow patterns may be present

The negative impact of each of these hydraulic conditions varies with pump specific

speed and size Application of the standard to design a sump does not generate a

problem free sump but provides a basis for the initial design

As there are no specific guidelines or criteria for design of trouble free intakes the

most common solution to potential problems in new design and rectification of

problems observed in existing designs is to construct a scaled model in a laboratory

observe and investigate the flow characteristics and propose modifications to the

intake geometry There are additional devices in the form of floor splitters or cones

back wall splitters fillets surface beams guide vanes etc aimed at controlling the

vortex and swirl formation Considerable prior experience and ingenuity are

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

5

required to solve the problems by this method and the cost and time required are

also significant

The intake structure refers to the channel leading from the water source which

may be a river or reservoir and all installations downstream including the pump

column or the intake pipe (the suction tube portion of a vertical intake) the

approach channel (upstream of pump bay) and the pump bay (bounded by the floor

the back wall side walls dividers walls separating adjacent pump column) Usually

the upstream end of the pump column has an inlet attachment called the suction bell

The function of the intake is to supply an evenly distributed flow of water to the

pump suction bell

Several types of pump intake basins are available and ANSIHI 98 provides

design requirements for each Intake structures can be categorized as being

clear liquids or solids-bearing liquids For clear liquids intakes are further

classified into rectangular formed circular and trench types as well as suction

tanks and cans For solid-bearing liquids trench type and rectangular wet wells are

usually considered

In this paper only rectangular intake structure for clear liquid is studied

21 Pump intake design

For pumps to achieve their optimum hydraulic performance across all operating

conditions the flow at the impeller must meet specific hydraulic conditions The

ideal flow entering the pump inlet should be of a uniform velocity distribution

without rotation and stable over time When designing a sump to achieve favorable

inflow to the pump or suction pipe bell various sump dimensions relative to the size

of the bell are required Using these standards of the distances to reduce the

probability of the occurence of strong submerged vortices and free-surface vortices

211 Inlet bell diameter design

The geometry is generally defined in terms of the pump inlet diameter as shown in

Table 21 and Fig 21 The bell diameter can be estimated based on an inlet pipe

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

6

Table 21 Acceptable velocity ranges for inlet bell diameter(ANSIHI 98)

Flow Rate

Qls

Recommended Inlet Bell Design

Velocityms

Acceptable Velocity

Range ms

Q<315 V=17 06leVle27

315 leQ<1260 V=17 09leVle24

Q ge1260 V=17 12leVle21

Fig 21 Recommended inlet bell design diameter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

7

212 Recommended dimensions for a rectangular sump

As the selected bell diameter has been determined the proportions of the inlet

structure can be estimated from Fig 22 and Fig 23 Table 22 shows the

recommended dimensions for a rectangular sump

Fig 22 Recommended intake structure layout

Fig 23 Filler wall details for proper bay width

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

8

Table 22 Recommended dimensions for a rectangular sump

Dimension Variable

Description Recommended value

A Distance from the pump inlet bell

centerline to the entrance A=5D minimum

a Length of constricted bay section near

the pump inlet 25D minimum

B Distance from the back wall to the

pump inlet bell centerline 075D

C Distance between the inlet bell and

floor 03D~05D

D Inlet bell outside diameter refer to section 211

H Minimum liquid depth H=S+C

h Minimum height of constricted bay

section near the pump inlet bell hgtH or 25D

S Minimum pump inlet bell submergence S=D(10+23Fr)

W Pump inlet bay entrance width 2D minimum

w Constricted bay width near the pump

inlet bell w=2D

X Pump inlet bay length 5D minimum

Y Distance from pump inlet bell

centerline to the through-flow traveling screen

4D minimum

Z1 Distance from sump inlet bell centerline to diverging walls

5D minimum

Z2 Distance from inlet bell centerline to

sloping floor 5D minimum

α Angle of floor slope -10deg~ +10deg

β Angle of wall convergence 0deg~ +10deg

empty Angle of convergence from constricted

area to by walls 10deg maximum

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

9

FD=Vb(gD)05 (21)

where

FD Froude number (dimensionless)

Vb velocity at suction inlet=FlowArea based on D

D Outside diameter of bell or pipe inlet

g gravitational acceleration

There is the design sequence summary for rectangular intake structures

middot Consider the flow patterns and boundary geometry of the body of liquid

from which the pump station is to receive flow Compare with the approach

flow condition and determine if a hydraulic physical model study is

required

middot Determine the number and size of pumps required to satisfy the range of

operating conditions likely to be encountered

middot Identify pump inlet bell diameter Refer to the section 211

middot Determine the bell-floor clearance 05D is a good preliminary design value

middot Determine the required bell submergence using the equation 21

middot Determine the minimum allowable liquid depth in the intake structure from

the sum of the floor clearance and the required bell submergence

middot Check bottom elevation near the entrance to the structure and determine if it

is necessary to slope the floor upstream of the bay entrance If the resulting

depth at the entrance to the intake structure is shallow then check to ensure

that gravity-driven flow is not restricted by the entrance condition

middot Check the pump bay velocity for the maximum single-pump flow and

minimum liquid depth with the bay width set to 2D If bay velocity exceeds

05 ms then increase the bay width to reduce the maximum velocity below

05 ms

middot If it is necessary to increase the pump bay width to greater than 2D then

decrease bay width in the vicinity of the pumps according to Fig 23

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

10

middot Compare cross-flow velocity (at maximum system flow) to average pump

bay velocity If cross-flow value exceeds 50 of the bay velocity a

physical hydraulic model study is necessary

middot Determine the length of the structure and dividing walls giving

consideration to minimum allowable distances to a sloping floor screening

equipment and length of dividing walls if dual flow traveling screens or

drum screens are to be used a physical hydraulic model study is required

middot If the final selected pump bell diameter and inlet velocity is within the

range given in section 211 then the sump dimensions (developed based on

the inlet bell design diameter) need not be changed and will comply with

these standards

22 Model test

The approach flow pattern in a pump intake facility is difficult to predict by

traditional mathematics or empirical formulae For large structures or those that

differ significantly from proven designs a model study is the only means to ensure

success A model study is used to identify adverse hydraulic conditions and

derive remedial measures for approach flow patterns generated by structures

upstream of the pump impeller Current model studies are not intended to

investigate flow patterns induced by the pump itself or the flow patterns within the

pump The objective of a model study is to ensure that the pump intake structure

generates favorable flow conditions at the inlet to the pump Evaluation for the

requirement of model test if

a Sump or piping geometry that deviates from this design standard

b Non-uniform or non-symmetric approach flow to the pump sump

exist

c The pumps have flows greater than 2520 ls per pump or the total

station flow with all pumps running would be greater than 6310 ls

d Situations that need ten times cost than the model study

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

11

Adverse hydraulic conditions that can affect pump performance include free and

submerged vortices swirl approaching the pump impeller flow separation at the

pump bell and a non-uniform axial velocity distribution at the suction

The purpose of model test is to examine these adverse hydraulic conditions and to

investigate measure against them

221 The scale effects

One possible difficulty in scale modeling is that it is impossible to reduce all

pertinent forces by the same factor With Froude-scale models the inertial and

gravitational forces are reduced similarly but the viscous and surface tension forces

cannot be simultaneously reduced The extra influence of these forces is known as

a ldquoscale effectrdquo A number of investigations have indicated that the scale effects for

the model are avoided if the Reynolds number based on the inlet flow and

submergence or intake diameter exceeds approximately 3x 104 and the Weber

number exceeds 150 [19]

In most engineering applications involving closed-conduit and open-channel flow

the Reynolds number limits are far exceeded and the flows are fully turbulent

In a sump model Froude number (Fr) and Reynolds number (Re) are the most

important non-dimensional parameters Some reduction in Re should be

introduced to the sump model However flow patterns are generally very similar

at high Re

No specific geometric scale rate is recommended but reasonable large

geometric scale to minimize viscous and surface tension scale effects and reproduce

the flow pattern

Re= uDν (22)

We= u2ρDσ (23)

where

u average axial velocity (ms)

ν kinematic viscosity of the liquid (m2s)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

12

ρ liquid density (kgm3)

σ surface tension of liquidair interface

According to the ANSIHI 98 the viscous scale effects on vortex may be negligible

if the resulting dimensionless numbers must meet these values Re>6times104 and

We>240

To allow practical visual observations of flow patterns accurate measurements of

swirl and velocity distribution and sufficient dimensional control the model scale

shall yield a bay width of at least 300mm a minimum liquid depth of at least

150mm and a pump throat or suction diameter of at least 80mm

222 Principle of similarity

In the similitude analysis the geometric and flow similarity requires the model and

the flow to be identical as the real model and flow This is to ensure the result

obtained from the model study is well presented in order to predict full scale

behavior Model involving a free surface are operated using Froude similarity since

the flow process is controlled by gravity and inertial forces

For similarity of the flow patterns the Froude number shall be equal in model and

prototype

Fr =

=

(24)

and as the scale ratio λ was selected in section 221 the following equation is

obtained

V = V times (

) = V times λ or Q = Q times (

) = Q times λ (25)

where Fr is the Froude number λ is the scale ratio of model dimensions to prototype

The test of 15 times Froude scaled flow should be conducted to ascertain the

potential scale affect on free surface vortices

Models of closed conduit piping system leading to a pump suction are not operated

based on Froude similitude However flow patterns are generally very similar at

high Re A minimum value of 1times105 for the Re at the pump suction is

recommended to ensure the flow patterns are correctly scaled

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

13

223 Preliminary model test

The initial design shall be tested first to identify any hydraulic problems

1) Vortex Formation Mechanisms

There are various investigations that provide insight into the fundamental processes

leading to the development of vortices both in experiment and numerical simulation

Shin et al [20] demonstrated that two basic mechanisms lead to inlet vortex

formation The first mechanism involves the development of an inlet vortex due to

the amplification of ambient vorticity in the approach flow as vortex lines

are convicted into the inlet The second mechanism involves the development of a

trailing vortex in the vicinity of the intake as a result of the variation in

circulation along the inlet For this second case a vortex can develop in a flow that

is irrotational upstream and the vortex development therefore does not depend

on the presence of ambient vorticity Shin et al investigation on kinematic

parameters indicate that the strength of an inlet-vortex or trailing vortex system

increases with decreasing distance from the surface However for an inlet in

an upstream irrotational flow two counter rotating vortices can still trail from

the rear of the inlet

2) Classification of vortex type

As illustrated in Fig 24 vortices in the vicinity of pump intakes may be adjacent to

the channel bottom or a channel wall (ie submerged vortices) or they may appear

adjacent to the free surface (ie free surface vortex)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

14

(a) Schematic diagram of vortices in a sump

(b) Experimental results of different vortex

Fig 24 Typical vortex in a sump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

15

For the model visual observation dye was injected around the intake structure near

the surface and submergence as Fig 24 (b) shows And the model test results of

vortex was classified as Fig 25

(a) Free Surface Vortices

(b) Submerged Vortices

Fig 25 Vortex classifications

There are a number of researches focused on the numerical theory to detect the

inception of a visible vortex As the diameters of vortices are usually much smaller

compared with the computing grid size to compensate for the lack of vortex

resolution in numerical simulation a stretching vortex model is applied to the local

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

16

flow field around semi-analytically identified vortex positions [21][22] The brief

explanations are as follows

a) Free surface vortex

The criteria expressed by equation 26 and 27 is applied to determine the

visible inception of a free surface vortex

≫ 1 (26)

=

gt 1 (27)

where P is the pressure drop at the vortex core and Ph is the static head at the

vortex element and f is the drag force acting at a unit length of water surface

vortex dimple based on Stokes approximation and caused by the down flow

Further f is the force acting on the bubble in the direction from the high

pressure region to the low pressure region like buoyancy

b) Submerged vortex

Equation 28 is applied to determine the visible inception of a submerged vortex

infin =

infin

infin gt 1 (28)

where is the static pressure at the vortex core infin is the atmospheric pressure

and is the critical cavitation inception pressure and nearly equal to the

saturated vapor pressure

3) Swirl angle

Swirl angle predicts the intensity of flow rotation Fig 26 shows experiment

calculation method with swirl meter and attempts of the CFD calculation method

were conducted as Fig 27 A four blade zero-pitch swirl meter which is supported

by a low friction bearing will be installed at about four suction pipe diameter (d)

downstream from the pump suction to measure swirl angle of pump approaching

flow For the swirl meter the tip to tip blade diameter is 075d and the length in

flow direction is 06 and one of the four blade is painted yellow as a reference to

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

17

count revolutions As flow swirl is usually unsteady the observation time of swirl

meter reading should be a continuous period of time

Fig 26 Typical swirl meter

(29)

(210)

(211)

Where qV flow rotational speed at swirl meter (ms)

zV average axial velocity at swirl meter (ms)

n Number of revolution per minute (rpm)

d Diameter of pipe at the swirl meter (m)

q flow rate (m3s)

zV

Vqq 1tan -=

)(60

smnd

Vtimestimes

=p

q

)(4

2

smd

qVz

times=

p

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

18

For the swirl angle calculating method with CFD the key point is to obtain the

average tangential velocity so the swirl check circles were created as illustrated in

Fig 27 The swirl check circle is located at the section of 4d height with diameter

of 025d 05d and 075d respectively

Swirl check circle

Fig 27 Schematic view of swirl angle numerical calculation

23

1 7501 501 250

acute++

=aringaringaring

N

i

N

i

NNVNVNV

Vi

q (212)

where

qV average tangential velocity (ms)

V circumferential velocity of check point in check circle

respectively (ms)

N check point total numbers in single check circle

4) Velocity

Velocity measurement at bell throat by Pitot tube was implemented by ANSIHI

method Typical Pitot tube configuration is shown in Fig 28 Eight measuring

points are defined as Fig 29 Time averaged velocity at eight points was obtained

and these eight data were again averaged to give throat sectional spatial velocity

average

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

19

Fig 28 Example of Pitot tube Fig 29 Measuring points at bell throat

224 Test criteria

There is the criteria for satisfactory performance of the sump model test and

summarized in Table 23

1) Free surface and submerged vortices

Free surface vortices entering the pump must be less severe than vortices with

coherent core Dye core vortices may be acceptable only if they occur less than 10

of the time or only for infrequent pump operating conditions

Submerged vortex which starts from the sump walls induce vibration and noise

problems Dye core submerged vortices extending into the pump bell mouth is not

allowed

2) Swirl angles

Swirl angle is used to check the flow rotation in the suction pipe The calculation

method is referred to section 223

Short-term maximum and long-term average values must be less than 5 degrees

Maximum short-term swirl angles up to 7 degrees may be acceptable only if they

occur less than 10 of the time or only for infrequent pump operating conditions

3)Velocity

Each point time-averaged velocity shall not be larger than 10 of sectional average

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

20

Standard deviation of time-varying velocity at each point was obtained and its

value shall not be larger than 10 of the time-averaged velocity at each point

1) Time-averaged

Variation[] = 100acute-

a

a

V

VV aV Averaged at section

2) Time-varying

Standard deviation= )1(

)(1i

2

-

-aring=

n

xxn

i

Variation[] = (Standard deviation X ) x 100

X Averaged at point

Table 23 Permission criteria of ANSIHI

Vortex occurence Swirl angle

Free surface vortex Submerged vortex

Observation

time Above 10 minutes Above 10 minutes

Above 10

minutes

Permission

criteria

Dye-core

Vortex(F3)

below 110 total

time

Dye-core

Vortex(S2)

below 110 total

time

Below

5degConditional

intermittent

below 7deg

225 Remedial measures for problem intakes

Usually in the preliminary test the worst flow condition is simulated And based on

the results remedial measures for problem intakes are discussed in this part

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

21

1) Overall approach flow patterns

The characteristics of flow approaching an intake structure is one of the foremost

considerations for the designer It is difficult to predict the effects of a given set of

flow conditions upstream from an intake structure Fig 210 depicts several typical

approach flow patterns of pump sump with different operating conditions

Fig 210 Examples of approach flow patterns

Generally speaking for the intake problems especially for the multiple pumps

sump there are the main measures using dividing walls placed between the pumps

as a partitioned structure trash racks with elongated bars guide piers or a number

of structural columns at the bay entrance can provide some assistance in controlling

cross-flow baffles flow distributors and screen mesh etc are applied to improve

the flow distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

22

2) Problems around pump intake

a) Free surface vortices

There are a number of factors that have an effect on the formation of surface

vortices To achieve a higher degree of certainty that objectionable surface vortices

do not form modifications can be made to intake structure to allow operation at

practical depths of submergence and the minimum submergence is referred in Table

22 the use of suction umbrellas vertical curtain walls and horizontal gratings

b) Submerged vortices

In the vicinity of intake bell there are the most complicated flow patterns and flow

makes the most changes in direction while maintaining a constant acceleration into

the pump bell to prevent local flow separation turbulence and vortex formation

Fig 211 represents a sample of various devices to inhibit the different submerged

vortex

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

23

Fig 211 Methods to reduce submerged vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

24

c) Preswirl

The approach flow distributions determine whether flow preswirl exists and a

sufficiently laterally skewed approach flow cause rotating around the intake bell

The most effective way of reducing preswirl is to establish a relatively uniform

approach flow within each pump bay by using the baffling schemes controlling

cross-flow and expanding concentrated flow discussed in the above part

To prevent non-uniform velocity distribution swirls and vortices necessary

modifications are conducted such as fore bay length baffle wall distributors

curtain wall mesh screen guide vane floor splitter wall fillet etc As the above

measures applied modification test will be conducted to check all the flow

conditions and compare with the preliminary results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

25

Chapter 3 Methodology

CFD (Computational fluid dynamics) is the technique which can design cycles to

better performance and reduce costs and time In CFD we solve the governing

equations of given physics (may be differential form or integral form) using some

numerical techniques like Finite Difference Method (FDM) Finite Element Method

(FEM) or Finite Volume Method (FVM) Although CFD plays a great part in the

design of turbomachinery CFD is not a replacement for experimental or analytical

approach In this study the numerical analysis was performed by the commercial

code ANSYS CFX 130 All aspects of the CFD study including modelling

meshing solving and post-processing of the results

31 Numerical modeling

ANSYS CFX solves the unsteady Navier-Stokes equations in their conservation

form The instantaneous equation of mass (continuity) in the stationary frame is

expressed as equation

ρ

δ + nabla ∙ (ρU) = 0 (31)

And the instantaneous equation for momentum is expressed as shown in equation

ρ

δ + nabla ∙ (ρU⨂U) = minusnablap + nabla ∙ τ + S (32)

Where τ is the stress tensor ( including both normal and shear components of the

stress)

These instantaneous equations are averaged for turbulent flows leading to additional

terms that need to be solved While the Navier-Stokes equations describe both

laminar and turbulent flows without addition terms realistic flows involve length

scales much smaller than the smallest finite volume mesh A Direct Numerical

Simulation of these flows would require significantly more computing power than

what is available now or in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

26

Therefore much research has been done to predict the effects of turbulence by

using turbulence models These models account for the effects of turbulence

without the use of a very fine mesh or direct numerical simulation

These turbulence models modify the transport equations by adding averaged and

fluctuating components The transport equations are changed to equations (33) and

(34)

ρ

δ + nabla ∙ (ρU) = 0 (33)

ρ

+ nabla ∙ ρ U ⨂U = minusnablap + nabla ∙ τ minus ρu⨂u + S (34)

The mass equation is not changed but the momentum equation contains extra terms

which are the Reynolds stresses ρ u⨂u and the Reynolds flux ρ u ϕ These

Reynolds stresses need to be modeled by additional equations to obtain closure

Obtaining closure implies that there are a sufficient number of equations to solve for

all the unknowns including the Reynolds stresses and Reynolds fluxes

Various turbulence models provide various ways to obtain closure In this

investigation Shear Stress Transport (SST) model and k-ε model were utilized The

advantage of SST model is that combines the advantages of other turbulence models

(the k-ε Wilcox k-ω) The 3 turbulence models will be discussed briefly

32 Turbulence models

Turbulence models are necessary because we cannot afford big enough computers

to directly capture every scale of motion There are three modelling frameworks

Direct Numerical Simulation (DNS) Reynolds-Averaged Navier-Stokes

Equations(RANS) Large Eddy Simulation(LES)

As RANS method is applied in most productionseveral RANS turbulence models

differ in their usage of wall functions the number of additional variables solved for

and what these variables represent

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

27

Two equation turbulence models

Two-equation turbulence models are widely used in CFD as they give a good

compromise between computational power needed and accuracy The term lsquotwo

equationrsquo refers to the fact that these models solve for the velocity and length scales

using separate transport equations The turbulence velocity scale is obtained by

solving the transport equation The turbulent length scale is estimated from two

properties of the turbulence field namely the turbulent kinetic energy and the

dissipation rate The dissipation rate of the kinetic energy is obtained from its

transport equation The most widely used are k-ε and k-ω two equation models In

the next section the k-ε Wilcox k-ω BSL k-ω and SST models will be briefly

discussed

321 k-ε turbulence model

The k-epsilon model solves for two variables k the turbulent kinetic energy and

epsilon the rate of dissipation of kinetic energy Wall functions are used in this

model so the flow in the buffer region is not simulated The k-epsilon model is very

popular for industrial applications due to its good convergence rate and relatively

low memory requirements It does not very accurately compute flow fields that

exhibit adverse pressure gradients strong curvature to the flow or jet flow It does

perform well for external flow problems around complex geometries

The k-ε model introduces k (m2s2) as the turbulence kinetic energy and ε (m2s3) as

the turbulence eddy dissipation The continuity equation remains the same

ρ

+ nabla ∙ (ρ U) = 0 (35)

The momentum equation changes as shown by equation

δρ

δ + nabla ∙ (ρU⨂U) = minusnablapprime + nabla ∙ μ

nablaU + (nablaU) + S (36)

Where SM is the sum of body forces μeff is the effective viscosity accounting for

turbulence and prsquo is the modified pressure The k-ε model uses the concept of eddy

viscosity giving the equation for effective viscosity as shown by equation

μ

= μ + μ (37)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

28

μt is the turbulence viscosity is linked to the turbulence kinetic energy and

dissipation by the equation

μ = Cμρ

ε (38)

Where Cμ is a constant

The values for k and ε come from the differential transport equations for the

turbulence kinetic energy and the turbulence dissipation rate

The turbulence kinetic energy equation is given as equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus ρε (39)

The turbulence dissipation rate is given by equation

(ρε)

+ nabla ∙ (ρUε) = nabla ∙ μ +

μ

σε nablaε +

ε

(Cε (P + Pε ) minus Cε ρε) (310)

Where Cε1 Cε2 σk σε are constants

Pk is the turbulence production due to viscous forces and is modeled by the

equation

P = μ nablaU ∙ nablaU + nablaU minus

nabla ∙ U 3μ

nabla ∙ U + ρk (311)

If a buoyancy term is added to the previous equation if the full buoyancy model is

used

The buoyancy term Pkb is modeled as

P = minusμ

ρσρg ∙ nablaρ (312)

322 Wilcox k-ω turbulence model

The k-omega model is similar to k-epsilon instead however it solves for omega mdash

the specific rate of dissipation of kinetic energy It also uses wall functions and

therefore has comparable memory requirements It has more difficulty converging

and is quite sensitive to the initial guess at the solution Hence the k-epsilon model

is often used first to find an initial condition for solving the k-omega model The k-

omega model is useful in many cases where the k-epsilon model is not accurate

such as internal flows flows that exhibit strong curvature separated flows and jets

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

29

This model has an advantage over the k-ε model where it does not involve complex

linear damping functions for near wall calculations at low Reynolds numbers

The k-ω model assumes that the turbulence viscosity is related to the turbulence

kinetic energy k and the turbulent frequency ω by the equation

μ = ρ

ω (313)

The transport equation for k is given by the equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus βprimeρ k ω (314)

The transport equation for ω is shown as equation

(ρω)

δ + nabla ∙ (ρ U ω) = nabla ∙ μ +

μ

μω

nablaω + αω

P + Pω minus β ρ ω (315)

The production rate of turbulence (Pk) is calculated as shown previously in the k-ε

section

The recommended values for the constants in the above equations are

βprime = 009

α = 59

β = 0075

σ = 2

σω = 2

The Reynolds stress tensor ρ u⨂u is calculated by

minusρ u⨂u = μ nablaU + (nablaU) minus

part ρk + μ

nabla ∙ U (316)

323 Shear stress transport model

The disadvantage of the Wilcox model is the strong sensitivity to free-stream

conditions Therefore a blending of the k-ω model near the surface and the k-ε in

the outer region was made by Menter [23] which resulted in the formulation of the

BSL k-ω turbulence model It consists of a transformation of the k-ε model to a k-ω

formulation and subsequently adding the resulting equations The Wilcox model is

multiplied by a blending function F1 and the transformed k-ε by another function 1-

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

30

F1 F1 is a function of wall distance (being the value of one near the surface and zero

outside the boundary layer) Outside the boundary and on the edge of the boundary

layer the standard k-ε model is used [24]

SST model is a combination of the k-epsilon in the free stream and the k-omega

models near the walls It does not use wall functions and tends to be most accurate

when solving the flow near the wall The SST model does not always converge to

the solution quickly so the k-epsilon or k-omega models are often solved first to

give good initial conditions

The k-ω based Shear Stress Transport (SST) model of Menter was applied for

turbulence treatment The transport equations for the SST model are expressed

below where the turbulent kinetic energy lsquokrsquo and turbulent frequency or dissipation

per unit turbulent kinetic energy are computed by using the following relations

For Turbulence Kinetic Energy

kkPukt

kTkk Ntilde+Ntilde+-=Ntilde+

para

para)[()( nsnwb (317)

where Pk is the production limiter

For Specific Dissipation Rate

ww

swnsnbwaww

ww NtildeNtilde-+Ntilde+Ntilde+-=Ntilde+para

para

1)1(2])[()( 21

22 kFSut

T (318)

The first blending function F1 is calculated from

iumlthorn

iumlyacute

uuml

iumlicirc

iumliacute

igrave

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

uacuteuacuteucirc

ugrave

ecircecirceuml

eacute

dividedivideoslash

oumlccedilccedilegrave

aelig=

4

22

21

4500maxmintanh

yCD

k

yy

kF

kw

wrs

w

n

wb (319)

y is the distance to the nearest wall and v is the kinematic viscosity

divideoslash

oumlccedilegrave

aeligNtildeNtilde= -10

2 101

2max ww

rsww kCDk (320)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

31

And Kinematic eddy viscosity

)max( 21

1

SF

kT

wa

an = (321)

where F2 is a blending function which restricts the limiter to the wall boundary and

S is the invariant measure of the strain rate

uacuteuacute

ucirc

ugrave

ecircecirc

euml

eacute

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

dividedivideoslash

oumlccedilccedilegrave

aelig=

2

22

5002maxtanh

w

n

wb yy

kF (322)

33 Brief introduction of simulation process

All aspects of CFD study including modelling meshing solving and post-

processing of the results are briefly introduced in this chapter

331 Pre-processing technique

The speed and accuracy of simulation results depending fundamentally on meshing

and the ability to prepare model geometry correctly

The steps in pre-processing are explained as follows

1) Domain Definitions

The domain definition depends mainly on the physics involved in the problem and

objective of the problem

a) Geometry creation or geometry import

b) Geometry cleanup

i) Removing the parts not necessary for simulation under consideration

ii) Closing the gaps in the geometry

iii) Removing small surfaces curves to make meshing process simple

2)Grid generation

To solve the governing equations the numerical methods like Finite Difference

Finite Volume or Finite Element Method to divide the domain into small parts and

solve the equations on one part at a time are applied

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 6: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

iii

List of Tables

Table 21 Acceptable velocity ranges for inlet bell diameter 6

Table 22 Recommended dimensions for a rectangular sump8

Table 23 Permission criteria of ANSIHI20

Table 41 Designed specifications of scaled sump model33

Table 42 Parameters of anti free surface vortex devices35

Table 43 Design specifications of the mixed flow pump38

Table 44 Mesh information for the pump sump model42

Table 51 Swirl angle of model test results54

Table 52 Summary of swirl angle CFD calculation results54

Table 53 Mixed flow pump performance results56

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

iv

List of Figures

Fig 21 Recommended inlet bell design diameter6

Fig 22 Recommended intake structure layout7

Fig 23 Filler wall details for proper bay width7

Fig 24 Typical vortex in a sump14

Fig 25 Vortex classifications15

Fig 26 Typical swirl meter17

Fig 27 Schematic view of swirl angle numerical calculation 18

Fig 28 Example of Pitot tube 19

Fig 29 Measuring point at bell throat19

Fig 210 Examples of approach flow patterns21

Fig 211 Methods to reduce submerged vortices23

Fig 41 Geometry of scaled sump model33

Fig 42 Schematic diagram of sump with anti free surface vortex devices34

Fig 43 Overall 3D drawing of the sump model35

Fig 44 2D drawing of the sump sketch36

Fig 45 Shape of the mixed flow pump37

Fig 46 Sump model with the submerged AVD38

Fig 47 Submerged AVD dimensions39

Fig 48 Overall mesh generation for scaled sump models40

Fig 49 Overall sump fluid domain mesh for the pump sump model41

Fig 410 Grid details of the pump sump model41

Fig 411 Boundaries of the scaled sump43

Fig 412 Experimental setup for the scaled sump model46

Fig 51 Streamline pattern of free surface47

Fig 52 Vortex core region of the sump model48

Fig 53 Velocity vectors on different sections48

Fig 54 Velocity vectors of PIV results51

Fig 55 Velocity vectors on different sections (PIV)52

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

v

Fig 56 Velocity vectors on free surface of different cases53

Fig 57 Tangential velocity profile of different models55

Fig 58 Performance curve of the mixed flow pump with the sump 57

Fig 59 Streamline pattern around the intake structure58

Fig 510 Vortex core region of the pump sump model59

Fig 511 Vorticity distribution in the flow direction60

Fig 512 Vorticity distribution in channel width direction61

Fig 513 Cavitation region of the mixed flow pump62

Fig 514 Cavitation performance curves of the mixed flow pump63

Fig 515 Vapor volume fraction distribution on blade surface64

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

vi

Study of Flow Field in Rectangular Sump Models and

Performance Analysis of a Mixed Flow Pump

Yuxin ZHAO

Department of Mechanical Engineering

Graduate School of Korea Maritime and Ocean University

Abstract

It is an accepted fact that quite a number of problems faced in a pumping station are

related to the design of sump or intake rather than pump design Head-capacity

curves provided by the pump manufacturer are obtained on the condition of no

vortices flowing into the pump intake The efficiency and performance of pumping

stations depend not only on the performance of the selected pumps but also on the

proper design of the intake sumps A faulty design of pump sump can lead to the

occurrence of swirl and vortices which reduce the pump performance and induce

vibration and additional noise Therefore sump model test is necessary to check the

flow condition around intake structure Numerical simulation is a good facility for

reducing the time and cost involved throughout the design process In this study the

commercial software ANSYS CFX-130 has been used for the CFD analysis of the

sump models

In a scaled sump model for air entrainment simulation numerical analysis of single

phase and two-phase with SST turbulence model was carried out to predict vortex

(both free surface vortex and submerged vortex) occurrence and location The

effectiveness of curtain walls and square bars to eliminate the free surface vortex is

evaluated Meanwhile the experiment including PIV system was performed in

KMOU to investigate the flow conditions around pump intake structure

In a full size sump model overall numerical analysis for the sump model with a

mixed flow pump installed was carried out Hydraulic performances of the mixed

flow pump for head rise shaft power pump efficiencies versus flow rate changed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

vii

from 50 to 140 of the design flow rate were studied by the performance curves

A trident shaped anti vortex device (AVD) composed of three wall fillets and one

center splitter was installed under the pump intake The effectiveness of the AVD

for the suppression of submerged vortex was evaluated In addition numerical

simulation of cavitation phenomenon in the mixed flow pump has been performed

by calculating the full cavitation model with k-ε turbulence model

According to the results details of the location size and strength of vortices were

predicted in the numerical simulation For the scaled sump model although there is

a little difference between the single phase and two-phase simulation all the results

predict the free surface vortex and submerged vortex formation and location The

CFD simulations of flow condition show good agreement with the PIV results The

effectiveness of curtain walls installed to inhibit the free surface vortex is confirmed

in the CFD results while the square bars installed in the channel failed to suppress

the free surface vortex For the mixed flow pump with sump the best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

The effectiveness of AVDs installed for the submerged vortex is confirmed by

comparisons of the sump model with and without AVD results

Cavitation phenomenon analysis shows the pump operating at the BEP condition

has a considerable extent over the requirement of cavitation performance With

numerical simulation the inception of cavitation in the blade passage is observed on

the suction surface where the leading edges meet the tips and then as the inlet total

pressure decrease the cavitation zone is spread out over the suction surface as well

as the leading edge of the impeller blade

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

viii

Nomenclature

B Distance from the back wall to the pump inlet bell centerline [m]

C Distance between the inlet bell and floor [m]

D Inlet bell design outside diameter [m]

Fr Froude number (=Vb(gD)05) [-]

g Acceleration due to gravity [ms2]

ht Total head [m]

n Number of revolution per min [-]

Pflow Flow power [W]

Pshaft Shaft power [W]

Pν Vapor pressure [Pa]

Q Volume flow rate [m3s]

Q0 Design volume flow rate [m3s]

Re Reynolds number (=VbDν) [-]

S Minimum submergence depth [m]

Vb Velocity at bell mouth [ms]

α Angle of floor slope [deg]

η Pump efficiency []

θ Swirl angle [deg]

λ The scale ratio of model dimensions to prototype [-]

μ Viscosity [Pa s]

ν Kinematic viscosity [m2s]

ρ Specific water density [kgm3]

σ Surface tension [Nm]

τ Stress tensor [-]

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

ix

Abbreviations

AVD Anti vortex device

BEP Best efficiency point

CFD Computational fluid Dynamics

NPSH Net positive suction head

NPSHR Required net positive suction head

PIV Particle Image Velocimetry

RANS Reynolds Averaged Navier- Stokes

RPM Number of revolutions per minute

SST Shear Stress Transport

VOF Volume of Fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

1

Chapter 1 Introduction

11 Background

It is an accepted fact that quite a number of problems faced in a pumping station are

related to the design of sump or intake rather than pump design As an important

part of the pumping station sump is designed to provide uniform swirl and

entrained air free flow to the pump intake Undesirable flow conditions (entrained

air non-uniform flow distribution vortices etc) can reduce the pump efficiency

induce vibration noise and cavitation even lead to excessive bearing loads of

impeller and structure damage Due to quite complex nature of the flow although

the flow conditions causing pump problems are well established there are no

specific solutions to eliminate them There are some design guidelines for the

specific geometrical and hydraulic constraints of the pump sump for any project

the model study is the only tool for solving potential problems in new designs and

rectifying problems observed in existing installations However the scaled model

tests are expensive and time-consuming so alternative Computational Fluid

Dynamics (CFD) methods for evaluating sump performance have been developed

With rapid progress in CFD numerical simulation is regarded as an effective tool in

solving fluid problems in pump sumps

12 Previous study

Various researches show the flow conditions around the sump intake structure are

quite complicated Shyam et al [1] used the commercial code CFX to carry out

two-phase flow simulation to capture air entrainment The air entrainments and its

location were well captured and numerical results were in accordance with the

experimental investigation Matsui J[2] and Okamura T[3] et al conducted a serial

research of the benchmark according to the Turbomachinery Society of Japan (TSJ)

standard including various commercial codes comparison performed in the flow

simulation and comparison analysis between the numerical and experimental results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

2

The purpose of the benchmark is to determine the accuracy and reliability of the

CFD codes used in universities and the industrial field It is also intended to identify

and investigate the formation of air-entrained and submerged vortices near the

pump Tanweer et al[4] have performed a numerical analysis of a multiple intake

pump sump to confirm that the geometry of sump affects the flow within sump

greatly Rajendran et al[5] have made a numerical analysis for the flow

characteristics of a sump model with pump intake and good agreements were

achieved by comparing the numerical results with the experiments Iwano et al[6]

have introduced a numerical method for the submerged vortex by analyzing the

flow in the pump sump with and without baffle plates Lee et al[7] have conducted

the CFD analysis of a multi-intake pump sump model to check the flow uniformity

by predicting the location number and vorticity of the vortex For the CFD

application in mixed flow pump design and performance analysis Oh H W [8]

conducted a mixed flow pump design optimization with specific parameters and

then compared the numerical results with the experiments The procedure presented

can be used as a practical design and analysis guide for the general turbomachinery

Kim et al[9] carried out a mixed flow pump performance analysis with numerical

simulation and improved the hydraulic efficiency by the modification of the

geometry

As an undesirable phenomenon caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage The cavitation

phenomenon has been studied by various investigators Van et al[10] have

investigated the cavitation inception behavior of a mixed-flow pump impellers

model and the calculated results matched well with the experiments Okamura et

al[11] have evaluated two numerical cavitation prediction methods used in the

pump industry manufacturers and attempts were made to improve the cavitation

performance Li et al[12] have conducted the numerical analysis on the basis of the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

3

development liquidvapor interface tracking method to predict the cavitation

characteristics with a centrifugal pump impeller model

Moreover there are a number of correlated studies [13]~[17]

13 Methodology of study

This study will investigate the flow phenomena of two sump models

A scaled sump model is adopted to carry out numerical simulation and PIV

experiment The flow conditions are predicted and the effectiveness of devices for

free surface vortex is evaluated by using numerical code a single intake sump

model with a mixed flow pump installed is modeled and the hydraulic performance

characteristics for the head rise shaft power and pump efficiencies versus flow rate

are studied In addition flow phenomena (vortices cavitation etc) are analyzed

and the effectiveness of an anti-vortex device (AVD) for the submerged vortex is

evaluated

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

4

Chapter 2 Basic theory of sump model

Specific hydraulic phenomena have been confirmed that they can lead to certain

common operational problems The common of the main problems are summarized

as follows

a Free surface vortices

b Submerged vortices

c Swirl of flow entering the pump

d Non-uniform distribution of velocity at the pump impeller

e Entrained air or gas bubbles

These problems encountered in the pump sump will affect the pump

performance and significantly increase the operational and maintenance costs In

order to identify sources of particular problems and find practical solution for

it the usual approach is to conduct the laboratory experiments on a scaled

physical model Typical design objectives are to ensure that a pump station is

designed according to best practice and conforms to the requirements set out in the

American National Standard for Pump Intake Design[18] These standards restrict

the degree to which the above-mentioned undesirable flow patterns may be present

The negative impact of each of these hydraulic conditions varies with pump specific

speed and size Application of the standard to design a sump does not generate a

problem free sump but provides a basis for the initial design

As there are no specific guidelines or criteria for design of trouble free intakes the

most common solution to potential problems in new design and rectification of

problems observed in existing designs is to construct a scaled model in a laboratory

observe and investigate the flow characteristics and propose modifications to the

intake geometry There are additional devices in the form of floor splitters or cones

back wall splitters fillets surface beams guide vanes etc aimed at controlling the

vortex and swirl formation Considerable prior experience and ingenuity are

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

5

required to solve the problems by this method and the cost and time required are

also significant

The intake structure refers to the channel leading from the water source which

may be a river or reservoir and all installations downstream including the pump

column or the intake pipe (the suction tube portion of a vertical intake) the

approach channel (upstream of pump bay) and the pump bay (bounded by the floor

the back wall side walls dividers walls separating adjacent pump column) Usually

the upstream end of the pump column has an inlet attachment called the suction bell

The function of the intake is to supply an evenly distributed flow of water to the

pump suction bell

Several types of pump intake basins are available and ANSIHI 98 provides

design requirements for each Intake structures can be categorized as being

clear liquids or solids-bearing liquids For clear liquids intakes are further

classified into rectangular formed circular and trench types as well as suction

tanks and cans For solid-bearing liquids trench type and rectangular wet wells are

usually considered

In this paper only rectangular intake structure for clear liquid is studied

21 Pump intake design

For pumps to achieve their optimum hydraulic performance across all operating

conditions the flow at the impeller must meet specific hydraulic conditions The

ideal flow entering the pump inlet should be of a uniform velocity distribution

without rotation and stable over time When designing a sump to achieve favorable

inflow to the pump or suction pipe bell various sump dimensions relative to the size

of the bell are required Using these standards of the distances to reduce the

probability of the occurence of strong submerged vortices and free-surface vortices

211 Inlet bell diameter design

The geometry is generally defined in terms of the pump inlet diameter as shown in

Table 21 and Fig 21 The bell diameter can be estimated based on an inlet pipe

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

6

Table 21 Acceptable velocity ranges for inlet bell diameter(ANSIHI 98)

Flow Rate

Qls

Recommended Inlet Bell Design

Velocityms

Acceptable Velocity

Range ms

Q<315 V=17 06leVle27

315 leQ<1260 V=17 09leVle24

Q ge1260 V=17 12leVle21

Fig 21 Recommended inlet bell design diameter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

7

212 Recommended dimensions for a rectangular sump

As the selected bell diameter has been determined the proportions of the inlet

structure can be estimated from Fig 22 and Fig 23 Table 22 shows the

recommended dimensions for a rectangular sump

Fig 22 Recommended intake structure layout

Fig 23 Filler wall details for proper bay width

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

8

Table 22 Recommended dimensions for a rectangular sump

Dimension Variable

Description Recommended value

A Distance from the pump inlet bell

centerline to the entrance A=5D minimum

a Length of constricted bay section near

the pump inlet 25D minimum

B Distance from the back wall to the

pump inlet bell centerline 075D

C Distance between the inlet bell and

floor 03D~05D

D Inlet bell outside diameter refer to section 211

H Minimum liquid depth H=S+C

h Minimum height of constricted bay

section near the pump inlet bell hgtH or 25D

S Minimum pump inlet bell submergence S=D(10+23Fr)

W Pump inlet bay entrance width 2D minimum

w Constricted bay width near the pump

inlet bell w=2D

X Pump inlet bay length 5D minimum

Y Distance from pump inlet bell

centerline to the through-flow traveling screen

4D minimum

Z1 Distance from sump inlet bell centerline to diverging walls

5D minimum

Z2 Distance from inlet bell centerline to

sloping floor 5D minimum

α Angle of floor slope -10deg~ +10deg

β Angle of wall convergence 0deg~ +10deg

empty Angle of convergence from constricted

area to by walls 10deg maximum

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

9

FD=Vb(gD)05 (21)

where

FD Froude number (dimensionless)

Vb velocity at suction inlet=FlowArea based on D

D Outside diameter of bell or pipe inlet

g gravitational acceleration

There is the design sequence summary for rectangular intake structures

middot Consider the flow patterns and boundary geometry of the body of liquid

from which the pump station is to receive flow Compare with the approach

flow condition and determine if a hydraulic physical model study is

required

middot Determine the number and size of pumps required to satisfy the range of

operating conditions likely to be encountered

middot Identify pump inlet bell diameter Refer to the section 211

middot Determine the bell-floor clearance 05D is a good preliminary design value

middot Determine the required bell submergence using the equation 21

middot Determine the minimum allowable liquid depth in the intake structure from

the sum of the floor clearance and the required bell submergence

middot Check bottom elevation near the entrance to the structure and determine if it

is necessary to slope the floor upstream of the bay entrance If the resulting

depth at the entrance to the intake structure is shallow then check to ensure

that gravity-driven flow is not restricted by the entrance condition

middot Check the pump bay velocity for the maximum single-pump flow and

minimum liquid depth with the bay width set to 2D If bay velocity exceeds

05 ms then increase the bay width to reduce the maximum velocity below

05 ms

middot If it is necessary to increase the pump bay width to greater than 2D then

decrease bay width in the vicinity of the pumps according to Fig 23

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

10

middot Compare cross-flow velocity (at maximum system flow) to average pump

bay velocity If cross-flow value exceeds 50 of the bay velocity a

physical hydraulic model study is necessary

middot Determine the length of the structure and dividing walls giving

consideration to minimum allowable distances to a sloping floor screening

equipment and length of dividing walls if dual flow traveling screens or

drum screens are to be used a physical hydraulic model study is required

middot If the final selected pump bell diameter and inlet velocity is within the

range given in section 211 then the sump dimensions (developed based on

the inlet bell design diameter) need not be changed and will comply with

these standards

22 Model test

The approach flow pattern in a pump intake facility is difficult to predict by

traditional mathematics or empirical formulae For large structures or those that

differ significantly from proven designs a model study is the only means to ensure

success A model study is used to identify adverse hydraulic conditions and

derive remedial measures for approach flow patterns generated by structures

upstream of the pump impeller Current model studies are not intended to

investigate flow patterns induced by the pump itself or the flow patterns within the

pump The objective of a model study is to ensure that the pump intake structure

generates favorable flow conditions at the inlet to the pump Evaluation for the

requirement of model test if

a Sump or piping geometry that deviates from this design standard

b Non-uniform or non-symmetric approach flow to the pump sump

exist

c The pumps have flows greater than 2520 ls per pump or the total

station flow with all pumps running would be greater than 6310 ls

d Situations that need ten times cost than the model study

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

11

Adverse hydraulic conditions that can affect pump performance include free and

submerged vortices swirl approaching the pump impeller flow separation at the

pump bell and a non-uniform axial velocity distribution at the suction

The purpose of model test is to examine these adverse hydraulic conditions and to

investigate measure against them

221 The scale effects

One possible difficulty in scale modeling is that it is impossible to reduce all

pertinent forces by the same factor With Froude-scale models the inertial and

gravitational forces are reduced similarly but the viscous and surface tension forces

cannot be simultaneously reduced The extra influence of these forces is known as

a ldquoscale effectrdquo A number of investigations have indicated that the scale effects for

the model are avoided if the Reynolds number based on the inlet flow and

submergence or intake diameter exceeds approximately 3x 104 and the Weber

number exceeds 150 [19]

In most engineering applications involving closed-conduit and open-channel flow

the Reynolds number limits are far exceeded and the flows are fully turbulent

In a sump model Froude number (Fr) and Reynolds number (Re) are the most

important non-dimensional parameters Some reduction in Re should be

introduced to the sump model However flow patterns are generally very similar

at high Re

No specific geometric scale rate is recommended but reasonable large

geometric scale to minimize viscous and surface tension scale effects and reproduce

the flow pattern

Re= uDν (22)

We= u2ρDσ (23)

where

u average axial velocity (ms)

ν kinematic viscosity of the liquid (m2s)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

12

ρ liquid density (kgm3)

σ surface tension of liquidair interface

According to the ANSIHI 98 the viscous scale effects on vortex may be negligible

if the resulting dimensionless numbers must meet these values Re>6times104 and

We>240

To allow practical visual observations of flow patterns accurate measurements of

swirl and velocity distribution and sufficient dimensional control the model scale

shall yield a bay width of at least 300mm a minimum liquid depth of at least

150mm and a pump throat or suction diameter of at least 80mm

222 Principle of similarity

In the similitude analysis the geometric and flow similarity requires the model and

the flow to be identical as the real model and flow This is to ensure the result

obtained from the model study is well presented in order to predict full scale

behavior Model involving a free surface are operated using Froude similarity since

the flow process is controlled by gravity and inertial forces

For similarity of the flow patterns the Froude number shall be equal in model and

prototype

Fr =

=

(24)

and as the scale ratio λ was selected in section 221 the following equation is

obtained

V = V times (

) = V times λ or Q = Q times (

) = Q times λ (25)

where Fr is the Froude number λ is the scale ratio of model dimensions to prototype

The test of 15 times Froude scaled flow should be conducted to ascertain the

potential scale affect on free surface vortices

Models of closed conduit piping system leading to a pump suction are not operated

based on Froude similitude However flow patterns are generally very similar at

high Re A minimum value of 1times105 for the Re at the pump suction is

recommended to ensure the flow patterns are correctly scaled

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

13

223 Preliminary model test

The initial design shall be tested first to identify any hydraulic problems

1) Vortex Formation Mechanisms

There are various investigations that provide insight into the fundamental processes

leading to the development of vortices both in experiment and numerical simulation

Shin et al [20] demonstrated that two basic mechanisms lead to inlet vortex

formation The first mechanism involves the development of an inlet vortex due to

the amplification of ambient vorticity in the approach flow as vortex lines

are convicted into the inlet The second mechanism involves the development of a

trailing vortex in the vicinity of the intake as a result of the variation in

circulation along the inlet For this second case a vortex can develop in a flow that

is irrotational upstream and the vortex development therefore does not depend

on the presence of ambient vorticity Shin et al investigation on kinematic

parameters indicate that the strength of an inlet-vortex or trailing vortex system

increases with decreasing distance from the surface However for an inlet in

an upstream irrotational flow two counter rotating vortices can still trail from

the rear of the inlet

2) Classification of vortex type

As illustrated in Fig 24 vortices in the vicinity of pump intakes may be adjacent to

the channel bottom or a channel wall (ie submerged vortices) or they may appear

adjacent to the free surface (ie free surface vortex)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

14

(a) Schematic diagram of vortices in a sump

(b) Experimental results of different vortex

Fig 24 Typical vortex in a sump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

15

For the model visual observation dye was injected around the intake structure near

the surface and submergence as Fig 24 (b) shows And the model test results of

vortex was classified as Fig 25

(a) Free Surface Vortices

(b) Submerged Vortices

Fig 25 Vortex classifications

There are a number of researches focused on the numerical theory to detect the

inception of a visible vortex As the diameters of vortices are usually much smaller

compared with the computing grid size to compensate for the lack of vortex

resolution in numerical simulation a stretching vortex model is applied to the local

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

16

flow field around semi-analytically identified vortex positions [21][22] The brief

explanations are as follows

a) Free surface vortex

The criteria expressed by equation 26 and 27 is applied to determine the

visible inception of a free surface vortex

≫ 1 (26)

=

gt 1 (27)

where P is the pressure drop at the vortex core and Ph is the static head at the

vortex element and f is the drag force acting at a unit length of water surface

vortex dimple based on Stokes approximation and caused by the down flow

Further f is the force acting on the bubble in the direction from the high

pressure region to the low pressure region like buoyancy

b) Submerged vortex

Equation 28 is applied to determine the visible inception of a submerged vortex

infin =

infin

infin gt 1 (28)

where is the static pressure at the vortex core infin is the atmospheric pressure

and is the critical cavitation inception pressure and nearly equal to the

saturated vapor pressure

3) Swirl angle

Swirl angle predicts the intensity of flow rotation Fig 26 shows experiment

calculation method with swirl meter and attempts of the CFD calculation method

were conducted as Fig 27 A four blade zero-pitch swirl meter which is supported

by a low friction bearing will be installed at about four suction pipe diameter (d)

downstream from the pump suction to measure swirl angle of pump approaching

flow For the swirl meter the tip to tip blade diameter is 075d and the length in

flow direction is 06 and one of the four blade is painted yellow as a reference to

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

17

count revolutions As flow swirl is usually unsteady the observation time of swirl

meter reading should be a continuous period of time

Fig 26 Typical swirl meter

(29)

(210)

(211)

Where qV flow rotational speed at swirl meter (ms)

zV average axial velocity at swirl meter (ms)

n Number of revolution per minute (rpm)

d Diameter of pipe at the swirl meter (m)

q flow rate (m3s)

zV

Vqq 1tan -=

)(60

smnd

Vtimestimes

=p

q

)(4

2

smd

qVz

times=

p

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

18

For the swirl angle calculating method with CFD the key point is to obtain the

average tangential velocity so the swirl check circles were created as illustrated in

Fig 27 The swirl check circle is located at the section of 4d height with diameter

of 025d 05d and 075d respectively

Swirl check circle

Fig 27 Schematic view of swirl angle numerical calculation

23

1 7501 501 250

acute++

=aringaringaring

N

i

N

i

NNVNVNV

Vi

q (212)

where

qV average tangential velocity (ms)

V circumferential velocity of check point in check circle

respectively (ms)

N check point total numbers in single check circle

4) Velocity

Velocity measurement at bell throat by Pitot tube was implemented by ANSIHI

method Typical Pitot tube configuration is shown in Fig 28 Eight measuring

points are defined as Fig 29 Time averaged velocity at eight points was obtained

and these eight data were again averaged to give throat sectional spatial velocity

average

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

19

Fig 28 Example of Pitot tube Fig 29 Measuring points at bell throat

224 Test criteria

There is the criteria for satisfactory performance of the sump model test and

summarized in Table 23

1) Free surface and submerged vortices

Free surface vortices entering the pump must be less severe than vortices with

coherent core Dye core vortices may be acceptable only if they occur less than 10

of the time or only for infrequent pump operating conditions

Submerged vortex which starts from the sump walls induce vibration and noise

problems Dye core submerged vortices extending into the pump bell mouth is not

allowed

2) Swirl angles

Swirl angle is used to check the flow rotation in the suction pipe The calculation

method is referred to section 223

Short-term maximum and long-term average values must be less than 5 degrees

Maximum short-term swirl angles up to 7 degrees may be acceptable only if they

occur less than 10 of the time or only for infrequent pump operating conditions

3)Velocity

Each point time-averaged velocity shall not be larger than 10 of sectional average

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

20

Standard deviation of time-varying velocity at each point was obtained and its

value shall not be larger than 10 of the time-averaged velocity at each point

1) Time-averaged

Variation[] = 100acute-

a

a

V

VV aV Averaged at section

2) Time-varying

Standard deviation= )1(

)(1i

2

-

-aring=

n

xxn

i

Variation[] = (Standard deviation X ) x 100

X Averaged at point

Table 23 Permission criteria of ANSIHI

Vortex occurence Swirl angle

Free surface vortex Submerged vortex

Observation

time Above 10 minutes Above 10 minutes

Above 10

minutes

Permission

criteria

Dye-core

Vortex(F3)

below 110 total

time

Dye-core

Vortex(S2)

below 110 total

time

Below

5degConditional

intermittent

below 7deg

225 Remedial measures for problem intakes

Usually in the preliminary test the worst flow condition is simulated And based on

the results remedial measures for problem intakes are discussed in this part

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

21

1) Overall approach flow patterns

The characteristics of flow approaching an intake structure is one of the foremost

considerations for the designer It is difficult to predict the effects of a given set of

flow conditions upstream from an intake structure Fig 210 depicts several typical

approach flow patterns of pump sump with different operating conditions

Fig 210 Examples of approach flow patterns

Generally speaking for the intake problems especially for the multiple pumps

sump there are the main measures using dividing walls placed between the pumps

as a partitioned structure trash racks with elongated bars guide piers or a number

of structural columns at the bay entrance can provide some assistance in controlling

cross-flow baffles flow distributors and screen mesh etc are applied to improve

the flow distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

22

2) Problems around pump intake

a) Free surface vortices

There are a number of factors that have an effect on the formation of surface

vortices To achieve a higher degree of certainty that objectionable surface vortices

do not form modifications can be made to intake structure to allow operation at

practical depths of submergence and the minimum submergence is referred in Table

22 the use of suction umbrellas vertical curtain walls and horizontal gratings

b) Submerged vortices

In the vicinity of intake bell there are the most complicated flow patterns and flow

makes the most changes in direction while maintaining a constant acceleration into

the pump bell to prevent local flow separation turbulence and vortex formation

Fig 211 represents a sample of various devices to inhibit the different submerged

vortex

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

23

Fig 211 Methods to reduce submerged vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

24

c) Preswirl

The approach flow distributions determine whether flow preswirl exists and a

sufficiently laterally skewed approach flow cause rotating around the intake bell

The most effective way of reducing preswirl is to establish a relatively uniform

approach flow within each pump bay by using the baffling schemes controlling

cross-flow and expanding concentrated flow discussed in the above part

To prevent non-uniform velocity distribution swirls and vortices necessary

modifications are conducted such as fore bay length baffle wall distributors

curtain wall mesh screen guide vane floor splitter wall fillet etc As the above

measures applied modification test will be conducted to check all the flow

conditions and compare with the preliminary results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

25

Chapter 3 Methodology

CFD (Computational fluid dynamics) is the technique which can design cycles to

better performance and reduce costs and time In CFD we solve the governing

equations of given physics (may be differential form or integral form) using some

numerical techniques like Finite Difference Method (FDM) Finite Element Method

(FEM) or Finite Volume Method (FVM) Although CFD plays a great part in the

design of turbomachinery CFD is not a replacement for experimental or analytical

approach In this study the numerical analysis was performed by the commercial

code ANSYS CFX 130 All aspects of the CFD study including modelling

meshing solving and post-processing of the results

31 Numerical modeling

ANSYS CFX solves the unsteady Navier-Stokes equations in their conservation

form The instantaneous equation of mass (continuity) in the stationary frame is

expressed as equation

ρ

δ + nabla ∙ (ρU) = 0 (31)

And the instantaneous equation for momentum is expressed as shown in equation

ρ

δ + nabla ∙ (ρU⨂U) = minusnablap + nabla ∙ τ + S (32)

Where τ is the stress tensor ( including both normal and shear components of the

stress)

These instantaneous equations are averaged for turbulent flows leading to additional

terms that need to be solved While the Navier-Stokes equations describe both

laminar and turbulent flows without addition terms realistic flows involve length

scales much smaller than the smallest finite volume mesh A Direct Numerical

Simulation of these flows would require significantly more computing power than

what is available now or in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

26

Therefore much research has been done to predict the effects of turbulence by

using turbulence models These models account for the effects of turbulence

without the use of a very fine mesh or direct numerical simulation

These turbulence models modify the transport equations by adding averaged and

fluctuating components The transport equations are changed to equations (33) and

(34)

ρ

δ + nabla ∙ (ρU) = 0 (33)

ρ

+ nabla ∙ ρ U ⨂U = minusnablap + nabla ∙ τ minus ρu⨂u + S (34)

The mass equation is not changed but the momentum equation contains extra terms

which are the Reynolds stresses ρ u⨂u and the Reynolds flux ρ u ϕ These

Reynolds stresses need to be modeled by additional equations to obtain closure

Obtaining closure implies that there are a sufficient number of equations to solve for

all the unknowns including the Reynolds stresses and Reynolds fluxes

Various turbulence models provide various ways to obtain closure In this

investigation Shear Stress Transport (SST) model and k-ε model were utilized The

advantage of SST model is that combines the advantages of other turbulence models

(the k-ε Wilcox k-ω) The 3 turbulence models will be discussed briefly

32 Turbulence models

Turbulence models are necessary because we cannot afford big enough computers

to directly capture every scale of motion There are three modelling frameworks

Direct Numerical Simulation (DNS) Reynolds-Averaged Navier-Stokes

Equations(RANS) Large Eddy Simulation(LES)

As RANS method is applied in most productionseveral RANS turbulence models

differ in their usage of wall functions the number of additional variables solved for

and what these variables represent

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

27

Two equation turbulence models

Two-equation turbulence models are widely used in CFD as they give a good

compromise between computational power needed and accuracy The term lsquotwo

equationrsquo refers to the fact that these models solve for the velocity and length scales

using separate transport equations The turbulence velocity scale is obtained by

solving the transport equation The turbulent length scale is estimated from two

properties of the turbulence field namely the turbulent kinetic energy and the

dissipation rate The dissipation rate of the kinetic energy is obtained from its

transport equation The most widely used are k-ε and k-ω two equation models In

the next section the k-ε Wilcox k-ω BSL k-ω and SST models will be briefly

discussed

321 k-ε turbulence model

The k-epsilon model solves for two variables k the turbulent kinetic energy and

epsilon the rate of dissipation of kinetic energy Wall functions are used in this

model so the flow in the buffer region is not simulated The k-epsilon model is very

popular for industrial applications due to its good convergence rate and relatively

low memory requirements It does not very accurately compute flow fields that

exhibit adverse pressure gradients strong curvature to the flow or jet flow It does

perform well for external flow problems around complex geometries

The k-ε model introduces k (m2s2) as the turbulence kinetic energy and ε (m2s3) as

the turbulence eddy dissipation The continuity equation remains the same

ρ

+ nabla ∙ (ρ U) = 0 (35)

The momentum equation changes as shown by equation

δρ

δ + nabla ∙ (ρU⨂U) = minusnablapprime + nabla ∙ μ

nablaU + (nablaU) + S (36)

Where SM is the sum of body forces μeff is the effective viscosity accounting for

turbulence and prsquo is the modified pressure The k-ε model uses the concept of eddy

viscosity giving the equation for effective viscosity as shown by equation

μ

= μ + μ (37)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

28

μt is the turbulence viscosity is linked to the turbulence kinetic energy and

dissipation by the equation

μ = Cμρ

ε (38)

Where Cμ is a constant

The values for k and ε come from the differential transport equations for the

turbulence kinetic energy and the turbulence dissipation rate

The turbulence kinetic energy equation is given as equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus ρε (39)

The turbulence dissipation rate is given by equation

(ρε)

+ nabla ∙ (ρUε) = nabla ∙ μ +

μ

σε nablaε +

ε

(Cε (P + Pε ) minus Cε ρε) (310)

Where Cε1 Cε2 σk σε are constants

Pk is the turbulence production due to viscous forces and is modeled by the

equation

P = μ nablaU ∙ nablaU + nablaU minus

nabla ∙ U 3μ

nabla ∙ U + ρk (311)

If a buoyancy term is added to the previous equation if the full buoyancy model is

used

The buoyancy term Pkb is modeled as

P = minusμ

ρσρg ∙ nablaρ (312)

322 Wilcox k-ω turbulence model

The k-omega model is similar to k-epsilon instead however it solves for omega mdash

the specific rate of dissipation of kinetic energy It also uses wall functions and

therefore has comparable memory requirements It has more difficulty converging

and is quite sensitive to the initial guess at the solution Hence the k-epsilon model

is often used first to find an initial condition for solving the k-omega model The k-

omega model is useful in many cases where the k-epsilon model is not accurate

such as internal flows flows that exhibit strong curvature separated flows and jets

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

29

This model has an advantage over the k-ε model where it does not involve complex

linear damping functions for near wall calculations at low Reynolds numbers

The k-ω model assumes that the turbulence viscosity is related to the turbulence

kinetic energy k and the turbulent frequency ω by the equation

μ = ρ

ω (313)

The transport equation for k is given by the equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus βprimeρ k ω (314)

The transport equation for ω is shown as equation

(ρω)

δ + nabla ∙ (ρ U ω) = nabla ∙ μ +

μ

μω

nablaω + αω

P + Pω minus β ρ ω (315)

The production rate of turbulence (Pk) is calculated as shown previously in the k-ε

section

The recommended values for the constants in the above equations are

βprime = 009

α = 59

β = 0075

σ = 2

σω = 2

The Reynolds stress tensor ρ u⨂u is calculated by

minusρ u⨂u = μ nablaU + (nablaU) minus

part ρk + μ

nabla ∙ U (316)

323 Shear stress transport model

The disadvantage of the Wilcox model is the strong sensitivity to free-stream

conditions Therefore a blending of the k-ω model near the surface and the k-ε in

the outer region was made by Menter [23] which resulted in the formulation of the

BSL k-ω turbulence model It consists of a transformation of the k-ε model to a k-ω

formulation and subsequently adding the resulting equations The Wilcox model is

multiplied by a blending function F1 and the transformed k-ε by another function 1-

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

30

F1 F1 is a function of wall distance (being the value of one near the surface and zero

outside the boundary layer) Outside the boundary and on the edge of the boundary

layer the standard k-ε model is used [24]

SST model is a combination of the k-epsilon in the free stream and the k-omega

models near the walls It does not use wall functions and tends to be most accurate

when solving the flow near the wall The SST model does not always converge to

the solution quickly so the k-epsilon or k-omega models are often solved first to

give good initial conditions

The k-ω based Shear Stress Transport (SST) model of Menter was applied for

turbulence treatment The transport equations for the SST model are expressed

below where the turbulent kinetic energy lsquokrsquo and turbulent frequency or dissipation

per unit turbulent kinetic energy are computed by using the following relations

For Turbulence Kinetic Energy

kkPukt

kTkk Ntilde+Ntilde+-=Ntilde+

para

para)[()( nsnwb (317)

where Pk is the production limiter

For Specific Dissipation Rate

ww

swnsnbwaww

ww NtildeNtilde-+Ntilde+Ntilde+-=Ntilde+para

para

1)1(2])[()( 21

22 kFSut

T (318)

The first blending function F1 is calculated from

iumlthorn

iumlyacute

uuml

iumlicirc

iumliacute

igrave

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

uacuteuacuteucirc

ugrave

ecircecirceuml

eacute

dividedivideoslash

oumlccedilccedilegrave

aelig=

4

22

21

4500maxmintanh

yCD

k

yy

kF

kw

wrs

w

n

wb (319)

y is the distance to the nearest wall and v is the kinematic viscosity

divideoslash

oumlccedilegrave

aeligNtildeNtilde= -10

2 101

2max ww

rsww kCDk (320)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

31

And Kinematic eddy viscosity

)max( 21

1

SF

kT

wa

an = (321)

where F2 is a blending function which restricts the limiter to the wall boundary and

S is the invariant measure of the strain rate

uacuteuacute

ucirc

ugrave

ecircecirc

euml

eacute

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

dividedivideoslash

oumlccedilccedilegrave

aelig=

2

22

5002maxtanh

w

n

wb yy

kF (322)

33 Brief introduction of simulation process

All aspects of CFD study including modelling meshing solving and post-

processing of the results are briefly introduced in this chapter

331 Pre-processing technique

The speed and accuracy of simulation results depending fundamentally on meshing

and the ability to prepare model geometry correctly

The steps in pre-processing are explained as follows

1) Domain Definitions

The domain definition depends mainly on the physics involved in the problem and

objective of the problem

a) Geometry creation or geometry import

b) Geometry cleanup

i) Removing the parts not necessary for simulation under consideration

ii) Closing the gaps in the geometry

iii) Removing small surfaces curves to make meshing process simple

2)Grid generation

To solve the governing equations the numerical methods like Finite Difference

Finite Volume or Finite Element Method to divide the domain into small parts and

solve the equations on one part at a time are applied

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 7: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

iv

List of Figures

Fig 21 Recommended inlet bell design diameter6

Fig 22 Recommended intake structure layout7

Fig 23 Filler wall details for proper bay width7

Fig 24 Typical vortex in a sump14

Fig 25 Vortex classifications15

Fig 26 Typical swirl meter17

Fig 27 Schematic view of swirl angle numerical calculation 18

Fig 28 Example of Pitot tube 19

Fig 29 Measuring point at bell throat19

Fig 210 Examples of approach flow patterns21

Fig 211 Methods to reduce submerged vortices23

Fig 41 Geometry of scaled sump model33

Fig 42 Schematic diagram of sump with anti free surface vortex devices34

Fig 43 Overall 3D drawing of the sump model35

Fig 44 2D drawing of the sump sketch36

Fig 45 Shape of the mixed flow pump37

Fig 46 Sump model with the submerged AVD38

Fig 47 Submerged AVD dimensions39

Fig 48 Overall mesh generation for scaled sump models40

Fig 49 Overall sump fluid domain mesh for the pump sump model41

Fig 410 Grid details of the pump sump model41

Fig 411 Boundaries of the scaled sump43

Fig 412 Experimental setup for the scaled sump model46

Fig 51 Streamline pattern of free surface47

Fig 52 Vortex core region of the sump model48

Fig 53 Velocity vectors on different sections48

Fig 54 Velocity vectors of PIV results51

Fig 55 Velocity vectors on different sections (PIV)52

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

v

Fig 56 Velocity vectors on free surface of different cases53

Fig 57 Tangential velocity profile of different models55

Fig 58 Performance curve of the mixed flow pump with the sump 57

Fig 59 Streamline pattern around the intake structure58

Fig 510 Vortex core region of the pump sump model59

Fig 511 Vorticity distribution in the flow direction60

Fig 512 Vorticity distribution in channel width direction61

Fig 513 Cavitation region of the mixed flow pump62

Fig 514 Cavitation performance curves of the mixed flow pump63

Fig 515 Vapor volume fraction distribution on blade surface64

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

vi

Study of Flow Field in Rectangular Sump Models and

Performance Analysis of a Mixed Flow Pump

Yuxin ZHAO

Department of Mechanical Engineering

Graduate School of Korea Maritime and Ocean University

Abstract

It is an accepted fact that quite a number of problems faced in a pumping station are

related to the design of sump or intake rather than pump design Head-capacity

curves provided by the pump manufacturer are obtained on the condition of no

vortices flowing into the pump intake The efficiency and performance of pumping

stations depend not only on the performance of the selected pumps but also on the

proper design of the intake sumps A faulty design of pump sump can lead to the

occurrence of swirl and vortices which reduce the pump performance and induce

vibration and additional noise Therefore sump model test is necessary to check the

flow condition around intake structure Numerical simulation is a good facility for

reducing the time and cost involved throughout the design process In this study the

commercial software ANSYS CFX-130 has been used for the CFD analysis of the

sump models

In a scaled sump model for air entrainment simulation numerical analysis of single

phase and two-phase with SST turbulence model was carried out to predict vortex

(both free surface vortex and submerged vortex) occurrence and location The

effectiveness of curtain walls and square bars to eliminate the free surface vortex is

evaluated Meanwhile the experiment including PIV system was performed in

KMOU to investigate the flow conditions around pump intake structure

In a full size sump model overall numerical analysis for the sump model with a

mixed flow pump installed was carried out Hydraulic performances of the mixed

flow pump for head rise shaft power pump efficiencies versus flow rate changed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

vii

from 50 to 140 of the design flow rate were studied by the performance curves

A trident shaped anti vortex device (AVD) composed of three wall fillets and one

center splitter was installed under the pump intake The effectiveness of the AVD

for the suppression of submerged vortex was evaluated In addition numerical

simulation of cavitation phenomenon in the mixed flow pump has been performed

by calculating the full cavitation model with k-ε turbulence model

According to the results details of the location size and strength of vortices were

predicted in the numerical simulation For the scaled sump model although there is

a little difference between the single phase and two-phase simulation all the results

predict the free surface vortex and submerged vortex formation and location The

CFD simulations of flow condition show good agreement with the PIV results The

effectiveness of curtain walls installed to inhibit the free surface vortex is confirmed

in the CFD results while the square bars installed in the channel failed to suppress

the free surface vortex For the mixed flow pump with sump the best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

The effectiveness of AVDs installed for the submerged vortex is confirmed by

comparisons of the sump model with and without AVD results

Cavitation phenomenon analysis shows the pump operating at the BEP condition

has a considerable extent over the requirement of cavitation performance With

numerical simulation the inception of cavitation in the blade passage is observed on

the suction surface where the leading edges meet the tips and then as the inlet total

pressure decrease the cavitation zone is spread out over the suction surface as well

as the leading edge of the impeller blade

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

viii

Nomenclature

B Distance from the back wall to the pump inlet bell centerline [m]

C Distance between the inlet bell and floor [m]

D Inlet bell design outside diameter [m]

Fr Froude number (=Vb(gD)05) [-]

g Acceleration due to gravity [ms2]

ht Total head [m]

n Number of revolution per min [-]

Pflow Flow power [W]

Pshaft Shaft power [W]

Pν Vapor pressure [Pa]

Q Volume flow rate [m3s]

Q0 Design volume flow rate [m3s]

Re Reynolds number (=VbDν) [-]

S Minimum submergence depth [m]

Vb Velocity at bell mouth [ms]

α Angle of floor slope [deg]

η Pump efficiency []

θ Swirl angle [deg]

λ The scale ratio of model dimensions to prototype [-]

μ Viscosity [Pa s]

ν Kinematic viscosity [m2s]

ρ Specific water density [kgm3]

σ Surface tension [Nm]

τ Stress tensor [-]

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

ix

Abbreviations

AVD Anti vortex device

BEP Best efficiency point

CFD Computational fluid Dynamics

NPSH Net positive suction head

NPSHR Required net positive suction head

PIV Particle Image Velocimetry

RANS Reynolds Averaged Navier- Stokes

RPM Number of revolutions per minute

SST Shear Stress Transport

VOF Volume of Fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

1

Chapter 1 Introduction

11 Background

It is an accepted fact that quite a number of problems faced in a pumping station are

related to the design of sump or intake rather than pump design As an important

part of the pumping station sump is designed to provide uniform swirl and

entrained air free flow to the pump intake Undesirable flow conditions (entrained

air non-uniform flow distribution vortices etc) can reduce the pump efficiency

induce vibration noise and cavitation even lead to excessive bearing loads of

impeller and structure damage Due to quite complex nature of the flow although

the flow conditions causing pump problems are well established there are no

specific solutions to eliminate them There are some design guidelines for the

specific geometrical and hydraulic constraints of the pump sump for any project

the model study is the only tool for solving potential problems in new designs and

rectifying problems observed in existing installations However the scaled model

tests are expensive and time-consuming so alternative Computational Fluid

Dynamics (CFD) methods for evaluating sump performance have been developed

With rapid progress in CFD numerical simulation is regarded as an effective tool in

solving fluid problems in pump sumps

12 Previous study

Various researches show the flow conditions around the sump intake structure are

quite complicated Shyam et al [1] used the commercial code CFX to carry out

two-phase flow simulation to capture air entrainment The air entrainments and its

location were well captured and numerical results were in accordance with the

experimental investigation Matsui J[2] and Okamura T[3] et al conducted a serial

research of the benchmark according to the Turbomachinery Society of Japan (TSJ)

standard including various commercial codes comparison performed in the flow

simulation and comparison analysis between the numerical and experimental results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

2

The purpose of the benchmark is to determine the accuracy and reliability of the

CFD codes used in universities and the industrial field It is also intended to identify

and investigate the formation of air-entrained and submerged vortices near the

pump Tanweer et al[4] have performed a numerical analysis of a multiple intake

pump sump to confirm that the geometry of sump affects the flow within sump

greatly Rajendran et al[5] have made a numerical analysis for the flow

characteristics of a sump model with pump intake and good agreements were

achieved by comparing the numerical results with the experiments Iwano et al[6]

have introduced a numerical method for the submerged vortex by analyzing the

flow in the pump sump with and without baffle plates Lee et al[7] have conducted

the CFD analysis of a multi-intake pump sump model to check the flow uniformity

by predicting the location number and vorticity of the vortex For the CFD

application in mixed flow pump design and performance analysis Oh H W [8]

conducted a mixed flow pump design optimization with specific parameters and

then compared the numerical results with the experiments The procedure presented

can be used as a practical design and analysis guide for the general turbomachinery

Kim et al[9] carried out a mixed flow pump performance analysis with numerical

simulation and improved the hydraulic efficiency by the modification of the

geometry

As an undesirable phenomenon caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage The cavitation

phenomenon has been studied by various investigators Van et al[10] have

investigated the cavitation inception behavior of a mixed-flow pump impellers

model and the calculated results matched well with the experiments Okamura et

al[11] have evaluated two numerical cavitation prediction methods used in the

pump industry manufacturers and attempts were made to improve the cavitation

performance Li et al[12] have conducted the numerical analysis on the basis of the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

3

development liquidvapor interface tracking method to predict the cavitation

characteristics with a centrifugal pump impeller model

Moreover there are a number of correlated studies [13]~[17]

13 Methodology of study

This study will investigate the flow phenomena of two sump models

A scaled sump model is adopted to carry out numerical simulation and PIV

experiment The flow conditions are predicted and the effectiveness of devices for

free surface vortex is evaluated by using numerical code a single intake sump

model with a mixed flow pump installed is modeled and the hydraulic performance

characteristics for the head rise shaft power and pump efficiencies versus flow rate

are studied In addition flow phenomena (vortices cavitation etc) are analyzed

and the effectiveness of an anti-vortex device (AVD) for the submerged vortex is

evaluated

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

4

Chapter 2 Basic theory of sump model

Specific hydraulic phenomena have been confirmed that they can lead to certain

common operational problems The common of the main problems are summarized

as follows

a Free surface vortices

b Submerged vortices

c Swirl of flow entering the pump

d Non-uniform distribution of velocity at the pump impeller

e Entrained air or gas bubbles

These problems encountered in the pump sump will affect the pump

performance and significantly increase the operational and maintenance costs In

order to identify sources of particular problems and find practical solution for

it the usual approach is to conduct the laboratory experiments on a scaled

physical model Typical design objectives are to ensure that a pump station is

designed according to best practice and conforms to the requirements set out in the

American National Standard for Pump Intake Design[18] These standards restrict

the degree to which the above-mentioned undesirable flow patterns may be present

The negative impact of each of these hydraulic conditions varies with pump specific

speed and size Application of the standard to design a sump does not generate a

problem free sump but provides a basis for the initial design

As there are no specific guidelines or criteria for design of trouble free intakes the

most common solution to potential problems in new design and rectification of

problems observed in existing designs is to construct a scaled model in a laboratory

observe and investigate the flow characteristics and propose modifications to the

intake geometry There are additional devices in the form of floor splitters or cones

back wall splitters fillets surface beams guide vanes etc aimed at controlling the

vortex and swirl formation Considerable prior experience and ingenuity are

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

5

required to solve the problems by this method and the cost and time required are

also significant

The intake structure refers to the channel leading from the water source which

may be a river or reservoir and all installations downstream including the pump

column or the intake pipe (the suction tube portion of a vertical intake) the

approach channel (upstream of pump bay) and the pump bay (bounded by the floor

the back wall side walls dividers walls separating adjacent pump column) Usually

the upstream end of the pump column has an inlet attachment called the suction bell

The function of the intake is to supply an evenly distributed flow of water to the

pump suction bell

Several types of pump intake basins are available and ANSIHI 98 provides

design requirements for each Intake structures can be categorized as being

clear liquids or solids-bearing liquids For clear liquids intakes are further

classified into rectangular formed circular and trench types as well as suction

tanks and cans For solid-bearing liquids trench type and rectangular wet wells are

usually considered

In this paper only rectangular intake structure for clear liquid is studied

21 Pump intake design

For pumps to achieve their optimum hydraulic performance across all operating

conditions the flow at the impeller must meet specific hydraulic conditions The

ideal flow entering the pump inlet should be of a uniform velocity distribution

without rotation and stable over time When designing a sump to achieve favorable

inflow to the pump or suction pipe bell various sump dimensions relative to the size

of the bell are required Using these standards of the distances to reduce the

probability of the occurence of strong submerged vortices and free-surface vortices

211 Inlet bell diameter design

The geometry is generally defined in terms of the pump inlet diameter as shown in

Table 21 and Fig 21 The bell diameter can be estimated based on an inlet pipe

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

6

Table 21 Acceptable velocity ranges for inlet bell diameter(ANSIHI 98)

Flow Rate

Qls

Recommended Inlet Bell Design

Velocityms

Acceptable Velocity

Range ms

Q<315 V=17 06leVle27

315 leQ<1260 V=17 09leVle24

Q ge1260 V=17 12leVle21

Fig 21 Recommended inlet bell design diameter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

7

212 Recommended dimensions for a rectangular sump

As the selected bell diameter has been determined the proportions of the inlet

structure can be estimated from Fig 22 and Fig 23 Table 22 shows the

recommended dimensions for a rectangular sump

Fig 22 Recommended intake structure layout

Fig 23 Filler wall details for proper bay width

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

8

Table 22 Recommended dimensions for a rectangular sump

Dimension Variable

Description Recommended value

A Distance from the pump inlet bell

centerline to the entrance A=5D minimum

a Length of constricted bay section near

the pump inlet 25D minimum

B Distance from the back wall to the

pump inlet bell centerline 075D

C Distance between the inlet bell and

floor 03D~05D

D Inlet bell outside diameter refer to section 211

H Minimum liquid depth H=S+C

h Minimum height of constricted bay

section near the pump inlet bell hgtH or 25D

S Minimum pump inlet bell submergence S=D(10+23Fr)

W Pump inlet bay entrance width 2D minimum

w Constricted bay width near the pump

inlet bell w=2D

X Pump inlet bay length 5D minimum

Y Distance from pump inlet bell

centerline to the through-flow traveling screen

4D minimum

Z1 Distance from sump inlet bell centerline to diverging walls

5D minimum

Z2 Distance from inlet bell centerline to

sloping floor 5D minimum

α Angle of floor slope -10deg~ +10deg

β Angle of wall convergence 0deg~ +10deg

empty Angle of convergence from constricted

area to by walls 10deg maximum

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

9

FD=Vb(gD)05 (21)

where

FD Froude number (dimensionless)

Vb velocity at suction inlet=FlowArea based on D

D Outside diameter of bell or pipe inlet

g gravitational acceleration

There is the design sequence summary for rectangular intake structures

middot Consider the flow patterns and boundary geometry of the body of liquid

from which the pump station is to receive flow Compare with the approach

flow condition and determine if a hydraulic physical model study is

required

middot Determine the number and size of pumps required to satisfy the range of

operating conditions likely to be encountered

middot Identify pump inlet bell diameter Refer to the section 211

middot Determine the bell-floor clearance 05D is a good preliminary design value

middot Determine the required bell submergence using the equation 21

middot Determine the minimum allowable liquid depth in the intake structure from

the sum of the floor clearance and the required bell submergence

middot Check bottom elevation near the entrance to the structure and determine if it

is necessary to slope the floor upstream of the bay entrance If the resulting

depth at the entrance to the intake structure is shallow then check to ensure

that gravity-driven flow is not restricted by the entrance condition

middot Check the pump bay velocity for the maximum single-pump flow and

minimum liquid depth with the bay width set to 2D If bay velocity exceeds

05 ms then increase the bay width to reduce the maximum velocity below

05 ms

middot If it is necessary to increase the pump bay width to greater than 2D then

decrease bay width in the vicinity of the pumps according to Fig 23

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

10

middot Compare cross-flow velocity (at maximum system flow) to average pump

bay velocity If cross-flow value exceeds 50 of the bay velocity a

physical hydraulic model study is necessary

middot Determine the length of the structure and dividing walls giving

consideration to minimum allowable distances to a sloping floor screening

equipment and length of dividing walls if dual flow traveling screens or

drum screens are to be used a physical hydraulic model study is required

middot If the final selected pump bell diameter and inlet velocity is within the

range given in section 211 then the sump dimensions (developed based on

the inlet bell design diameter) need not be changed and will comply with

these standards

22 Model test

The approach flow pattern in a pump intake facility is difficult to predict by

traditional mathematics or empirical formulae For large structures or those that

differ significantly from proven designs a model study is the only means to ensure

success A model study is used to identify adverse hydraulic conditions and

derive remedial measures for approach flow patterns generated by structures

upstream of the pump impeller Current model studies are not intended to

investigate flow patterns induced by the pump itself or the flow patterns within the

pump The objective of a model study is to ensure that the pump intake structure

generates favorable flow conditions at the inlet to the pump Evaluation for the

requirement of model test if

a Sump or piping geometry that deviates from this design standard

b Non-uniform or non-symmetric approach flow to the pump sump

exist

c The pumps have flows greater than 2520 ls per pump or the total

station flow with all pumps running would be greater than 6310 ls

d Situations that need ten times cost than the model study

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

11

Adverse hydraulic conditions that can affect pump performance include free and

submerged vortices swirl approaching the pump impeller flow separation at the

pump bell and a non-uniform axial velocity distribution at the suction

The purpose of model test is to examine these adverse hydraulic conditions and to

investigate measure against them

221 The scale effects

One possible difficulty in scale modeling is that it is impossible to reduce all

pertinent forces by the same factor With Froude-scale models the inertial and

gravitational forces are reduced similarly but the viscous and surface tension forces

cannot be simultaneously reduced The extra influence of these forces is known as

a ldquoscale effectrdquo A number of investigations have indicated that the scale effects for

the model are avoided if the Reynolds number based on the inlet flow and

submergence or intake diameter exceeds approximately 3x 104 and the Weber

number exceeds 150 [19]

In most engineering applications involving closed-conduit and open-channel flow

the Reynolds number limits are far exceeded and the flows are fully turbulent

In a sump model Froude number (Fr) and Reynolds number (Re) are the most

important non-dimensional parameters Some reduction in Re should be

introduced to the sump model However flow patterns are generally very similar

at high Re

No specific geometric scale rate is recommended but reasonable large

geometric scale to minimize viscous and surface tension scale effects and reproduce

the flow pattern

Re= uDν (22)

We= u2ρDσ (23)

where

u average axial velocity (ms)

ν kinematic viscosity of the liquid (m2s)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

12

ρ liquid density (kgm3)

σ surface tension of liquidair interface

According to the ANSIHI 98 the viscous scale effects on vortex may be negligible

if the resulting dimensionless numbers must meet these values Re>6times104 and

We>240

To allow practical visual observations of flow patterns accurate measurements of

swirl and velocity distribution and sufficient dimensional control the model scale

shall yield a bay width of at least 300mm a minimum liquid depth of at least

150mm and a pump throat or suction diameter of at least 80mm

222 Principle of similarity

In the similitude analysis the geometric and flow similarity requires the model and

the flow to be identical as the real model and flow This is to ensure the result

obtained from the model study is well presented in order to predict full scale

behavior Model involving a free surface are operated using Froude similarity since

the flow process is controlled by gravity and inertial forces

For similarity of the flow patterns the Froude number shall be equal in model and

prototype

Fr =

=

(24)

and as the scale ratio λ was selected in section 221 the following equation is

obtained

V = V times (

) = V times λ or Q = Q times (

) = Q times λ (25)

where Fr is the Froude number λ is the scale ratio of model dimensions to prototype

The test of 15 times Froude scaled flow should be conducted to ascertain the

potential scale affect on free surface vortices

Models of closed conduit piping system leading to a pump suction are not operated

based on Froude similitude However flow patterns are generally very similar at

high Re A minimum value of 1times105 for the Re at the pump suction is

recommended to ensure the flow patterns are correctly scaled

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

13

223 Preliminary model test

The initial design shall be tested first to identify any hydraulic problems

1) Vortex Formation Mechanisms

There are various investigations that provide insight into the fundamental processes

leading to the development of vortices both in experiment and numerical simulation

Shin et al [20] demonstrated that two basic mechanisms lead to inlet vortex

formation The first mechanism involves the development of an inlet vortex due to

the amplification of ambient vorticity in the approach flow as vortex lines

are convicted into the inlet The second mechanism involves the development of a

trailing vortex in the vicinity of the intake as a result of the variation in

circulation along the inlet For this second case a vortex can develop in a flow that

is irrotational upstream and the vortex development therefore does not depend

on the presence of ambient vorticity Shin et al investigation on kinematic

parameters indicate that the strength of an inlet-vortex or trailing vortex system

increases with decreasing distance from the surface However for an inlet in

an upstream irrotational flow two counter rotating vortices can still trail from

the rear of the inlet

2) Classification of vortex type

As illustrated in Fig 24 vortices in the vicinity of pump intakes may be adjacent to

the channel bottom or a channel wall (ie submerged vortices) or they may appear

adjacent to the free surface (ie free surface vortex)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

14

(a) Schematic diagram of vortices in a sump

(b) Experimental results of different vortex

Fig 24 Typical vortex in a sump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

15

For the model visual observation dye was injected around the intake structure near

the surface and submergence as Fig 24 (b) shows And the model test results of

vortex was classified as Fig 25

(a) Free Surface Vortices

(b) Submerged Vortices

Fig 25 Vortex classifications

There are a number of researches focused on the numerical theory to detect the

inception of a visible vortex As the diameters of vortices are usually much smaller

compared with the computing grid size to compensate for the lack of vortex

resolution in numerical simulation a stretching vortex model is applied to the local

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

16

flow field around semi-analytically identified vortex positions [21][22] The brief

explanations are as follows

a) Free surface vortex

The criteria expressed by equation 26 and 27 is applied to determine the

visible inception of a free surface vortex

≫ 1 (26)

=

gt 1 (27)

where P is the pressure drop at the vortex core and Ph is the static head at the

vortex element and f is the drag force acting at a unit length of water surface

vortex dimple based on Stokes approximation and caused by the down flow

Further f is the force acting on the bubble in the direction from the high

pressure region to the low pressure region like buoyancy

b) Submerged vortex

Equation 28 is applied to determine the visible inception of a submerged vortex

infin =

infin

infin gt 1 (28)

where is the static pressure at the vortex core infin is the atmospheric pressure

and is the critical cavitation inception pressure and nearly equal to the

saturated vapor pressure

3) Swirl angle

Swirl angle predicts the intensity of flow rotation Fig 26 shows experiment

calculation method with swirl meter and attempts of the CFD calculation method

were conducted as Fig 27 A four blade zero-pitch swirl meter which is supported

by a low friction bearing will be installed at about four suction pipe diameter (d)

downstream from the pump suction to measure swirl angle of pump approaching

flow For the swirl meter the tip to tip blade diameter is 075d and the length in

flow direction is 06 and one of the four blade is painted yellow as a reference to

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

17

count revolutions As flow swirl is usually unsteady the observation time of swirl

meter reading should be a continuous period of time

Fig 26 Typical swirl meter

(29)

(210)

(211)

Where qV flow rotational speed at swirl meter (ms)

zV average axial velocity at swirl meter (ms)

n Number of revolution per minute (rpm)

d Diameter of pipe at the swirl meter (m)

q flow rate (m3s)

zV

Vqq 1tan -=

)(60

smnd

Vtimestimes

=p

q

)(4

2

smd

qVz

times=

p

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

18

For the swirl angle calculating method with CFD the key point is to obtain the

average tangential velocity so the swirl check circles were created as illustrated in

Fig 27 The swirl check circle is located at the section of 4d height with diameter

of 025d 05d and 075d respectively

Swirl check circle

Fig 27 Schematic view of swirl angle numerical calculation

23

1 7501 501 250

acute++

=aringaringaring

N

i

N

i

NNVNVNV

Vi

q (212)

where

qV average tangential velocity (ms)

V circumferential velocity of check point in check circle

respectively (ms)

N check point total numbers in single check circle

4) Velocity

Velocity measurement at bell throat by Pitot tube was implemented by ANSIHI

method Typical Pitot tube configuration is shown in Fig 28 Eight measuring

points are defined as Fig 29 Time averaged velocity at eight points was obtained

and these eight data were again averaged to give throat sectional spatial velocity

average

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

19

Fig 28 Example of Pitot tube Fig 29 Measuring points at bell throat

224 Test criteria

There is the criteria for satisfactory performance of the sump model test and

summarized in Table 23

1) Free surface and submerged vortices

Free surface vortices entering the pump must be less severe than vortices with

coherent core Dye core vortices may be acceptable only if they occur less than 10

of the time or only for infrequent pump operating conditions

Submerged vortex which starts from the sump walls induce vibration and noise

problems Dye core submerged vortices extending into the pump bell mouth is not

allowed

2) Swirl angles

Swirl angle is used to check the flow rotation in the suction pipe The calculation

method is referred to section 223

Short-term maximum and long-term average values must be less than 5 degrees

Maximum short-term swirl angles up to 7 degrees may be acceptable only if they

occur less than 10 of the time or only for infrequent pump operating conditions

3)Velocity

Each point time-averaged velocity shall not be larger than 10 of sectional average

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

20

Standard deviation of time-varying velocity at each point was obtained and its

value shall not be larger than 10 of the time-averaged velocity at each point

1) Time-averaged

Variation[] = 100acute-

a

a

V

VV aV Averaged at section

2) Time-varying

Standard deviation= )1(

)(1i

2

-

-aring=

n

xxn

i

Variation[] = (Standard deviation X ) x 100

X Averaged at point

Table 23 Permission criteria of ANSIHI

Vortex occurence Swirl angle

Free surface vortex Submerged vortex

Observation

time Above 10 minutes Above 10 minutes

Above 10

minutes

Permission

criteria

Dye-core

Vortex(F3)

below 110 total

time

Dye-core

Vortex(S2)

below 110 total

time

Below

5degConditional

intermittent

below 7deg

225 Remedial measures for problem intakes

Usually in the preliminary test the worst flow condition is simulated And based on

the results remedial measures for problem intakes are discussed in this part

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

21

1) Overall approach flow patterns

The characteristics of flow approaching an intake structure is one of the foremost

considerations for the designer It is difficult to predict the effects of a given set of

flow conditions upstream from an intake structure Fig 210 depicts several typical

approach flow patterns of pump sump with different operating conditions

Fig 210 Examples of approach flow patterns

Generally speaking for the intake problems especially for the multiple pumps

sump there are the main measures using dividing walls placed between the pumps

as a partitioned structure trash racks with elongated bars guide piers or a number

of structural columns at the bay entrance can provide some assistance in controlling

cross-flow baffles flow distributors and screen mesh etc are applied to improve

the flow distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

22

2) Problems around pump intake

a) Free surface vortices

There are a number of factors that have an effect on the formation of surface

vortices To achieve a higher degree of certainty that objectionable surface vortices

do not form modifications can be made to intake structure to allow operation at

practical depths of submergence and the minimum submergence is referred in Table

22 the use of suction umbrellas vertical curtain walls and horizontal gratings

b) Submerged vortices

In the vicinity of intake bell there are the most complicated flow patterns and flow

makes the most changes in direction while maintaining a constant acceleration into

the pump bell to prevent local flow separation turbulence and vortex formation

Fig 211 represents a sample of various devices to inhibit the different submerged

vortex

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

23

Fig 211 Methods to reduce submerged vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

24

c) Preswirl

The approach flow distributions determine whether flow preswirl exists and a

sufficiently laterally skewed approach flow cause rotating around the intake bell

The most effective way of reducing preswirl is to establish a relatively uniform

approach flow within each pump bay by using the baffling schemes controlling

cross-flow and expanding concentrated flow discussed in the above part

To prevent non-uniform velocity distribution swirls and vortices necessary

modifications are conducted such as fore bay length baffle wall distributors

curtain wall mesh screen guide vane floor splitter wall fillet etc As the above

measures applied modification test will be conducted to check all the flow

conditions and compare with the preliminary results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

25

Chapter 3 Methodology

CFD (Computational fluid dynamics) is the technique which can design cycles to

better performance and reduce costs and time In CFD we solve the governing

equations of given physics (may be differential form or integral form) using some

numerical techniques like Finite Difference Method (FDM) Finite Element Method

(FEM) or Finite Volume Method (FVM) Although CFD plays a great part in the

design of turbomachinery CFD is not a replacement for experimental or analytical

approach In this study the numerical analysis was performed by the commercial

code ANSYS CFX 130 All aspects of the CFD study including modelling

meshing solving and post-processing of the results

31 Numerical modeling

ANSYS CFX solves the unsteady Navier-Stokes equations in their conservation

form The instantaneous equation of mass (continuity) in the stationary frame is

expressed as equation

ρ

δ + nabla ∙ (ρU) = 0 (31)

And the instantaneous equation for momentum is expressed as shown in equation

ρ

δ + nabla ∙ (ρU⨂U) = minusnablap + nabla ∙ τ + S (32)

Where τ is the stress tensor ( including both normal and shear components of the

stress)

These instantaneous equations are averaged for turbulent flows leading to additional

terms that need to be solved While the Navier-Stokes equations describe both

laminar and turbulent flows without addition terms realistic flows involve length

scales much smaller than the smallest finite volume mesh A Direct Numerical

Simulation of these flows would require significantly more computing power than

what is available now or in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

26

Therefore much research has been done to predict the effects of turbulence by

using turbulence models These models account for the effects of turbulence

without the use of a very fine mesh or direct numerical simulation

These turbulence models modify the transport equations by adding averaged and

fluctuating components The transport equations are changed to equations (33) and

(34)

ρ

δ + nabla ∙ (ρU) = 0 (33)

ρ

+ nabla ∙ ρ U ⨂U = minusnablap + nabla ∙ τ minus ρu⨂u + S (34)

The mass equation is not changed but the momentum equation contains extra terms

which are the Reynolds stresses ρ u⨂u and the Reynolds flux ρ u ϕ These

Reynolds stresses need to be modeled by additional equations to obtain closure

Obtaining closure implies that there are a sufficient number of equations to solve for

all the unknowns including the Reynolds stresses and Reynolds fluxes

Various turbulence models provide various ways to obtain closure In this

investigation Shear Stress Transport (SST) model and k-ε model were utilized The

advantage of SST model is that combines the advantages of other turbulence models

(the k-ε Wilcox k-ω) The 3 turbulence models will be discussed briefly

32 Turbulence models

Turbulence models are necessary because we cannot afford big enough computers

to directly capture every scale of motion There are three modelling frameworks

Direct Numerical Simulation (DNS) Reynolds-Averaged Navier-Stokes

Equations(RANS) Large Eddy Simulation(LES)

As RANS method is applied in most productionseveral RANS turbulence models

differ in their usage of wall functions the number of additional variables solved for

and what these variables represent

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

27

Two equation turbulence models

Two-equation turbulence models are widely used in CFD as they give a good

compromise between computational power needed and accuracy The term lsquotwo

equationrsquo refers to the fact that these models solve for the velocity and length scales

using separate transport equations The turbulence velocity scale is obtained by

solving the transport equation The turbulent length scale is estimated from two

properties of the turbulence field namely the turbulent kinetic energy and the

dissipation rate The dissipation rate of the kinetic energy is obtained from its

transport equation The most widely used are k-ε and k-ω two equation models In

the next section the k-ε Wilcox k-ω BSL k-ω and SST models will be briefly

discussed

321 k-ε turbulence model

The k-epsilon model solves for two variables k the turbulent kinetic energy and

epsilon the rate of dissipation of kinetic energy Wall functions are used in this

model so the flow in the buffer region is not simulated The k-epsilon model is very

popular for industrial applications due to its good convergence rate and relatively

low memory requirements It does not very accurately compute flow fields that

exhibit adverse pressure gradients strong curvature to the flow or jet flow It does

perform well for external flow problems around complex geometries

The k-ε model introduces k (m2s2) as the turbulence kinetic energy and ε (m2s3) as

the turbulence eddy dissipation The continuity equation remains the same

ρ

+ nabla ∙ (ρ U) = 0 (35)

The momentum equation changes as shown by equation

δρ

δ + nabla ∙ (ρU⨂U) = minusnablapprime + nabla ∙ μ

nablaU + (nablaU) + S (36)

Where SM is the sum of body forces μeff is the effective viscosity accounting for

turbulence and prsquo is the modified pressure The k-ε model uses the concept of eddy

viscosity giving the equation for effective viscosity as shown by equation

μ

= μ + μ (37)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

28

μt is the turbulence viscosity is linked to the turbulence kinetic energy and

dissipation by the equation

μ = Cμρ

ε (38)

Where Cμ is a constant

The values for k and ε come from the differential transport equations for the

turbulence kinetic energy and the turbulence dissipation rate

The turbulence kinetic energy equation is given as equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus ρε (39)

The turbulence dissipation rate is given by equation

(ρε)

+ nabla ∙ (ρUε) = nabla ∙ μ +

μ

σε nablaε +

ε

(Cε (P + Pε ) minus Cε ρε) (310)

Where Cε1 Cε2 σk σε are constants

Pk is the turbulence production due to viscous forces and is modeled by the

equation

P = μ nablaU ∙ nablaU + nablaU minus

nabla ∙ U 3μ

nabla ∙ U + ρk (311)

If a buoyancy term is added to the previous equation if the full buoyancy model is

used

The buoyancy term Pkb is modeled as

P = minusμ

ρσρg ∙ nablaρ (312)

322 Wilcox k-ω turbulence model

The k-omega model is similar to k-epsilon instead however it solves for omega mdash

the specific rate of dissipation of kinetic energy It also uses wall functions and

therefore has comparable memory requirements It has more difficulty converging

and is quite sensitive to the initial guess at the solution Hence the k-epsilon model

is often used first to find an initial condition for solving the k-omega model The k-

omega model is useful in many cases where the k-epsilon model is not accurate

such as internal flows flows that exhibit strong curvature separated flows and jets

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

29

This model has an advantage over the k-ε model where it does not involve complex

linear damping functions for near wall calculations at low Reynolds numbers

The k-ω model assumes that the turbulence viscosity is related to the turbulence

kinetic energy k and the turbulent frequency ω by the equation

μ = ρ

ω (313)

The transport equation for k is given by the equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus βprimeρ k ω (314)

The transport equation for ω is shown as equation

(ρω)

δ + nabla ∙ (ρ U ω) = nabla ∙ μ +

μ

μω

nablaω + αω

P + Pω minus β ρ ω (315)

The production rate of turbulence (Pk) is calculated as shown previously in the k-ε

section

The recommended values for the constants in the above equations are

βprime = 009

α = 59

β = 0075

σ = 2

σω = 2

The Reynolds stress tensor ρ u⨂u is calculated by

minusρ u⨂u = μ nablaU + (nablaU) minus

part ρk + μ

nabla ∙ U (316)

323 Shear stress transport model

The disadvantage of the Wilcox model is the strong sensitivity to free-stream

conditions Therefore a blending of the k-ω model near the surface and the k-ε in

the outer region was made by Menter [23] which resulted in the formulation of the

BSL k-ω turbulence model It consists of a transformation of the k-ε model to a k-ω

formulation and subsequently adding the resulting equations The Wilcox model is

multiplied by a blending function F1 and the transformed k-ε by another function 1-

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

30

F1 F1 is a function of wall distance (being the value of one near the surface and zero

outside the boundary layer) Outside the boundary and on the edge of the boundary

layer the standard k-ε model is used [24]

SST model is a combination of the k-epsilon in the free stream and the k-omega

models near the walls It does not use wall functions and tends to be most accurate

when solving the flow near the wall The SST model does not always converge to

the solution quickly so the k-epsilon or k-omega models are often solved first to

give good initial conditions

The k-ω based Shear Stress Transport (SST) model of Menter was applied for

turbulence treatment The transport equations for the SST model are expressed

below where the turbulent kinetic energy lsquokrsquo and turbulent frequency or dissipation

per unit turbulent kinetic energy are computed by using the following relations

For Turbulence Kinetic Energy

kkPukt

kTkk Ntilde+Ntilde+-=Ntilde+

para

para)[()( nsnwb (317)

where Pk is the production limiter

For Specific Dissipation Rate

ww

swnsnbwaww

ww NtildeNtilde-+Ntilde+Ntilde+-=Ntilde+para

para

1)1(2])[()( 21

22 kFSut

T (318)

The first blending function F1 is calculated from

iumlthorn

iumlyacute

uuml

iumlicirc

iumliacute

igrave

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

uacuteuacuteucirc

ugrave

ecircecirceuml

eacute

dividedivideoslash

oumlccedilccedilegrave

aelig=

4

22

21

4500maxmintanh

yCD

k

yy

kF

kw

wrs

w

n

wb (319)

y is the distance to the nearest wall and v is the kinematic viscosity

divideoslash

oumlccedilegrave

aeligNtildeNtilde= -10

2 101

2max ww

rsww kCDk (320)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

31

And Kinematic eddy viscosity

)max( 21

1

SF

kT

wa

an = (321)

where F2 is a blending function which restricts the limiter to the wall boundary and

S is the invariant measure of the strain rate

uacuteuacute

ucirc

ugrave

ecircecirc

euml

eacute

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

dividedivideoslash

oumlccedilccedilegrave

aelig=

2

22

5002maxtanh

w

n

wb yy

kF (322)

33 Brief introduction of simulation process

All aspects of CFD study including modelling meshing solving and post-

processing of the results are briefly introduced in this chapter

331 Pre-processing technique

The speed and accuracy of simulation results depending fundamentally on meshing

and the ability to prepare model geometry correctly

The steps in pre-processing are explained as follows

1) Domain Definitions

The domain definition depends mainly on the physics involved in the problem and

objective of the problem

a) Geometry creation or geometry import

b) Geometry cleanup

i) Removing the parts not necessary for simulation under consideration

ii) Closing the gaps in the geometry

iii) Removing small surfaces curves to make meshing process simple

2)Grid generation

To solve the governing equations the numerical methods like Finite Difference

Finite Volume or Finite Element Method to divide the domain into small parts and

solve the equations on one part at a time are applied

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 8: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

v

Fig 56 Velocity vectors on free surface of different cases53

Fig 57 Tangential velocity profile of different models55

Fig 58 Performance curve of the mixed flow pump with the sump 57

Fig 59 Streamline pattern around the intake structure58

Fig 510 Vortex core region of the pump sump model59

Fig 511 Vorticity distribution in the flow direction60

Fig 512 Vorticity distribution in channel width direction61

Fig 513 Cavitation region of the mixed flow pump62

Fig 514 Cavitation performance curves of the mixed flow pump63

Fig 515 Vapor volume fraction distribution on blade surface64

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

vi

Study of Flow Field in Rectangular Sump Models and

Performance Analysis of a Mixed Flow Pump

Yuxin ZHAO

Department of Mechanical Engineering

Graduate School of Korea Maritime and Ocean University

Abstract

It is an accepted fact that quite a number of problems faced in a pumping station are

related to the design of sump or intake rather than pump design Head-capacity

curves provided by the pump manufacturer are obtained on the condition of no

vortices flowing into the pump intake The efficiency and performance of pumping

stations depend not only on the performance of the selected pumps but also on the

proper design of the intake sumps A faulty design of pump sump can lead to the

occurrence of swirl and vortices which reduce the pump performance and induce

vibration and additional noise Therefore sump model test is necessary to check the

flow condition around intake structure Numerical simulation is a good facility for

reducing the time and cost involved throughout the design process In this study the

commercial software ANSYS CFX-130 has been used for the CFD analysis of the

sump models

In a scaled sump model for air entrainment simulation numerical analysis of single

phase and two-phase with SST turbulence model was carried out to predict vortex

(both free surface vortex and submerged vortex) occurrence and location The

effectiveness of curtain walls and square bars to eliminate the free surface vortex is

evaluated Meanwhile the experiment including PIV system was performed in

KMOU to investigate the flow conditions around pump intake structure

In a full size sump model overall numerical analysis for the sump model with a

mixed flow pump installed was carried out Hydraulic performances of the mixed

flow pump for head rise shaft power pump efficiencies versus flow rate changed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

vii

from 50 to 140 of the design flow rate were studied by the performance curves

A trident shaped anti vortex device (AVD) composed of three wall fillets and one

center splitter was installed under the pump intake The effectiveness of the AVD

for the suppression of submerged vortex was evaluated In addition numerical

simulation of cavitation phenomenon in the mixed flow pump has been performed

by calculating the full cavitation model with k-ε turbulence model

According to the results details of the location size and strength of vortices were

predicted in the numerical simulation For the scaled sump model although there is

a little difference between the single phase and two-phase simulation all the results

predict the free surface vortex and submerged vortex formation and location The

CFD simulations of flow condition show good agreement with the PIV results The

effectiveness of curtain walls installed to inhibit the free surface vortex is confirmed

in the CFD results while the square bars installed in the channel failed to suppress

the free surface vortex For the mixed flow pump with sump the best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

The effectiveness of AVDs installed for the submerged vortex is confirmed by

comparisons of the sump model with and without AVD results

Cavitation phenomenon analysis shows the pump operating at the BEP condition

has a considerable extent over the requirement of cavitation performance With

numerical simulation the inception of cavitation in the blade passage is observed on

the suction surface where the leading edges meet the tips and then as the inlet total

pressure decrease the cavitation zone is spread out over the suction surface as well

as the leading edge of the impeller blade

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

viii

Nomenclature

B Distance from the back wall to the pump inlet bell centerline [m]

C Distance between the inlet bell and floor [m]

D Inlet bell design outside diameter [m]

Fr Froude number (=Vb(gD)05) [-]

g Acceleration due to gravity [ms2]

ht Total head [m]

n Number of revolution per min [-]

Pflow Flow power [W]

Pshaft Shaft power [W]

Pν Vapor pressure [Pa]

Q Volume flow rate [m3s]

Q0 Design volume flow rate [m3s]

Re Reynolds number (=VbDν) [-]

S Minimum submergence depth [m]

Vb Velocity at bell mouth [ms]

α Angle of floor slope [deg]

η Pump efficiency []

θ Swirl angle [deg]

λ The scale ratio of model dimensions to prototype [-]

μ Viscosity [Pa s]

ν Kinematic viscosity [m2s]

ρ Specific water density [kgm3]

σ Surface tension [Nm]

τ Stress tensor [-]

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

ix

Abbreviations

AVD Anti vortex device

BEP Best efficiency point

CFD Computational fluid Dynamics

NPSH Net positive suction head

NPSHR Required net positive suction head

PIV Particle Image Velocimetry

RANS Reynolds Averaged Navier- Stokes

RPM Number of revolutions per minute

SST Shear Stress Transport

VOF Volume of Fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

1

Chapter 1 Introduction

11 Background

It is an accepted fact that quite a number of problems faced in a pumping station are

related to the design of sump or intake rather than pump design As an important

part of the pumping station sump is designed to provide uniform swirl and

entrained air free flow to the pump intake Undesirable flow conditions (entrained

air non-uniform flow distribution vortices etc) can reduce the pump efficiency

induce vibration noise and cavitation even lead to excessive bearing loads of

impeller and structure damage Due to quite complex nature of the flow although

the flow conditions causing pump problems are well established there are no

specific solutions to eliminate them There are some design guidelines for the

specific geometrical and hydraulic constraints of the pump sump for any project

the model study is the only tool for solving potential problems in new designs and

rectifying problems observed in existing installations However the scaled model

tests are expensive and time-consuming so alternative Computational Fluid

Dynamics (CFD) methods for evaluating sump performance have been developed

With rapid progress in CFD numerical simulation is regarded as an effective tool in

solving fluid problems in pump sumps

12 Previous study

Various researches show the flow conditions around the sump intake structure are

quite complicated Shyam et al [1] used the commercial code CFX to carry out

two-phase flow simulation to capture air entrainment The air entrainments and its

location were well captured and numerical results were in accordance with the

experimental investigation Matsui J[2] and Okamura T[3] et al conducted a serial

research of the benchmark according to the Turbomachinery Society of Japan (TSJ)

standard including various commercial codes comparison performed in the flow

simulation and comparison analysis between the numerical and experimental results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

2

The purpose of the benchmark is to determine the accuracy and reliability of the

CFD codes used in universities and the industrial field It is also intended to identify

and investigate the formation of air-entrained and submerged vortices near the

pump Tanweer et al[4] have performed a numerical analysis of a multiple intake

pump sump to confirm that the geometry of sump affects the flow within sump

greatly Rajendran et al[5] have made a numerical analysis for the flow

characteristics of a sump model with pump intake and good agreements were

achieved by comparing the numerical results with the experiments Iwano et al[6]

have introduced a numerical method for the submerged vortex by analyzing the

flow in the pump sump with and without baffle plates Lee et al[7] have conducted

the CFD analysis of a multi-intake pump sump model to check the flow uniformity

by predicting the location number and vorticity of the vortex For the CFD

application in mixed flow pump design and performance analysis Oh H W [8]

conducted a mixed flow pump design optimization with specific parameters and

then compared the numerical results with the experiments The procedure presented

can be used as a practical design and analysis guide for the general turbomachinery

Kim et al[9] carried out a mixed flow pump performance analysis with numerical

simulation and improved the hydraulic efficiency by the modification of the

geometry

As an undesirable phenomenon caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage The cavitation

phenomenon has been studied by various investigators Van et al[10] have

investigated the cavitation inception behavior of a mixed-flow pump impellers

model and the calculated results matched well with the experiments Okamura et

al[11] have evaluated two numerical cavitation prediction methods used in the

pump industry manufacturers and attempts were made to improve the cavitation

performance Li et al[12] have conducted the numerical analysis on the basis of the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

3

development liquidvapor interface tracking method to predict the cavitation

characteristics with a centrifugal pump impeller model

Moreover there are a number of correlated studies [13]~[17]

13 Methodology of study

This study will investigate the flow phenomena of two sump models

A scaled sump model is adopted to carry out numerical simulation and PIV

experiment The flow conditions are predicted and the effectiveness of devices for

free surface vortex is evaluated by using numerical code a single intake sump

model with a mixed flow pump installed is modeled and the hydraulic performance

characteristics for the head rise shaft power and pump efficiencies versus flow rate

are studied In addition flow phenomena (vortices cavitation etc) are analyzed

and the effectiveness of an anti-vortex device (AVD) for the submerged vortex is

evaluated

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

4

Chapter 2 Basic theory of sump model

Specific hydraulic phenomena have been confirmed that they can lead to certain

common operational problems The common of the main problems are summarized

as follows

a Free surface vortices

b Submerged vortices

c Swirl of flow entering the pump

d Non-uniform distribution of velocity at the pump impeller

e Entrained air or gas bubbles

These problems encountered in the pump sump will affect the pump

performance and significantly increase the operational and maintenance costs In

order to identify sources of particular problems and find practical solution for

it the usual approach is to conduct the laboratory experiments on a scaled

physical model Typical design objectives are to ensure that a pump station is

designed according to best practice and conforms to the requirements set out in the

American National Standard for Pump Intake Design[18] These standards restrict

the degree to which the above-mentioned undesirable flow patterns may be present

The negative impact of each of these hydraulic conditions varies with pump specific

speed and size Application of the standard to design a sump does not generate a

problem free sump but provides a basis for the initial design

As there are no specific guidelines or criteria for design of trouble free intakes the

most common solution to potential problems in new design and rectification of

problems observed in existing designs is to construct a scaled model in a laboratory

observe and investigate the flow characteristics and propose modifications to the

intake geometry There are additional devices in the form of floor splitters or cones

back wall splitters fillets surface beams guide vanes etc aimed at controlling the

vortex and swirl formation Considerable prior experience and ingenuity are

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

5

required to solve the problems by this method and the cost and time required are

also significant

The intake structure refers to the channel leading from the water source which

may be a river or reservoir and all installations downstream including the pump

column or the intake pipe (the suction tube portion of a vertical intake) the

approach channel (upstream of pump bay) and the pump bay (bounded by the floor

the back wall side walls dividers walls separating adjacent pump column) Usually

the upstream end of the pump column has an inlet attachment called the suction bell

The function of the intake is to supply an evenly distributed flow of water to the

pump suction bell

Several types of pump intake basins are available and ANSIHI 98 provides

design requirements for each Intake structures can be categorized as being

clear liquids or solids-bearing liquids For clear liquids intakes are further

classified into rectangular formed circular and trench types as well as suction

tanks and cans For solid-bearing liquids trench type and rectangular wet wells are

usually considered

In this paper only rectangular intake structure for clear liquid is studied

21 Pump intake design

For pumps to achieve their optimum hydraulic performance across all operating

conditions the flow at the impeller must meet specific hydraulic conditions The

ideal flow entering the pump inlet should be of a uniform velocity distribution

without rotation and stable over time When designing a sump to achieve favorable

inflow to the pump or suction pipe bell various sump dimensions relative to the size

of the bell are required Using these standards of the distances to reduce the

probability of the occurence of strong submerged vortices and free-surface vortices

211 Inlet bell diameter design

The geometry is generally defined in terms of the pump inlet diameter as shown in

Table 21 and Fig 21 The bell diameter can be estimated based on an inlet pipe

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

6

Table 21 Acceptable velocity ranges for inlet bell diameter(ANSIHI 98)

Flow Rate

Qls

Recommended Inlet Bell Design

Velocityms

Acceptable Velocity

Range ms

Q<315 V=17 06leVle27

315 leQ<1260 V=17 09leVle24

Q ge1260 V=17 12leVle21

Fig 21 Recommended inlet bell design diameter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

7

212 Recommended dimensions for a rectangular sump

As the selected bell diameter has been determined the proportions of the inlet

structure can be estimated from Fig 22 and Fig 23 Table 22 shows the

recommended dimensions for a rectangular sump

Fig 22 Recommended intake structure layout

Fig 23 Filler wall details for proper bay width

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

8

Table 22 Recommended dimensions for a rectangular sump

Dimension Variable

Description Recommended value

A Distance from the pump inlet bell

centerline to the entrance A=5D minimum

a Length of constricted bay section near

the pump inlet 25D minimum

B Distance from the back wall to the

pump inlet bell centerline 075D

C Distance between the inlet bell and

floor 03D~05D

D Inlet bell outside diameter refer to section 211

H Minimum liquid depth H=S+C

h Minimum height of constricted bay

section near the pump inlet bell hgtH or 25D

S Minimum pump inlet bell submergence S=D(10+23Fr)

W Pump inlet bay entrance width 2D minimum

w Constricted bay width near the pump

inlet bell w=2D

X Pump inlet bay length 5D minimum

Y Distance from pump inlet bell

centerline to the through-flow traveling screen

4D minimum

Z1 Distance from sump inlet bell centerline to diverging walls

5D minimum

Z2 Distance from inlet bell centerline to

sloping floor 5D minimum

α Angle of floor slope -10deg~ +10deg

β Angle of wall convergence 0deg~ +10deg

empty Angle of convergence from constricted

area to by walls 10deg maximum

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

9

FD=Vb(gD)05 (21)

where

FD Froude number (dimensionless)

Vb velocity at suction inlet=FlowArea based on D

D Outside diameter of bell or pipe inlet

g gravitational acceleration

There is the design sequence summary for rectangular intake structures

middot Consider the flow patterns and boundary geometry of the body of liquid

from which the pump station is to receive flow Compare with the approach

flow condition and determine if a hydraulic physical model study is

required

middot Determine the number and size of pumps required to satisfy the range of

operating conditions likely to be encountered

middot Identify pump inlet bell diameter Refer to the section 211

middot Determine the bell-floor clearance 05D is a good preliminary design value

middot Determine the required bell submergence using the equation 21

middot Determine the minimum allowable liquid depth in the intake structure from

the sum of the floor clearance and the required bell submergence

middot Check bottom elevation near the entrance to the structure and determine if it

is necessary to slope the floor upstream of the bay entrance If the resulting

depth at the entrance to the intake structure is shallow then check to ensure

that gravity-driven flow is not restricted by the entrance condition

middot Check the pump bay velocity for the maximum single-pump flow and

minimum liquid depth with the bay width set to 2D If bay velocity exceeds

05 ms then increase the bay width to reduce the maximum velocity below

05 ms

middot If it is necessary to increase the pump bay width to greater than 2D then

decrease bay width in the vicinity of the pumps according to Fig 23

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

10

middot Compare cross-flow velocity (at maximum system flow) to average pump

bay velocity If cross-flow value exceeds 50 of the bay velocity a

physical hydraulic model study is necessary

middot Determine the length of the structure and dividing walls giving

consideration to minimum allowable distances to a sloping floor screening

equipment and length of dividing walls if dual flow traveling screens or

drum screens are to be used a physical hydraulic model study is required

middot If the final selected pump bell diameter and inlet velocity is within the

range given in section 211 then the sump dimensions (developed based on

the inlet bell design diameter) need not be changed and will comply with

these standards

22 Model test

The approach flow pattern in a pump intake facility is difficult to predict by

traditional mathematics or empirical formulae For large structures or those that

differ significantly from proven designs a model study is the only means to ensure

success A model study is used to identify adverse hydraulic conditions and

derive remedial measures for approach flow patterns generated by structures

upstream of the pump impeller Current model studies are not intended to

investigate flow patterns induced by the pump itself or the flow patterns within the

pump The objective of a model study is to ensure that the pump intake structure

generates favorable flow conditions at the inlet to the pump Evaluation for the

requirement of model test if

a Sump or piping geometry that deviates from this design standard

b Non-uniform or non-symmetric approach flow to the pump sump

exist

c The pumps have flows greater than 2520 ls per pump or the total

station flow with all pumps running would be greater than 6310 ls

d Situations that need ten times cost than the model study

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

11

Adverse hydraulic conditions that can affect pump performance include free and

submerged vortices swirl approaching the pump impeller flow separation at the

pump bell and a non-uniform axial velocity distribution at the suction

The purpose of model test is to examine these adverse hydraulic conditions and to

investigate measure against them

221 The scale effects

One possible difficulty in scale modeling is that it is impossible to reduce all

pertinent forces by the same factor With Froude-scale models the inertial and

gravitational forces are reduced similarly but the viscous and surface tension forces

cannot be simultaneously reduced The extra influence of these forces is known as

a ldquoscale effectrdquo A number of investigations have indicated that the scale effects for

the model are avoided if the Reynolds number based on the inlet flow and

submergence or intake diameter exceeds approximately 3x 104 and the Weber

number exceeds 150 [19]

In most engineering applications involving closed-conduit and open-channel flow

the Reynolds number limits are far exceeded and the flows are fully turbulent

In a sump model Froude number (Fr) and Reynolds number (Re) are the most

important non-dimensional parameters Some reduction in Re should be

introduced to the sump model However flow patterns are generally very similar

at high Re

No specific geometric scale rate is recommended but reasonable large

geometric scale to minimize viscous and surface tension scale effects and reproduce

the flow pattern

Re= uDν (22)

We= u2ρDσ (23)

where

u average axial velocity (ms)

ν kinematic viscosity of the liquid (m2s)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

12

ρ liquid density (kgm3)

σ surface tension of liquidair interface

According to the ANSIHI 98 the viscous scale effects on vortex may be negligible

if the resulting dimensionless numbers must meet these values Re>6times104 and

We>240

To allow practical visual observations of flow patterns accurate measurements of

swirl and velocity distribution and sufficient dimensional control the model scale

shall yield a bay width of at least 300mm a minimum liquid depth of at least

150mm and a pump throat or suction diameter of at least 80mm

222 Principle of similarity

In the similitude analysis the geometric and flow similarity requires the model and

the flow to be identical as the real model and flow This is to ensure the result

obtained from the model study is well presented in order to predict full scale

behavior Model involving a free surface are operated using Froude similarity since

the flow process is controlled by gravity and inertial forces

For similarity of the flow patterns the Froude number shall be equal in model and

prototype

Fr =

=

(24)

and as the scale ratio λ was selected in section 221 the following equation is

obtained

V = V times (

) = V times λ or Q = Q times (

) = Q times λ (25)

where Fr is the Froude number λ is the scale ratio of model dimensions to prototype

The test of 15 times Froude scaled flow should be conducted to ascertain the

potential scale affect on free surface vortices

Models of closed conduit piping system leading to a pump suction are not operated

based on Froude similitude However flow patterns are generally very similar at

high Re A minimum value of 1times105 for the Re at the pump suction is

recommended to ensure the flow patterns are correctly scaled

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

13

223 Preliminary model test

The initial design shall be tested first to identify any hydraulic problems

1) Vortex Formation Mechanisms

There are various investigations that provide insight into the fundamental processes

leading to the development of vortices both in experiment and numerical simulation

Shin et al [20] demonstrated that two basic mechanisms lead to inlet vortex

formation The first mechanism involves the development of an inlet vortex due to

the amplification of ambient vorticity in the approach flow as vortex lines

are convicted into the inlet The second mechanism involves the development of a

trailing vortex in the vicinity of the intake as a result of the variation in

circulation along the inlet For this second case a vortex can develop in a flow that

is irrotational upstream and the vortex development therefore does not depend

on the presence of ambient vorticity Shin et al investigation on kinematic

parameters indicate that the strength of an inlet-vortex or trailing vortex system

increases with decreasing distance from the surface However for an inlet in

an upstream irrotational flow two counter rotating vortices can still trail from

the rear of the inlet

2) Classification of vortex type

As illustrated in Fig 24 vortices in the vicinity of pump intakes may be adjacent to

the channel bottom or a channel wall (ie submerged vortices) or they may appear

adjacent to the free surface (ie free surface vortex)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

14

(a) Schematic diagram of vortices in a sump

(b) Experimental results of different vortex

Fig 24 Typical vortex in a sump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

15

For the model visual observation dye was injected around the intake structure near

the surface and submergence as Fig 24 (b) shows And the model test results of

vortex was classified as Fig 25

(a) Free Surface Vortices

(b) Submerged Vortices

Fig 25 Vortex classifications

There are a number of researches focused on the numerical theory to detect the

inception of a visible vortex As the diameters of vortices are usually much smaller

compared with the computing grid size to compensate for the lack of vortex

resolution in numerical simulation a stretching vortex model is applied to the local

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

16

flow field around semi-analytically identified vortex positions [21][22] The brief

explanations are as follows

a) Free surface vortex

The criteria expressed by equation 26 and 27 is applied to determine the

visible inception of a free surface vortex

≫ 1 (26)

=

gt 1 (27)

where P is the pressure drop at the vortex core and Ph is the static head at the

vortex element and f is the drag force acting at a unit length of water surface

vortex dimple based on Stokes approximation and caused by the down flow

Further f is the force acting on the bubble in the direction from the high

pressure region to the low pressure region like buoyancy

b) Submerged vortex

Equation 28 is applied to determine the visible inception of a submerged vortex

infin =

infin

infin gt 1 (28)

where is the static pressure at the vortex core infin is the atmospheric pressure

and is the critical cavitation inception pressure and nearly equal to the

saturated vapor pressure

3) Swirl angle

Swirl angle predicts the intensity of flow rotation Fig 26 shows experiment

calculation method with swirl meter and attempts of the CFD calculation method

were conducted as Fig 27 A four blade zero-pitch swirl meter which is supported

by a low friction bearing will be installed at about four suction pipe diameter (d)

downstream from the pump suction to measure swirl angle of pump approaching

flow For the swirl meter the tip to tip blade diameter is 075d and the length in

flow direction is 06 and one of the four blade is painted yellow as a reference to

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

17

count revolutions As flow swirl is usually unsteady the observation time of swirl

meter reading should be a continuous period of time

Fig 26 Typical swirl meter

(29)

(210)

(211)

Where qV flow rotational speed at swirl meter (ms)

zV average axial velocity at swirl meter (ms)

n Number of revolution per minute (rpm)

d Diameter of pipe at the swirl meter (m)

q flow rate (m3s)

zV

Vqq 1tan -=

)(60

smnd

Vtimestimes

=p

q

)(4

2

smd

qVz

times=

p

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

18

For the swirl angle calculating method with CFD the key point is to obtain the

average tangential velocity so the swirl check circles were created as illustrated in

Fig 27 The swirl check circle is located at the section of 4d height with diameter

of 025d 05d and 075d respectively

Swirl check circle

Fig 27 Schematic view of swirl angle numerical calculation

23

1 7501 501 250

acute++

=aringaringaring

N

i

N

i

NNVNVNV

Vi

q (212)

where

qV average tangential velocity (ms)

V circumferential velocity of check point in check circle

respectively (ms)

N check point total numbers in single check circle

4) Velocity

Velocity measurement at bell throat by Pitot tube was implemented by ANSIHI

method Typical Pitot tube configuration is shown in Fig 28 Eight measuring

points are defined as Fig 29 Time averaged velocity at eight points was obtained

and these eight data were again averaged to give throat sectional spatial velocity

average

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

19

Fig 28 Example of Pitot tube Fig 29 Measuring points at bell throat

224 Test criteria

There is the criteria for satisfactory performance of the sump model test and

summarized in Table 23

1) Free surface and submerged vortices

Free surface vortices entering the pump must be less severe than vortices with

coherent core Dye core vortices may be acceptable only if they occur less than 10

of the time or only for infrequent pump operating conditions

Submerged vortex which starts from the sump walls induce vibration and noise

problems Dye core submerged vortices extending into the pump bell mouth is not

allowed

2) Swirl angles

Swirl angle is used to check the flow rotation in the suction pipe The calculation

method is referred to section 223

Short-term maximum and long-term average values must be less than 5 degrees

Maximum short-term swirl angles up to 7 degrees may be acceptable only if they

occur less than 10 of the time or only for infrequent pump operating conditions

3)Velocity

Each point time-averaged velocity shall not be larger than 10 of sectional average

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

20

Standard deviation of time-varying velocity at each point was obtained and its

value shall not be larger than 10 of the time-averaged velocity at each point

1) Time-averaged

Variation[] = 100acute-

a

a

V

VV aV Averaged at section

2) Time-varying

Standard deviation= )1(

)(1i

2

-

-aring=

n

xxn

i

Variation[] = (Standard deviation X ) x 100

X Averaged at point

Table 23 Permission criteria of ANSIHI

Vortex occurence Swirl angle

Free surface vortex Submerged vortex

Observation

time Above 10 minutes Above 10 minutes

Above 10

minutes

Permission

criteria

Dye-core

Vortex(F3)

below 110 total

time

Dye-core

Vortex(S2)

below 110 total

time

Below

5degConditional

intermittent

below 7deg

225 Remedial measures for problem intakes

Usually in the preliminary test the worst flow condition is simulated And based on

the results remedial measures for problem intakes are discussed in this part

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

21

1) Overall approach flow patterns

The characteristics of flow approaching an intake structure is one of the foremost

considerations for the designer It is difficult to predict the effects of a given set of

flow conditions upstream from an intake structure Fig 210 depicts several typical

approach flow patterns of pump sump with different operating conditions

Fig 210 Examples of approach flow patterns

Generally speaking for the intake problems especially for the multiple pumps

sump there are the main measures using dividing walls placed between the pumps

as a partitioned structure trash racks with elongated bars guide piers or a number

of structural columns at the bay entrance can provide some assistance in controlling

cross-flow baffles flow distributors and screen mesh etc are applied to improve

the flow distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

22

2) Problems around pump intake

a) Free surface vortices

There are a number of factors that have an effect on the formation of surface

vortices To achieve a higher degree of certainty that objectionable surface vortices

do not form modifications can be made to intake structure to allow operation at

practical depths of submergence and the minimum submergence is referred in Table

22 the use of suction umbrellas vertical curtain walls and horizontal gratings

b) Submerged vortices

In the vicinity of intake bell there are the most complicated flow patterns and flow

makes the most changes in direction while maintaining a constant acceleration into

the pump bell to prevent local flow separation turbulence and vortex formation

Fig 211 represents a sample of various devices to inhibit the different submerged

vortex

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

23

Fig 211 Methods to reduce submerged vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

24

c) Preswirl

The approach flow distributions determine whether flow preswirl exists and a

sufficiently laterally skewed approach flow cause rotating around the intake bell

The most effective way of reducing preswirl is to establish a relatively uniform

approach flow within each pump bay by using the baffling schemes controlling

cross-flow and expanding concentrated flow discussed in the above part

To prevent non-uniform velocity distribution swirls and vortices necessary

modifications are conducted such as fore bay length baffle wall distributors

curtain wall mesh screen guide vane floor splitter wall fillet etc As the above

measures applied modification test will be conducted to check all the flow

conditions and compare with the preliminary results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

25

Chapter 3 Methodology

CFD (Computational fluid dynamics) is the technique which can design cycles to

better performance and reduce costs and time In CFD we solve the governing

equations of given physics (may be differential form or integral form) using some

numerical techniques like Finite Difference Method (FDM) Finite Element Method

(FEM) or Finite Volume Method (FVM) Although CFD plays a great part in the

design of turbomachinery CFD is not a replacement for experimental or analytical

approach In this study the numerical analysis was performed by the commercial

code ANSYS CFX 130 All aspects of the CFD study including modelling

meshing solving and post-processing of the results

31 Numerical modeling

ANSYS CFX solves the unsteady Navier-Stokes equations in their conservation

form The instantaneous equation of mass (continuity) in the stationary frame is

expressed as equation

ρ

δ + nabla ∙ (ρU) = 0 (31)

And the instantaneous equation for momentum is expressed as shown in equation

ρ

δ + nabla ∙ (ρU⨂U) = minusnablap + nabla ∙ τ + S (32)

Where τ is the stress tensor ( including both normal and shear components of the

stress)

These instantaneous equations are averaged for turbulent flows leading to additional

terms that need to be solved While the Navier-Stokes equations describe both

laminar and turbulent flows without addition terms realistic flows involve length

scales much smaller than the smallest finite volume mesh A Direct Numerical

Simulation of these flows would require significantly more computing power than

what is available now or in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

26

Therefore much research has been done to predict the effects of turbulence by

using turbulence models These models account for the effects of turbulence

without the use of a very fine mesh or direct numerical simulation

These turbulence models modify the transport equations by adding averaged and

fluctuating components The transport equations are changed to equations (33) and

(34)

ρ

δ + nabla ∙ (ρU) = 0 (33)

ρ

+ nabla ∙ ρ U ⨂U = minusnablap + nabla ∙ τ minus ρu⨂u + S (34)

The mass equation is not changed but the momentum equation contains extra terms

which are the Reynolds stresses ρ u⨂u and the Reynolds flux ρ u ϕ These

Reynolds stresses need to be modeled by additional equations to obtain closure

Obtaining closure implies that there are a sufficient number of equations to solve for

all the unknowns including the Reynolds stresses and Reynolds fluxes

Various turbulence models provide various ways to obtain closure In this

investigation Shear Stress Transport (SST) model and k-ε model were utilized The

advantage of SST model is that combines the advantages of other turbulence models

(the k-ε Wilcox k-ω) The 3 turbulence models will be discussed briefly

32 Turbulence models

Turbulence models are necessary because we cannot afford big enough computers

to directly capture every scale of motion There are three modelling frameworks

Direct Numerical Simulation (DNS) Reynolds-Averaged Navier-Stokes

Equations(RANS) Large Eddy Simulation(LES)

As RANS method is applied in most productionseveral RANS turbulence models

differ in their usage of wall functions the number of additional variables solved for

and what these variables represent

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

27

Two equation turbulence models

Two-equation turbulence models are widely used in CFD as they give a good

compromise between computational power needed and accuracy The term lsquotwo

equationrsquo refers to the fact that these models solve for the velocity and length scales

using separate transport equations The turbulence velocity scale is obtained by

solving the transport equation The turbulent length scale is estimated from two

properties of the turbulence field namely the turbulent kinetic energy and the

dissipation rate The dissipation rate of the kinetic energy is obtained from its

transport equation The most widely used are k-ε and k-ω two equation models In

the next section the k-ε Wilcox k-ω BSL k-ω and SST models will be briefly

discussed

321 k-ε turbulence model

The k-epsilon model solves for two variables k the turbulent kinetic energy and

epsilon the rate of dissipation of kinetic energy Wall functions are used in this

model so the flow in the buffer region is not simulated The k-epsilon model is very

popular for industrial applications due to its good convergence rate and relatively

low memory requirements It does not very accurately compute flow fields that

exhibit adverse pressure gradients strong curvature to the flow or jet flow It does

perform well for external flow problems around complex geometries

The k-ε model introduces k (m2s2) as the turbulence kinetic energy and ε (m2s3) as

the turbulence eddy dissipation The continuity equation remains the same

ρ

+ nabla ∙ (ρ U) = 0 (35)

The momentum equation changes as shown by equation

δρ

δ + nabla ∙ (ρU⨂U) = minusnablapprime + nabla ∙ μ

nablaU + (nablaU) + S (36)

Where SM is the sum of body forces μeff is the effective viscosity accounting for

turbulence and prsquo is the modified pressure The k-ε model uses the concept of eddy

viscosity giving the equation for effective viscosity as shown by equation

μ

= μ + μ (37)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

28

μt is the turbulence viscosity is linked to the turbulence kinetic energy and

dissipation by the equation

μ = Cμρ

ε (38)

Where Cμ is a constant

The values for k and ε come from the differential transport equations for the

turbulence kinetic energy and the turbulence dissipation rate

The turbulence kinetic energy equation is given as equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus ρε (39)

The turbulence dissipation rate is given by equation

(ρε)

+ nabla ∙ (ρUε) = nabla ∙ μ +

μ

σε nablaε +

ε

(Cε (P + Pε ) minus Cε ρε) (310)

Where Cε1 Cε2 σk σε are constants

Pk is the turbulence production due to viscous forces and is modeled by the

equation

P = μ nablaU ∙ nablaU + nablaU minus

nabla ∙ U 3μ

nabla ∙ U + ρk (311)

If a buoyancy term is added to the previous equation if the full buoyancy model is

used

The buoyancy term Pkb is modeled as

P = minusμ

ρσρg ∙ nablaρ (312)

322 Wilcox k-ω turbulence model

The k-omega model is similar to k-epsilon instead however it solves for omega mdash

the specific rate of dissipation of kinetic energy It also uses wall functions and

therefore has comparable memory requirements It has more difficulty converging

and is quite sensitive to the initial guess at the solution Hence the k-epsilon model

is often used first to find an initial condition for solving the k-omega model The k-

omega model is useful in many cases where the k-epsilon model is not accurate

such as internal flows flows that exhibit strong curvature separated flows and jets

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

29

This model has an advantage over the k-ε model where it does not involve complex

linear damping functions for near wall calculations at low Reynolds numbers

The k-ω model assumes that the turbulence viscosity is related to the turbulence

kinetic energy k and the turbulent frequency ω by the equation

μ = ρ

ω (313)

The transport equation for k is given by the equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus βprimeρ k ω (314)

The transport equation for ω is shown as equation

(ρω)

δ + nabla ∙ (ρ U ω) = nabla ∙ μ +

μ

μω

nablaω + αω

P + Pω minus β ρ ω (315)

The production rate of turbulence (Pk) is calculated as shown previously in the k-ε

section

The recommended values for the constants in the above equations are

βprime = 009

α = 59

β = 0075

σ = 2

σω = 2

The Reynolds stress tensor ρ u⨂u is calculated by

minusρ u⨂u = μ nablaU + (nablaU) minus

part ρk + μ

nabla ∙ U (316)

323 Shear stress transport model

The disadvantage of the Wilcox model is the strong sensitivity to free-stream

conditions Therefore a blending of the k-ω model near the surface and the k-ε in

the outer region was made by Menter [23] which resulted in the formulation of the

BSL k-ω turbulence model It consists of a transformation of the k-ε model to a k-ω

formulation and subsequently adding the resulting equations The Wilcox model is

multiplied by a blending function F1 and the transformed k-ε by another function 1-

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

30

F1 F1 is a function of wall distance (being the value of one near the surface and zero

outside the boundary layer) Outside the boundary and on the edge of the boundary

layer the standard k-ε model is used [24]

SST model is a combination of the k-epsilon in the free stream and the k-omega

models near the walls It does not use wall functions and tends to be most accurate

when solving the flow near the wall The SST model does not always converge to

the solution quickly so the k-epsilon or k-omega models are often solved first to

give good initial conditions

The k-ω based Shear Stress Transport (SST) model of Menter was applied for

turbulence treatment The transport equations for the SST model are expressed

below where the turbulent kinetic energy lsquokrsquo and turbulent frequency or dissipation

per unit turbulent kinetic energy are computed by using the following relations

For Turbulence Kinetic Energy

kkPukt

kTkk Ntilde+Ntilde+-=Ntilde+

para

para)[()( nsnwb (317)

where Pk is the production limiter

For Specific Dissipation Rate

ww

swnsnbwaww

ww NtildeNtilde-+Ntilde+Ntilde+-=Ntilde+para

para

1)1(2])[()( 21

22 kFSut

T (318)

The first blending function F1 is calculated from

iumlthorn

iumlyacute

uuml

iumlicirc

iumliacute

igrave

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

uacuteuacuteucirc

ugrave

ecircecirceuml

eacute

dividedivideoslash

oumlccedilccedilegrave

aelig=

4

22

21

4500maxmintanh

yCD

k

yy

kF

kw

wrs

w

n

wb (319)

y is the distance to the nearest wall and v is the kinematic viscosity

divideoslash

oumlccedilegrave

aeligNtildeNtilde= -10

2 101

2max ww

rsww kCDk (320)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

31

And Kinematic eddy viscosity

)max( 21

1

SF

kT

wa

an = (321)

where F2 is a blending function which restricts the limiter to the wall boundary and

S is the invariant measure of the strain rate

uacuteuacute

ucirc

ugrave

ecircecirc

euml

eacute

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

dividedivideoslash

oumlccedilccedilegrave

aelig=

2

22

5002maxtanh

w

n

wb yy

kF (322)

33 Brief introduction of simulation process

All aspects of CFD study including modelling meshing solving and post-

processing of the results are briefly introduced in this chapter

331 Pre-processing technique

The speed and accuracy of simulation results depending fundamentally on meshing

and the ability to prepare model geometry correctly

The steps in pre-processing are explained as follows

1) Domain Definitions

The domain definition depends mainly on the physics involved in the problem and

objective of the problem

a) Geometry creation or geometry import

b) Geometry cleanup

i) Removing the parts not necessary for simulation under consideration

ii) Closing the gaps in the geometry

iii) Removing small surfaces curves to make meshing process simple

2)Grid generation

To solve the governing equations the numerical methods like Finite Difference

Finite Volume or Finite Element Method to divide the domain into small parts and

solve the equations on one part at a time are applied

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 9: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

vi

Study of Flow Field in Rectangular Sump Models and

Performance Analysis of a Mixed Flow Pump

Yuxin ZHAO

Department of Mechanical Engineering

Graduate School of Korea Maritime and Ocean University

Abstract

It is an accepted fact that quite a number of problems faced in a pumping station are

related to the design of sump or intake rather than pump design Head-capacity

curves provided by the pump manufacturer are obtained on the condition of no

vortices flowing into the pump intake The efficiency and performance of pumping

stations depend not only on the performance of the selected pumps but also on the

proper design of the intake sumps A faulty design of pump sump can lead to the

occurrence of swirl and vortices which reduce the pump performance and induce

vibration and additional noise Therefore sump model test is necessary to check the

flow condition around intake structure Numerical simulation is a good facility for

reducing the time and cost involved throughout the design process In this study the

commercial software ANSYS CFX-130 has been used for the CFD analysis of the

sump models

In a scaled sump model for air entrainment simulation numerical analysis of single

phase and two-phase with SST turbulence model was carried out to predict vortex

(both free surface vortex and submerged vortex) occurrence and location The

effectiveness of curtain walls and square bars to eliminate the free surface vortex is

evaluated Meanwhile the experiment including PIV system was performed in

KMOU to investigate the flow conditions around pump intake structure

In a full size sump model overall numerical analysis for the sump model with a

mixed flow pump installed was carried out Hydraulic performances of the mixed

flow pump for head rise shaft power pump efficiencies versus flow rate changed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

vii

from 50 to 140 of the design flow rate were studied by the performance curves

A trident shaped anti vortex device (AVD) composed of three wall fillets and one

center splitter was installed under the pump intake The effectiveness of the AVD

for the suppression of submerged vortex was evaluated In addition numerical

simulation of cavitation phenomenon in the mixed flow pump has been performed

by calculating the full cavitation model with k-ε turbulence model

According to the results details of the location size and strength of vortices were

predicted in the numerical simulation For the scaled sump model although there is

a little difference between the single phase and two-phase simulation all the results

predict the free surface vortex and submerged vortex formation and location The

CFD simulations of flow condition show good agreement with the PIV results The

effectiveness of curtain walls installed to inhibit the free surface vortex is confirmed

in the CFD results while the square bars installed in the channel failed to suppress

the free surface vortex For the mixed flow pump with sump the best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

The effectiveness of AVDs installed for the submerged vortex is confirmed by

comparisons of the sump model with and without AVD results

Cavitation phenomenon analysis shows the pump operating at the BEP condition

has a considerable extent over the requirement of cavitation performance With

numerical simulation the inception of cavitation in the blade passage is observed on

the suction surface where the leading edges meet the tips and then as the inlet total

pressure decrease the cavitation zone is spread out over the suction surface as well

as the leading edge of the impeller blade

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

viii

Nomenclature

B Distance from the back wall to the pump inlet bell centerline [m]

C Distance between the inlet bell and floor [m]

D Inlet bell design outside diameter [m]

Fr Froude number (=Vb(gD)05) [-]

g Acceleration due to gravity [ms2]

ht Total head [m]

n Number of revolution per min [-]

Pflow Flow power [W]

Pshaft Shaft power [W]

Pν Vapor pressure [Pa]

Q Volume flow rate [m3s]

Q0 Design volume flow rate [m3s]

Re Reynolds number (=VbDν) [-]

S Minimum submergence depth [m]

Vb Velocity at bell mouth [ms]

α Angle of floor slope [deg]

η Pump efficiency []

θ Swirl angle [deg]

λ The scale ratio of model dimensions to prototype [-]

μ Viscosity [Pa s]

ν Kinematic viscosity [m2s]

ρ Specific water density [kgm3]

σ Surface tension [Nm]

τ Stress tensor [-]

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

ix

Abbreviations

AVD Anti vortex device

BEP Best efficiency point

CFD Computational fluid Dynamics

NPSH Net positive suction head

NPSHR Required net positive suction head

PIV Particle Image Velocimetry

RANS Reynolds Averaged Navier- Stokes

RPM Number of revolutions per minute

SST Shear Stress Transport

VOF Volume of Fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

1

Chapter 1 Introduction

11 Background

It is an accepted fact that quite a number of problems faced in a pumping station are

related to the design of sump or intake rather than pump design As an important

part of the pumping station sump is designed to provide uniform swirl and

entrained air free flow to the pump intake Undesirable flow conditions (entrained

air non-uniform flow distribution vortices etc) can reduce the pump efficiency

induce vibration noise and cavitation even lead to excessive bearing loads of

impeller and structure damage Due to quite complex nature of the flow although

the flow conditions causing pump problems are well established there are no

specific solutions to eliminate them There are some design guidelines for the

specific geometrical and hydraulic constraints of the pump sump for any project

the model study is the only tool for solving potential problems in new designs and

rectifying problems observed in existing installations However the scaled model

tests are expensive and time-consuming so alternative Computational Fluid

Dynamics (CFD) methods for evaluating sump performance have been developed

With rapid progress in CFD numerical simulation is regarded as an effective tool in

solving fluid problems in pump sumps

12 Previous study

Various researches show the flow conditions around the sump intake structure are

quite complicated Shyam et al [1] used the commercial code CFX to carry out

two-phase flow simulation to capture air entrainment The air entrainments and its

location were well captured and numerical results were in accordance with the

experimental investigation Matsui J[2] and Okamura T[3] et al conducted a serial

research of the benchmark according to the Turbomachinery Society of Japan (TSJ)

standard including various commercial codes comparison performed in the flow

simulation and comparison analysis between the numerical and experimental results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

2

The purpose of the benchmark is to determine the accuracy and reliability of the

CFD codes used in universities and the industrial field It is also intended to identify

and investigate the formation of air-entrained and submerged vortices near the

pump Tanweer et al[4] have performed a numerical analysis of a multiple intake

pump sump to confirm that the geometry of sump affects the flow within sump

greatly Rajendran et al[5] have made a numerical analysis for the flow

characteristics of a sump model with pump intake and good agreements were

achieved by comparing the numerical results with the experiments Iwano et al[6]

have introduced a numerical method for the submerged vortex by analyzing the

flow in the pump sump with and without baffle plates Lee et al[7] have conducted

the CFD analysis of a multi-intake pump sump model to check the flow uniformity

by predicting the location number and vorticity of the vortex For the CFD

application in mixed flow pump design and performance analysis Oh H W [8]

conducted a mixed flow pump design optimization with specific parameters and

then compared the numerical results with the experiments The procedure presented

can be used as a practical design and analysis guide for the general turbomachinery

Kim et al[9] carried out a mixed flow pump performance analysis with numerical

simulation and improved the hydraulic efficiency by the modification of the

geometry

As an undesirable phenomenon caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage The cavitation

phenomenon has been studied by various investigators Van et al[10] have

investigated the cavitation inception behavior of a mixed-flow pump impellers

model and the calculated results matched well with the experiments Okamura et

al[11] have evaluated two numerical cavitation prediction methods used in the

pump industry manufacturers and attempts were made to improve the cavitation

performance Li et al[12] have conducted the numerical analysis on the basis of the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

3

development liquidvapor interface tracking method to predict the cavitation

characteristics with a centrifugal pump impeller model

Moreover there are a number of correlated studies [13]~[17]

13 Methodology of study

This study will investigate the flow phenomena of two sump models

A scaled sump model is adopted to carry out numerical simulation and PIV

experiment The flow conditions are predicted and the effectiveness of devices for

free surface vortex is evaluated by using numerical code a single intake sump

model with a mixed flow pump installed is modeled and the hydraulic performance

characteristics for the head rise shaft power and pump efficiencies versus flow rate

are studied In addition flow phenomena (vortices cavitation etc) are analyzed

and the effectiveness of an anti-vortex device (AVD) for the submerged vortex is

evaluated

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

4

Chapter 2 Basic theory of sump model

Specific hydraulic phenomena have been confirmed that they can lead to certain

common operational problems The common of the main problems are summarized

as follows

a Free surface vortices

b Submerged vortices

c Swirl of flow entering the pump

d Non-uniform distribution of velocity at the pump impeller

e Entrained air or gas bubbles

These problems encountered in the pump sump will affect the pump

performance and significantly increase the operational and maintenance costs In

order to identify sources of particular problems and find practical solution for

it the usual approach is to conduct the laboratory experiments on a scaled

physical model Typical design objectives are to ensure that a pump station is

designed according to best practice and conforms to the requirements set out in the

American National Standard for Pump Intake Design[18] These standards restrict

the degree to which the above-mentioned undesirable flow patterns may be present

The negative impact of each of these hydraulic conditions varies with pump specific

speed and size Application of the standard to design a sump does not generate a

problem free sump but provides a basis for the initial design

As there are no specific guidelines or criteria for design of trouble free intakes the

most common solution to potential problems in new design and rectification of

problems observed in existing designs is to construct a scaled model in a laboratory

observe and investigate the flow characteristics and propose modifications to the

intake geometry There are additional devices in the form of floor splitters or cones

back wall splitters fillets surface beams guide vanes etc aimed at controlling the

vortex and swirl formation Considerable prior experience and ingenuity are

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

5

required to solve the problems by this method and the cost and time required are

also significant

The intake structure refers to the channel leading from the water source which

may be a river or reservoir and all installations downstream including the pump

column or the intake pipe (the suction tube portion of a vertical intake) the

approach channel (upstream of pump bay) and the pump bay (bounded by the floor

the back wall side walls dividers walls separating adjacent pump column) Usually

the upstream end of the pump column has an inlet attachment called the suction bell

The function of the intake is to supply an evenly distributed flow of water to the

pump suction bell

Several types of pump intake basins are available and ANSIHI 98 provides

design requirements for each Intake structures can be categorized as being

clear liquids or solids-bearing liquids For clear liquids intakes are further

classified into rectangular formed circular and trench types as well as suction

tanks and cans For solid-bearing liquids trench type and rectangular wet wells are

usually considered

In this paper only rectangular intake structure for clear liquid is studied

21 Pump intake design

For pumps to achieve their optimum hydraulic performance across all operating

conditions the flow at the impeller must meet specific hydraulic conditions The

ideal flow entering the pump inlet should be of a uniform velocity distribution

without rotation and stable over time When designing a sump to achieve favorable

inflow to the pump or suction pipe bell various sump dimensions relative to the size

of the bell are required Using these standards of the distances to reduce the

probability of the occurence of strong submerged vortices and free-surface vortices

211 Inlet bell diameter design

The geometry is generally defined in terms of the pump inlet diameter as shown in

Table 21 and Fig 21 The bell diameter can be estimated based on an inlet pipe

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

6

Table 21 Acceptable velocity ranges for inlet bell diameter(ANSIHI 98)

Flow Rate

Qls

Recommended Inlet Bell Design

Velocityms

Acceptable Velocity

Range ms

Q<315 V=17 06leVle27

315 leQ<1260 V=17 09leVle24

Q ge1260 V=17 12leVle21

Fig 21 Recommended inlet bell design diameter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

7

212 Recommended dimensions for a rectangular sump

As the selected bell diameter has been determined the proportions of the inlet

structure can be estimated from Fig 22 and Fig 23 Table 22 shows the

recommended dimensions for a rectangular sump

Fig 22 Recommended intake structure layout

Fig 23 Filler wall details for proper bay width

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

8

Table 22 Recommended dimensions for a rectangular sump

Dimension Variable

Description Recommended value

A Distance from the pump inlet bell

centerline to the entrance A=5D minimum

a Length of constricted bay section near

the pump inlet 25D minimum

B Distance from the back wall to the

pump inlet bell centerline 075D

C Distance between the inlet bell and

floor 03D~05D

D Inlet bell outside diameter refer to section 211

H Minimum liquid depth H=S+C

h Minimum height of constricted bay

section near the pump inlet bell hgtH or 25D

S Minimum pump inlet bell submergence S=D(10+23Fr)

W Pump inlet bay entrance width 2D minimum

w Constricted bay width near the pump

inlet bell w=2D

X Pump inlet bay length 5D minimum

Y Distance from pump inlet bell

centerline to the through-flow traveling screen

4D minimum

Z1 Distance from sump inlet bell centerline to diverging walls

5D minimum

Z2 Distance from inlet bell centerline to

sloping floor 5D minimum

α Angle of floor slope -10deg~ +10deg

β Angle of wall convergence 0deg~ +10deg

empty Angle of convergence from constricted

area to by walls 10deg maximum

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

9

FD=Vb(gD)05 (21)

where

FD Froude number (dimensionless)

Vb velocity at suction inlet=FlowArea based on D

D Outside diameter of bell or pipe inlet

g gravitational acceleration

There is the design sequence summary for rectangular intake structures

middot Consider the flow patterns and boundary geometry of the body of liquid

from which the pump station is to receive flow Compare with the approach

flow condition and determine if a hydraulic physical model study is

required

middot Determine the number and size of pumps required to satisfy the range of

operating conditions likely to be encountered

middot Identify pump inlet bell diameter Refer to the section 211

middot Determine the bell-floor clearance 05D is a good preliminary design value

middot Determine the required bell submergence using the equation 21

middot Determine the minimum allowable liquid depth in the intake structure from

the sum of the floor clearance and the required bell submergence

middot Check bottom elevation near the entrance to the structure and determine if it

is necessary to slope the floor upstream of the bay entrance If the resulting

depth at the entrance to the intake structure is shallow then check to ensure

that gravity-driven flow is not restricted by the entrance condition

middot Check the pump bay velocity for the maximum single-pump flow and

minimum liquid depth with the bay width set to 2D If bay velocity exceeds

05 ms then increase the bay width to reduce the maximum velocity below

05 ms

middot If it is necessary to increase the pump bay width to greater than 2D then

decrease bay width in the vicinity of the pumps according to Fig 23

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

10

middot Compare cross-flow velocity (at maximum system flow) to average pump

bay velocity If cross-flow value exceeds 50 of the bay velocity a

physical hydraulic model study is necessary

middot Determine the length of the structure and dividing walls giving

consideration to minimum allowable distances to a sloping floor screening

equipment and length of dividing walls if dual flow traveling screens or

drum screens are to be used a physical hydraulic model study is required

middot If the final selected pump bell diameter and inlet velocity is within the

range given in section 211 then the sump dimensions (developed based on

the inlet bell design diameter) need not be changed and will comply with

these standards

22 Model test

The approach flow pattern in a pump intake facility is difficult to predict by

traditional mathematics or empirical formulae For large structures or those that

differ significantly from proven designs a model study is the only means to ensure

success A model study is used to identify adverse hydraulic conditions and

derive remedial measures for approach flow patterns generated by structures

upstream of the pump impeller Current model studies are not intended to

investigate flow patterns induced by the pump itself or the flow patterns within the

pump The objective of a model study is to ensure that the pump intake structure

generates favorable flow conditions at the inlet to the pump Evaluation for the

requirement of model test if

a Sump or piping geometry that deviates from this design standard

b Non-uniform or non-symmetric approach flow to the pump sump

exist

c The pumps have flows greater than 2520 ls per pump or the total

station flow with all pumps running would be greater than 6310 ls

d Situations that need ten times cost than the model study

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

11

Adverse hydraulic conditions that can affect pump performance include free and

submerged vortices swirl approaching the pump impeller flow separation at the

pump bell and a non-uniform axial velocity distribution at the suction

The purpose of model test is to examine these adverse hydraulic conditions and to

investigate measure against them

221 The scale effects

One possible difficulty in scale modeling is that it is impossible to reduce all

pertinent forces by the same factor With Froude-scale models the inertial and

gravitational forces are reduced similarly but the viscous and surface tension forces

cannot be simultaneously reduced The extra influence of these forces is known as

a ldquoscale effectrdquo A number of investigations have indicated that the scale effects for

the model are avoided if the Reynolds number based on the inlet flow and

submergence or intake diameter exceeds approximately 3x 104 and the Weber

number exceeds 150 [19]

In most engineering applications involving closed-conduit and open-channel flow

the Reynolds number limits are far exceeded and the flows are fully turbulent

In a sump model Froude number (Fr) and Reynolds number (Re) are the most

important non-dimensional parameters Some reduction in Re should be

introduced to the sump model However flow patterns are generally very similar

at high Re

No specific geometric scale rate is recommended but reasonable large

geometric scale to minimize viscous and surface tension scale effects and reproduce

the flow pattern

Re= uDν (22)

We= u2ρDσ (23)

where

u average axial velocity (ms)

ν kinematic viscosity of the liquid (m2s)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

12

ρ liquid density (kgm3)

σ surface tension of liquidair interface

According to the ANSIHI 98 the viscous scale effects on vortex may be negligible

if the resulting dimensionless numbers must meet these values Re>6times104 and

We>240

To allow practical visual observations of flow patterns accurate measurements of

swirl and velocity distribution and sufficient dimensional control the model scale

shall yield a bay width of at least 300mm a minimum liquid depth of at least

150mm and a pump throat or suction diameter of at least 80mm

222 Principle of similarity

In the similitude analysis the geometric and flow similarity requires the model and

the flow to be identical as the real model and flow This is to ensure the result

obtained from the model study is well presented in order to predict full scale

behavior Model involving a free surface are operated using Froude similarity since

the flow process is controlled by gravity and inertial forces

For similarity of the flow patterns the Froude number shall be equal in model and

prototype

Fr =

=

(24)

and as the scale ratio λ was selected in section 221 the following equation is

obtained

V = V times (

) = V times λ or Q = Q times (

) = Q times λ (25)

where Fr is the Froude number λ is the scale ratio of model dimensions to prototype

The test of 15 times Froude scaled flow should be conducted to ascertain the

potential scale affect on free surface vortices

Models of closed conduit piping system leading to a pump suction are not operated

based on Froude similitude However flow patterns are generally very similar at

high Re A minimum value of 1times105 for the Re at the pump suction is

recommended to ensure the flow patterns are correctly scaled

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

13

223 Preliminary model test

The initial design shall be tested first to identify any hydraulic problems

1) Vortex Formation Mechanisms

There are various investigations that provide insight into the fundamental processes

leading to the development of vortices both in experiment and numerical simulation

Shin et al [20] demonstrated that two basic mechanisms lead to inlet vortex

formation The first mechanism involves the development of an inlet vortex due to

the amplification of ambient vorticity in the approach flow as vortex lines

are convicted into the inlet The second mechanism involves the development of a

trailing vortex in the vicinity of the intake as a result of the variation in

circulation along the inlet For this second case a vortex can develop in a flow that

is irrotational upstream and the vortex development therefore does not depend

on the presence of ambient vorticity Shin et al investigation on kinematic

parameters indicate that the strength of an inlet-vortex or trailing vortex system

increases with decreasing distance from the surface However for an inlet in

an upstream irrotational flow two counter rotating vortices can still trail from

the rear of the inlet

2) Classification of vortex type

As illustrated in Fig 24 vortices in the vicinity of pump intakes may be adjacent to

the channel bottom or a channel wall (ie submerged vortices) or they may appear

adjacent to the free surface (ie free surface vortex)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

14

(a) Schematic diagram of vortices in a sump

(b) Experimental results of different vortex

Fig 24 Typical vortex in a sump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

15

For the model visual observation dye was injected around the intake structure near

the surface and submergence as Fig 24 (b) shows And the model test results of

vortex was classified as Fig 25

(a) Free Surface Vortices

(b) Submerged Vortices

Fig 25 Vortex classifications

There are a number of researches focused on the numerical theory to detect the

inception of a visible vortex As the diameters of vortices are usually much smaller

compared with the computing grid size to compensate for the lack of vortex

resolution in numerical simulation a stretching vortex model is applied to the local

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

16

flow field around semi-analytically identified vortex positions [21][22] The brief

explanations are as follows

a) Free surface vortex

The criteria expressed by equation 26 and 27 is applied to determine the

visible inception of a free surface vortex

≫ 1 (26)

=

gt 1 (27)

where P is the pressure drop at the vortex core and Ph is the static head at the

vortex element and f is the drag force acting at a unit length of water surface

vortex dimple based on Stokes approximation and caused by the down flow

Further f is the force acting on the bubble in the direction from the high

pressure region to the low pressure region like buoyancy

b) Submerged vortex

Equation 28 is applied to determine the visible inception of a submerged vortex

infin =

infin

infin gt 1 (28)

where is the static pressure at the vortex core infin is the atmospheric pressure

and is the critical cavitation inception pressure and nearly equal to the

saturated vapor pressure

3) Swirl angle

Swirl angle predicts the intensity of flow rotation Fig 26 shows experiment

calculation method with swirl meter and attempts of the CFD calculation method

were conducted as Fig 27 A four blade zero-pitch swirl meter which is supported

by a low friction bearing will be installed at about four suction pipe diameter (d)

downstream from the pump suction to measure swirl angle of pump approaching

flow For the swirl meter the tip to tip blade diameter is 075d and the length in

flow direction is 06 and one of the four blade is painted yellow as a reference to

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

17

count revolutions As flow swirl is usually unsteady the observation time of swirl

meter reading should be a continuous period of time

Fig 26 Typical swirl meter

(29)

(210)

(211)

Where qV flow rotational speed at swirl meter (ms)

zV average axial velocity at swirl meter (ms)

n Number of revolution per minute (rpm)

d Diameter of pipe at the swirl meter (m)

q flow rate (m3s)

zV

Vqq 1tan -=

)(60

smnd

Vtimestimes

=p

q

)(4

2

smd

qVz

times=

p

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

18

For the swirl angle calculating method with CFD the key point is to obtain the

average tangential velocity so the swirl check circles were created as illustrated in

Fig 27 The swirl check circle is located at the section of 4d height with diameter

of 025d 05d and 075d respectively

Swirl check circle

Fig 27 Schematic view of swirl angle numerical calculation

23

1 7501 501 250

acute++

=aringaringaring

N

i

N

i

NNVNVNV

Vi

q (212)

where

qV average tangential velocity (ms)

V circumferential velocity of check point in check circle

respectively (ms)

N check point total numbers in single check circle

4) Velocity

Velocity measurement at bell throat by Pitot tube was implemented by ANSIHI

method Typical Pitot tube configuration is shown in Fig 28 Eight measuring

points are defined as Fig 29 Time averaged velocity at eight points was obtained

and these eight data were again averaged to give throat sectional spatial velocity

average

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

19

Fig 28 Example of Pitot tube Fig 29 Measuring points at bell throat

224 Test criteria

There is the criteria for satisfactory performance of the sump model test and

summarized in Table 23

1) Free surface and submerged vortices

Free surface vortices entering the pump must be less severe than vortices with

coherent core Dye core vortices may be acceptable only if they occur less than 10

of the time or only for infrequent pump operating conditions

Submerged vortex which starts from the sump walls induce vibration and noise

problems Dye core submerged vortices extending into the pump bell mouth is not

allowed

2) Swirl angles

Swirl angle is used to check the flow rotation in the suction pipe The calculation

method is referred to section 223

Short-term maximum and long-term average values must be less than 5 degrees

Maximum short-term swirl angles up to 7 degrees may be acceptable only if they

occur less than 10 of the time or only for infrequent pump operating conditions

3)Velocity

Each point time-averaged velocity shall not be larger than 10 of sectional average

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

20

Standard deviation of time-varying velocity at each point was obtained and its

value shall not be larger than 10 of the time-averaged velocity at each point

1) Time-averaged

Variation[] = 100acute-

a

a

V

VV aV Averaged at section

2) Time-varying

Standard deviation= )1(

)(1i

2

-

-aring=

n

xxn

i

Variation[] = (Standard deviation X ) x 100

X Averaged at point

Table 23 Permission criteria of ANSIHI

Vortex occurence Swirl angle

Free surface vortex Submerged vortex

Observation

time Above 10 minutes Above 10 minutes

Above 10

minutes

Permission

criteria

Dye-core

Vortex(F3)

below 110 total

time

Dye-core

Vortex(S2)

below 110 total

time

Below

5degConditional

intermittent

below 7deg

225 Remedial measures for problem intakes

Usually in the preliminary test the worst flow condition is simulated And based on

the results remedial measures for problem intakes are discussed in this part

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

21

1) Overall approach flow patterns

The characteristics of flow approaching an intake structure is one of the foremost

considerations for the designer It is difficult to predict the effects of a given set of

flow conditions upstream from an intake structure Fig 210 depicts several typical

approach flow patterns of pump sump with different operating conditions

Fig 210 Examples of approach flow patterns

Generally speaking for the intake problems especially for the multiple pumps

sump there are the main measures using dividing walls placed between the pumps

as a partitioned structure trash racks with elongated bars guide piers or a number

of structural columns at the bay entrance can provide some assistance in controlling

cross-flow baffles flow distributors and screen mesh etc are applied to improve

the flow distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

22

2) Problems around pump intake

a) Free surface vortices

There are a number of factors that have an effect on the formation of surface

vortices To achieve a higher degree of certainty that objectionable surface vortices

do not form modifications can be made to intake structure to allow operation at

practical depths of submergence and the minimum submergence is referred in Table

22 the use of suction umbrellas vertical curtain walls and horizontal gratings

b) Submerged vortices

In the vicinity of intake bell there are the most complicated flow patterns and flow

makes the most changes in direction while maintaining a constant acceleration into

the pump bell to prevent local flow separation turbulence and vortex formation

Fig 211 represents a sample of various devices to inhibit the different submerged

vortex

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

23

Fig 211 Methods to reduce submerged vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

24

c) Preswirl

The approach flow distributions determine whether flow preswirl exists and a

sufficiently laterally skewed approach flow cause rotating around the intake bell

The most effective way of reducing preswirl is to establish a relatively uniform

approach flow within each pump bay by using the baffling schemes controlling

cross-flow and expanding concentrated flow discussed in the above part

To prevent non-uniform velocity distribution swirls and vortices necessary

modifications are conducted such as fore bay length baffle wall distributors

curtain wall mesh screen guide vane floor splitter wall fillet etc As the above

measures applied modification test will be conducted to check all the flow

conditions and compare with the preliminary results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

25

Chapter 3 Methodology

CFD (Computational fluid dynamics) is the technique which can design cycles to

better performance and reduce costs and time In CFD we solve the governing

equations of given physics (may be differential form or integral form) using some

numerical techniques like Finite Difference Method (FDM) Finite Element Method

(FEM) or Finite Volume Method (FVM) Although CFD plays a great part in the

design of turbomachinery CFD is not a replacement for experimental or analytical

approach In this study the numerical analysis was performed by the commercial

code ANSYS CFX 130 All aspects of the CFD study including modelling

meshing solving and post-processing of the results

31 Numerical modeling

ANSYS CFX solves the unsteady Navier-Stokes equations in their conservation

form The instantaneous equation of mass (continuity) in the stationary frame is

expressed as equation

ρ

δ + nabla ∙ (ρU) = 0 (31)

And the instantaneous equation for momentum is expressed as shown in equation

ρ

δ + nabla ∙ (ρU⨂U) = minusnablap + nabla ∙ τ + S (32)

Where τ is the stress tensor ( including both normal and shear components of the

stress)

These instantaneous equations are averaged for turbulent flows leading to additional

terms that need to be solved While the Navier-Stokes equations describe both

laminar and turbulent flows without addition terms realistic flows involve length

scales much smaller than the smallest finite volume mesh A Direct Numerical

Simulation of these flows would require significantly more computing power than

what is available now or in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

26

Therefore much research has been done to predict the effects of turbulence by

using turbulence models These models account for the effects of turbulence

without the use of a very fine mesh or direct numerical simulation

These turbulence models modify the transport equations by adding averaged and

fluctuating components The transport equations are changed to equations (33) and

(34)

ρ

δ + nabla ∙ (ρU) = 0 (33)

ρ

+ nabla ∙ ρ U ⨂U = minusnablap + nabla ∙ τ minus ρu⨂u + S (34)

The mass equation is not changed but the momentum equation contains extra terms

which are the Reynolds stresses ρ u⨂u and the Reynolds flux ρ u ϕ These

Reynolds stresses need to be modeled by additional equations to obtain closure

Obtaining closure implies that there are a sufficient number of equations to solve for

all the unknowns including the Reynolds stresses and Reynolds fluxes

Various turbulence models provide various ways to obtain closure In this

investigation Shear Stress Transport (SST) model and k-ε model were utilized The

advantage of SST model is that combines the advantages of other turbulence models

(the k-ε Wilcox k-ω) The 3 turbulence models will be discussed briefly

32 Turbulence models

Turbulence models are necessary because we cannot afford big enough computers

to directly capture every scale of motion There are three modelling frameworks

Direct Numerical Simulation (DNS) Reynolds-Averaged Navier-Stokes

Equations(RANS) Large Eddy Simulation(LES)

As RANS method is applied in most productionseveral RANS turbulence models

differ in their usage of wall functions the number of additional variables solved for

and what these variables represent

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

27

Two equation turbulence models

Two-equation turbulence models are widely used in CFD as they give a good

compromise between computational power needed and accuracy The term lsquotwo

equationrsquo refers to the fact that these models solve for the velocity and length scales

using separate transport equations The turbulence velocity scale is obtained by

solving the transport equation The turbulent length scale is estimated from two

properties of the turbulence field namely the turbulent kinetic energy and the

dissipation rate The dissipation rate of the kinetic energy is obtained from its

transport equation The most widely used are k-ε and k-ω two equation models In

the next section the k-ε Wilcox k-ω BSL k-ω and SST models will be briefly

discussed

321 k-ε turbulence model

The k-epsilon model solves for two variables k the turbulent kinetic energy and

epsilon the rate of dissipation of kinetic energy Wall functions are used in this

model so the flow in the buffer region is not simulated The k-epsilon model is very

popular for industrial applications due to its good convergence rate and relatively

low memory requirements It does not very accurately compute flow fields that

exhibit adverse pressure gradients strong curvature to the flow or jet flow It does

perform well for external flow problems around complex geometries

The k-ε model introduces k (m2s2) as the turbulence kinetic energy and ε (m2s3) as

the turbulence eddy dissipation The continuity equation remains the same

ρ

+ nabla ∙ (ρ U) = 0 (35)

The momentum equation changes as shown by equation

δρ

δ + nabla ∙ (ρU⨂U) = minusnablapprime + nabla ∙ μ

nablaU + (nablaU) + S (36)

Where SM is the sum of body forces μeff is the effective viscosity accounting for

turbulence and prsquo is the modified pressure The k-ε model uses the concept of eddy

viscosity giving the equation for effective viscosity as shown by equation

μ

= μ + μ (37)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

28

μt is the turbulence viscosity is linked to the turbulence kinetic energy and

dissipation by the equation

μ = Cμρ

ε (38)

Where Cμ is a constant

The values for k and ε come from the differential transport equations for the

turbulence kinetic energy and the turbulence dissipation rate

The turbulence kinetic energy equation is given as equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus ρε (39)

The turbulence dissipation rate is given by equation

(ρε)

+ nabla ∙ (ρUε) = nabla ∙ μ +

μ

σε nablaε +

ε

(Cε (P + Pε ) minus Cε ρε) (310)

Where Cε1 Cε2 σk σε are constants

Pk is the turbulence production due to viscous forces and is modeled by the

equation

P = μ nablaU ∙ nablaU + nablaU minus

nabla ∙ U 3μ

nabla ∙ U + ρk (311)

If a buoyancy term is added to the previous equation if the full buoyancy model is

used

The buoyancy term Pkb is modeled as

P = minusμ

ρσρg ∙ nablaρ (312)

322 Wilcox k-ω turbulence model

The k-omega model is similar to k-epsilon instead however it solves for omega mdash

the specific rate of dissipation of kinetic energy It also uses wall functions and

therefore has comparable memory requirements It has more difficulty converging

and is quite sensitive to the initial guess at the solution Hence the k-epsilon model

is often used first to find an initial condition for solving the k-omega model The k-

omega model is useful in many cases where the k-epsilon model is not accurate

such as internal flows flows that exhibit strong curvature separated flows and jets

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

29

This model has an advantage over the k-ε model where it does not involve complex

linear damping functions for near wall calculations at low Reynolds numbers

The k-ω model assumes that the turbulence viscosity is related to the turbulence

kinetic energy k and the turbulent frequency ω by the equation

μ = ρ

ω (313)

The transport equation for k is given by the equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus βprimeρ k ω (314)

The transport equation for ω is shown as equation

(ρω)

δ + nabla ∙ (ρ U ω) = nabla ∙ μ +

μ

μω

nablaω + αω

P + Pω minus β ρ ω (315)

The production rate of turbulence (Pk) is calculated as shown previously in the k-ε

section

The recommended values for the constants in the above equations are

βprime = 009

α = 59

β = 0075

σ = 2

σω = 2

The Reynolds stress tensor ρ u⨂u is calculated by

minusρ u⨂u = μ nablaU + (nablaU) minus

part ρk + μ

nabla ∙ U (316)

323 Shear stress transport model

The disadvantage of the Wilcox model is the strong sensitivity to free-stream

conditions Therefore a blending of the k-ω model near the surface and the k-ε in

the outer region was made by Menter [23] which resulted in the formulation of the

BSL k-ω turbulence model It consists of a transformation of the k-ε model to a k-ω

formulation and subsequently adding the resulting equations The Wilcox model is

multiplied by a blending function F1 and the transformed k-ε by another function 1-

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

30

F1 F1 is a function of wall distance (being the value of one near the surface and zero

outside the boundary layer) Outside the boundary and on the edge of the boundary

layer the standard k-ε model is used [24]

SST model is a combination of the k-epsilon in the free stream and the k-omega

models near the walls It does not use wall functions and tends to be most accurate

when solving the flow near the wall The SST model does not always converge to

the solution quickly so the k-epsilon or k-omega models are often solved first to

give good initial conditions

The k-ω based Shear Stress Transport (SST) model of Menter was applied for

turbulence treatment The transport equations for the SST model are expressed

below where the turbulent kinetic energy lsquokrsquo and turbulent frequency or dissipation

per unit turbulent kinetic energy are computed by using the following relations

For Turbulence Kinetic Energy

kkPukt

kTkk Ntilde+Ntilde+-=Ntilde+

para

para)[()( nsnwb (317)

where Pk is the production limiter

For Specific Dissipation Rate

ww

swnsnbwaww

ww NtildeNtilde-+Ntilde+Ntilde+-=Ntilde+para

para

1)1(2])[()( 21

22 kFSut

T (318)

The first blending function F1 is calculated from

iumlthorn

iumlyacute

uuml

iumlicirc

iumliacute

igrave

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

uacuteuacuteucirc

ugrave

ecircecirceuml

eacute

dividedivideoslash

oumlccedilccedilegrave

aelig=

4

22

21

4500maxmintanh

yCD

k

yy

kF

kw

wrs

w

n

wb (319)

y is the distance to the nearest wall and v is the kinematic viscosity

divideoslash

oumlccedilegrave

aeligNtildeNtilde= -10

2 101

2max ww

rsww kCDk (320)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

31

And Kinematic eddy viscosity

)max( 21

1

SF

kT

wa

an = (321)

where F2 is a blending function which restricts the limiter to the wall boundary and

S is the invariant measure of the strain rate

uacuteuacute

ucirc

ugrave

ecircecirc

euml

eacute

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

dividedivideoslash

oumlccedilccedilegrave

aelig=

2

22

5002maxtanh

w

n

wb yy

kF (322)

33 Brief introduction of simulation process

All aspects of CFD study including modelling meshing solving and post-

processing of the results are briefly introduced in this chapter

331 Pre-processing technique

The speed and accuracy of simulation results depending fundamentally on meshing

and the ability to prepare model geometry correctly

The steps in pre-processing are explained as follows

1) Domain Definitions

The domain definition depends mainly on the physics involved in the problem and

objective of the problem

a) Geometry creation or geometry import

b) Geometry cleanup

i) Removing the parts not necessary for simulation under consideration

ii) Closing the gaps in the geometry

iii) Removing small surfaces curves to make meshing process simple

2)Grid generation

To solve the governing equations the numerical methods like Finite Difference

Finite Volume or Finite Element Method to divide the domain into small parts and

solve the equations on one part at a time are applied

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 10: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

vii

from 50 to 140 of the design flow rate were studied by the performance curves

A trident shaped anti vortex device (AVD) composed of three wall fillets and one

center splitter was installed under the pump intake The effectiveness of the AVD

for the suppression of submerged vortex was evaluated In addition numerical

simulation of cavitation phenomenon in the mixed flow pump has been performed

by calculating the full cavitation model with k-ε turbulence model

According to the results details of the location size and strength of vortices were

predicted in the numerical simulation For the scaled sump model although there is

a little difference between the single phase and two-phase simulation all the results

predict the free surface vortex and submerged vortex formation and location The

CFD simulations of flow condition show good agreement with the PIV results The

effectiveness of curtain walls installed to inhibit the free surface vortex is confirmed

in the CFD results while the square bars installed in the channel failed to suppress

the free surface vortex For the mixed flow pump with sump the best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

The effectiveness of AVDs installed for the submerged vortex is confirmed by

comparisons of the sump model with and without AVD results

Cavitation phenomenon analysis shows the pump operating at the BEP condition

has a considerable extent over the requirement of cavitation performance With

numerical simulation the inception of cavitation in the blade passage is observed on

the suction surface where the leading edges meet the tips and then as the inlet total

pressure decrease the cavitation zone is spread out over the suction surface as well

as the leading edge of the impeller blade

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

viii

Nomenclature

B Distance from the back wall to the pump inlet bell centerline [m]

C Distance between the inlet bell and floor [m]

D Inlet bell design outside diameter [m]

Fr Froude number (=Vb(gD)05) [-]

g Acceleration due to gravity [ms2]

ht Total head [m]

n Number of revolution per min [-]

Pflow Flow power [W]

Pshaft Shaft power [W]

Pν Vapor pressure [Pa]

Q Volume flow rate [m3s]

Q0 Design volume flow rate [m3s]

Re Reynolds number (=VbDν) [-]

S Minimum submergence depth [m]

Vb Velocity at bell mouth [ms]

α Angle of floor slope [deg]

η Pump efficiency []

θ Swirl angle [deg]

λ The scale ratio of model dimensions to prototype [-]

μ Viscosity [Pa s]

ν Kinematic viscosity [m2s]

ρ Specific water density [kgm3]

σ Surface tension [Nm]

τ Stress tensor [-]

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

ix

Abbreviations

AVD Anti vortex device

BEP Best efficiency point

CFD Computational fluid Dynamics

NPSH Net positive suction head

NPSHR Required net positive suction head

PIV Particle Image Velocimetry

RANS Reynolds Averaged Navier- Stokes

RPM Number of revolutions per minute

SST Shear Stress Transport

VOF Volume of Fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

1

Chapter 1 Introduction

11 Background

It is an accepted fact that quite a number of problems faced in a pumping station are

related to the design of sump or intake rather than pump design As an important

part of the pumping station sump is designed to provide uniform swirl and

entrained air free flow to the pump intake Undesirable flow conditions (entrained

air non-uniform flow distribution vortices etc) can reduce the pump efficiency

induce vibration noise and cavitation even lead to excessive bearing loads of

impeller and structure damage Due to quite complex nature of the flow although

the flow conditions causing pump problems are well established there are no

specific solutions to eliminate them There are some design guidelines for the

specific geometrical and hydraulic constraints of the pump sump for any project

the model study is the only tool for solving potential problems in new designs and

rectifying problems observed in existing installations However the scaled model

tests are expensive and time-consuming so alternative Computational Fluid

Dynamics (CFD) methods for evaluating sump performance have been developed

With rapid progress in CFD numerical simulation is regarded as an effective tool in

solving fluid problems in pump sumps

12 Previous study

Various researches show the flow conditions around the sump intake structure are

quite complicated Shyam et al [1] used the commercial code CFX to carry out

two-phase flow simulation to capture air entrainment The air entrainments and its

location were well captured and numerical results were in accordance with the

experimental investigation Matsui J[2] and Okamura T[3] et al conducted a serial

research of the benchmark according to the Turbomachinery Society of Japan (TSJ)

standard including various commercial codes comparison performed in the flow

simulation and comparison analysis between the numerical and experimental results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

2

The purpose of the benchmark is to determine the accuracy and reliability of the

CFD codes used in universities and the industrial field It is also intended to identify

and investigate the formation of air-entrained and submerged vortices near the

pump Tanweer et al[4] have performed a numerical analysis of a multiple intake

pump sump to confirm that the geometry of sump affects the flow within sump

greatly Rajendran et al[5] have made a numerical analysis for the flow

characteristics of a sump model with pump intake and good agreements were

achieved by comparing the numerical results with the experiments Iwano et al[6]

have introduced a numerical method for the submerged vortex by analyzing the

flow in the pump sump with and without baffle plates Lee et al[7] have conducted

the CFD analysis of a multi-intake pump sump model to check the flow uniformity

by predicting the location number and vorticity of the vortex For the CFD

application in mixed flow pump design and performance analysis Oh H W [8]

conducted a mixed flow pump design optimization with specific parameters and

then compared the numerical results with the experiments The procedure presented

can be used as a practical design and analysis guide for the general turbomachinery

Kim et al[9] carried out a mixed flow pump performance analysis with numerical

simulation and improved the hydraulic efficiency by the modification of the

geometry

As an undesirable phenomenon caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage The cavitation

phenomenon has been studied by various investigators Van et al[10] have

investigated the cavitation inception behavior of a mixed-flow pump impellers

model and the calculated results matched well with the experiments Okamura et

al[11] have evaluated two numerical cavitation prediction methods used in the

pump industry manufacturers and attempts were made to improve the cavitation

performance Li et al[12] have conducted the numerical analysis on the basis of the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

3

development liquidvapor interface tracking method to predict the cavitation

characteristics with a centrifugal pump impeller model

Moreover there are a number of correlated studies [13]~[17]

13 Methodology of study

This study will investigate the flow phenomena of two sump models

A scaled sump model is adopted to carry out numerical simulation and PIV

experiment The flow conditions are predicted and the effectiveness of devices for

free surface vortex is evaluated by using numerical code a single intake sump

model with a mixed flow pump installed is modeled and the hydraulic performance

characteristics for the head rise shaft power and pump efficiencies versus flow rate

are studied In addition flow phenomena (vortices cavitation etc) are analyzed

and the effectiveness of an anti-vortex device (AVD) for the submerged vortex is

evaluated

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

4

Chapter 2 Basic theory of sump model

Specific hydraulic phenomena have been confirmed that they can lead to certain

common operational problems The common of the main problems are summarized

as follows

a Free surface vortices

b Submerged vortices

c Swirl of flow entering the pump

d Non-uniform distribution of velocity at the pump impeller

e Entrained air or gas bubbles

These problems encountered in the pump sump will affect the pump

performance and significantly increase the operational and maintenance costs In

order to identify sources of particular problems and find practical solution for

it the usual approach is to conduct the laboratory experiments on a scaled

physical model Typical design objectives are to ensure that a pump station is

designed according to best practice and conforms to the requirements set out in the

American National Standard for Pump Intake Design[18] These standards restrict

the degree to which the above-mentioned undesirable flow patterns may be present

The negative impact of each of these hydraulic conditions varies with pump specific

speed and size Application of the standard to design a sump does not generate a

problem free sump but provides a basis for the initial design

As there are no specific guidelines or criteria for design of trouble free intakes the

most common solution to potential problems in new design and rectification of

problems observed in existing designs is to construct a scaled model in a laboratory

observe and investigate the flow characteristics and propose modifications to the

intake geometry There are additional devices in the form of floor splitters or cones

back wall splitters fillets surface beams guide vanes etc aimed at controlling the

vortex and swirl formation Considerable prior experience and ingenuity are

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

5

required to solve the problems by this method and the cost and time required are

also significant

The intake structure refers to the channel leading from the water source which

may be a river or reservoir and all installations downstream including the pump

column or the intake pipe (the suction tube portion of a vertical intake) the

approach channel (upstream of pump bay) and the pump bay (bounded by the floor

the back wall side walls dividers walls separating adjacent pump column) Usually

the upstream end of the pump column has an inlet attachment called the suction bell

The function of the intake is to supply an evenly distributed flow of water to the

pump suction bell

Several types of pump intake basins are available and ANSIHI 98 provides

design requirements for each Intake structures can be categorized as being

clear liquids or solids-bearing liquids For clear liquids intakes are further

classified into rectangular formed circular and trench types as well as suction

tanks and cans For solid-bearing liquids trench type and rectangular wet wells are

usually considered

In this paper only rectangular intake structure for clear liquid is studied

21 Pump intake design

For pumps to achieve their optimum hydraulic performance across all operating

conditions the flow at the impeller must meet specific hydraulic conditions The

ideal flow entering the pump inlet should be of a uniform velocity distribution

without rotation and stable over time When designing a sump to achieve favorable

inflow to the pump or suction pipe bell various sump dimensions relative to the size

of the bell are required Using these standards of the distances to reduce the

probability of the occurence of strong submerged vortices and free-surface vortices

211 Inlet bell diameter design

The geometry is generally defined in terms of the pump inlet diameter as shown in

Table 21 and Fig 21 The bell diameter can be estimated based on an inlet pipe

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

6

Table 21 Acceptable velocity ranges for inlet bell diameter(ANSIHI 98)

Flow Rate

Qls

Recommended Inlet Bell Design

Velocityms

Acceptable Velocity

Range ms

Q<315 V=17 06leVle27

315 leQ<1260 V=17 09leVle24

Q ge1260 V=17 12leVle21

Fig 21 Recommended inlet bell design diameter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

7

212 Recommended dimensions for a rectangular sump

As the selected bell diameter has been determined the proportions of the inlet

structure can be estimated from Fig 22 and Fig 23 Table 22 shows the

recommended dimensions for a rectangular sump

Fig 22 Recommended intake structure layout

Fig 23 Filler wall details for proper bay width

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

8

Table 22 Recommended dimensions for a rectangular sump

Dimension Variable

Description Recommended value

A Distance from the pump inlet bell

centerline to the entrance A=5D minimum

a Length of constricted bay section near

the pump inlet 25D minimum

B Distance from the back wall to the

pump inlet bell centerline 075D

C Distance between the inlet bell and

floor 03D~05D

D Inlet bell outside diameter refer to section 211

H Minimum liquid depth H=S+C

h Minimum height of constricted bay

section near the pump inlet bell hgtH or 25D

S Minimum pump inlet bell submergence S=D(10+23Fr)

W Pump inlet bay entrance width 2D minimum

w Constricted bay width near the pump

inlet bell w=2D

X Pump inlet bay length 5D minimum

Y Distance from pump inlet bell

centerline to the through-flow traveling screen

4D minimum

Z1 Distance from sump inlet bell centerline to diverging walls

5D minimum

Z2 Distance from inlet bell centerline to

sloping floor 5D minimum

α Angle of floor slope -10deg~ +10deg

β Angle of wall convergence 0deg~ +10deg

empty Angle of convergence from constricted

area to by walls 10deg maximum

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

9

FD=Vb(gD)05 (21)

where

FD Froude number (dimensionless)

Vb velocity at suction inlet=FlowArea based on D

D Outside diameter of bell or pipe inlet

g gravitational acceleration

There is the design sequence summary for rectangular intake structures

middot Consider the flow patterns and boundary geometry of the body of liquid

from which the pump station is to receive flow Compare with the approach

flow condition and determine if a hydraulic physical model study is

required

middot Determine the number and size of pumps required to satisfy the range of

operating conditions likely to be encountered

middot Identify pump inlet bell diameter Refer to the section 211

middot Determine the bell-floor clearance 05D is a good preliminary design value

middot Determine the required bell submergence using the equation 21

middot Determine the minimum allowable liquid depth in the intake structure from

the sum of the floor clearance and the required bell submergence

middot Check bottom elevation near the entrance to the structure and determine if it

is necessary to slope the floor upstream of the bay entrance If the resulting

depth at the entrance to the intake structure is shallow then check to ensure

that gravity-driven flow is not restricted by the entrance condition

middot Check the pump bay velocity for the maximum single-pump flow and

minimum liquid depth with the bay width set to 2D If bay velocity exceeds

05 ms then increase the bay width to reduce the maximum velocity below

05 ms

middot If it is necessary to increase the pump bay width to greater than 2D then

decrease bay width in the vicinity of the pumps according to Fig 23

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

10

middot Compare cross-flow velocity (at maximum system flow) to average pump

bay velocity If cross-flow value exceeds 50 of the bay velocity a

physical hydraulic model study is necessary

middot Determine the length of the structure and dividing walls giving

consideration to minimum allowable distances to a sloping floor screening

equipment and length of dividing walls if dual flow traveling screens or

drum screens are to be used a physical hydraulic model study is required

middot If the final selected pump bell diameter and inlet velocity is within the

range given in section 211 then the sump dimensions (developed based on

the inlet bell design diameter) need not be changed and will comply with

these standards

22 Model test

The approach flow pattern in a pump intake facility is difficult to predict by

traditional mathematics or empirical formulae For large structures or those that

differ significantly from proven designs a model study is the only means to ensure

success A model study is used to identify adverse hydraulic conditions and

derive remedial measures for approach flow patterns generated by structures

upstream of the pump impeller Current model studies are not intended to

investigate flow patterns induced by the pump itself or the flow patterns within the

pump The objective of a model study is to ensure that the pump intake structure

generates favorable flow conditions at the inlet to the pump Evaluation for the

requirement of model test if

a Sump or piping geometry that deviates from this design standard

b Non-uniform or non-symmetric approach flow to the pump sump

exist

c The pumps have flows greater than 2520 ls per pump or the total

station flow with all pumps running would be greater than 6310 ls

d Situations that need ten times cost than the model study

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

11

Adverse hydraulic conditions that can affect pump performance include free and

submerged vortices swirl approaching the pump impeller flow separation at the

pump bell and a non-uniform axial velocity distribution at the suction

The purpose of model test is to examine these adverse hydraulic conditions and to

investigate measure against them

221 The scale effects

One possible difficulty in scale modeling is that it is impossible to reduce all

pertinent forces by the same factor With Froude-scale models the inertial and

gravitational forces are reduced similarly but the viscous and surface tension forces

cannot be simultaneously reduced The extra influence of these forces is known as

a ldquoscale effectrdquo A number of investigations have indicated that the scale effects for

the model are avoided if the Reynolds number based on the inlet flow and

submergence or intake diameter exceeds approximately 3x 104 and the Weber

number exceeds 150 [19]

In most engineering applications involving closed-conduit and open-channel flow

the Reynolds number limits are far exceeded and the flows are fully turbulent

In a sump model Froude number (Fr) and Reynolds number (Re) are the most

important non-dimensional parameters Some reduction in Re should be

introduced to the sump model However flow patterns are generally very similar

at high Re

No specific geometric scale rate is recommended but reasonable large

geometric scale to minimize viscous and surface tension scale effects and reproduce

the flow pattern

Re= uDν (22)

We= u2ρDσ (23)

where

u average axial velocity (ms)

ν kinematic viscosity of the liquid (m2s)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

12

ρ liquid density (kgm3)

σ surface tension of liquidair interface

According to the ANSIHI 98 the viscous scale effects on vortex may be negligible

if the resulting dimensionless numbers must meet these values Re>6times104 and

We>240

To allow practical visual observations of flow patterns accurate measurements of

swirl and velocity distribution and sufficient dimensional control the model scale

shall yield a bay width of at least 300mm a minimum liquid depth of at least

150mm and a pump throat or suction diameter of at least 80mm

222 Principle of similarity

In the similitude analysis the geometric and flow similarity requires the model and

the flow to be identical as the real model and flow This is to ensure the result

obtained from the model study is well presented in order to predict full scale

behavior Model involving a free surface are operated using Froude similarity since

the flow process is controlled by gravity and inertial forces

For similarity of the flow patterns the Froude number shall be equal in model and

prototype

Fr =

=

(24)

and as the scale ratio λ was selected in section 221 the following equation is

obtained

V = V times (

) = V times λ or Q = Q times (

) = Q times λ (25)

where Fr is the Froude number λ is the scale ratio of model dimensions to prototype

The test of 15 times Froude scaled flow should be conducted to ascertain the

potential scale affect on free surface vortices

Models of closed conduit piping system leading to a pump suction are not operated

based on Froude similitude However flow patterns are generally very similar at

high Re A minimum value of 1times105 for the Re at the pump suction is

recommended to ensure the flow patterns are correctly scaled

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

13

223 Preliminary model test

The initial design shall be tested first to identify any hydraulic problems

1) Vortex Formation Mechanisms

There are various investigations that provide insight into the fundamental processes

leading to the development of vortices both in experiment and numerical simulation

Shin et al [20] demonstrated that two basic mechanisms lead to inlet vortex

formation The first mechanism involves the development of an inlet vortex due to

the amplification of ambient vorticity in the approach flow as vortex lines

are convicted into the inlet The second mechanism involves the development of a

trailing vortex in the vicinity of the intake as a result of the variation in

circulation along the inlet For this second case a vortex can develop in a flow that

is irrotational upstream and the vortex development therefore does not depend

on the presence of ambient vorticity Shin et al investigation on kinematic

parameters indicate that the strength of an inlet-vortex or trailing vortex system

increases with decreasing distance from the surface However for an inlet in

an upstream irrotational flow two counter rotating vortices can still trail from

the rear of the inlet

2) Classification of vortex type

As illustrated in Fig 24 vortices in the vicinity of pump intakes may be adjacent to

the channel bottom or a channel wall (ie submerged vortices) or they may appear

adjacent to the free surface (ie free surface vortex)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

14

(a) Schematic diagram of vortices in a sump

(b) Experimental results of different vortex

Fig 24 Typical vortex in a sump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

15

For the model visual observation dye was injected around the intake structure near

the surface and submergence as Fig 24 (b) shows And the model test results of

vortex was classified as Fig 25

(a) Free Surface Vortices

(b) Submerged Vortices

Fig 25 Vortex classifications

There are a number of researches focused on the numerical theory to detect the

inception of a visible vortex As the diameters of vortices are usually much smaller

compared with the computing grid size to compensate for the lack of vortex

resolution in numerical simulation a stretching vortex model is applied to the local

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

16

flow field around semi-analytically identified vortex positions [21][22] The brief

explanations are as follows

a) Free surface vortex

The criteria expressed by equation 26 and 27 is applied to determine the

visible inception of a free surface vortex

≫ 1 (26)

=

gt 1 (27)

where P is the pressure drop at the vortex core and Ph is the static head at the

vortex element and f is the drag force acting at a unit length of water surface

vortex dimple based on Stokes approximation and caused by the down flow

Further f is the force acting on the bubble in the direction from the high

pressure region to the low pressure region like buoyancy

b) Submerged vortex

Equation 28 is applied to determine the visible inception of a submerged vortex

infin =

infin

infin gt 1 (28)

where is the static pressure at the vortex core infin is the atmospheric pressure

and is the critical cavitation inception pressure and nearly equal to the

saturated vapor pressure

3) Swirl angle

Swirl angle predicts the intensity of flow rotation Fig 26 shows experiment

calculation method with swirl meter and attempts of the CFD calculation method

were conducted as Fig 27 A four blade zero-pitch swirl meter which is supported

by a low friction bearing will be installed at about four suction pipe diameter (d)

downstream from the pump suction to measure swirl angle of pump approaching

flow For the swirl meter the tip to tip blade diameter is 075d and the length in

flow direction is 06 and one of the four blade is painted yellow as a reference to

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

17

count revolutions As flow swirl is usually unsteady the observation time of swirl

meter reading should be a continuous period of time

Fig 26 Typical swirl meter

(29)

(210)

(211)

Where qV flow rotational speed at swirl meter (ms)

zV average axial velocity at swirl meter (ms)

n Number of revolution per minute (rpm)

d Diameter of pipe at the swirl meter (m)

q flow rate (m3s)

zV

Vqq 1tan -=

)(60

smnd

Vtimestimes

=p

q

)(4

2

smd

qVz

times=

p

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

18

For the swirl angle calculating method with CFD the key point is to obtain the

average tangential velocity so the swirl check circles were created as illustrated in

Fig 27 The swirl check circle is located at the section of 4d height with diameter

of 025d 05d and 075d respectively

Swirl check circle

Fig 27 Schematic view of swirl angle numerical calculation

23

1 7501 501 250

acute++

=aringaringaring

N

i

N

i

NNVNVNV

Vi

q (212)

where

qV average tangential velocity (ms)

V circumferential velocity of check point in check circle

respectively (ms)

N check point total numbers in single check circle

4) Velocity

Velocity measurement at bell throat by Pitot tube was implemented by ANSIHI

method Typical Pitot tube configuration is shown in Fig 28 Eight measuring

points are defined as Fig 29 Time averaged velocity at eight points was obtained

and these eight data were again averaged to give throat sectional spatial velocity

average

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

19

Fig 28 Example of Pitot tube Fig 29 Measuring points at bell throat

224 Test criteria

There is the criteria for satisfactory performance of the sump model test and

summarized in Table 23

1) Free surface and submerged vortices

Free surface vortices entering the pump must be less severe than vortices with

coherent core Dye core vortices may be acceptable only if they occur less than 10

of the time or only for infrequent pump operating conditions

Submerged vortex which starts from the sump walls induce vibration and noise

problems Dye core submerged vortices extending into the pump bell mouth is not

allowed

2) Swirl angles

Swirl angle is used to check the flow rotation in the suction pipe The calculation

method is referred to section 223

Short-term maximum and long-term average values must be less than 5 degrees

Maximum short-term swirl angles up to 7 degrees may be acceptable only if they

occur less than 10 of the time or only for infrequent pump operating conditions

3)Velocity

Each point time-averaged velocity shall not be larger than 10 of sectional average

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

20

Standard deviation of time-varying velocity at each point was obtained and its

value shall not be larger than 10 of the time-averaged velocity at each point

1) Time-averaged

Variation[] = 100acute-

a

a

V

VV aV Averaged at section

2) Time-varying

Standard deviation= )1(

)(1i

2

-

-aring=

n

xxn

i

Variation[] = (Standard deviation X ) x 100

X Averaged at point

Table 23 Permission criteria of ANSIHI

Vortex occurence Swirl angle

Free surface vortex Submerged vortex

Observation

time Above 10 minutes Above 10 minutes

Above 10

minutes

Permission

criteria

Dye-core

Vortex(F3)

below 110 total

time

Dye-core

Vortex(S2)

below 110 total

time

Below

5degConditional

intermittent

below 7deg

225 Remedial measures for problem intakes

Usually in the preliminary test the worst flow condition is simulated And based on

the results remedial measures for problem intakes are discussed in this part

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

21

1) Overall approach flow patterns

The characteristics of flow approaching an intake structure is one of the foremost

considerations for the designer It is difficult to predict the effects of a given set of

flow conditions upstream from an intake structure Fig 210 depicts several typical

approach flow patterns of pump sump with different operating conditions

Fig 210 Examples of approach flow patterns

Generally speaking for the intake problems especially for the multiple pumps

sump there are the main measures using dividing walls placed between the pumps

as a partitioned structure trash racks with elongated bars guide piers or a number

of structural columns at the bay entrance can provide some assistance in controlling

cross-flow baffles flow distributors and screen mesh etc are applied to improve

the flow distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

22

2) Problems around pump intake

a) Free surface vortices

There are a number of factors that have an effect on the formation of surface

vortices To achieve a higher degree of certainty that objectionable surface vortices

do not form modifications can be made to intake structure to allow operation at

practical depths of submergence and the minimum submergence is referred in Table

22 the use of suction umbrellas vertical curtain walls and horizontal gratings

b) Submerged vortices

In the vicinity of intake bell there are the most complicated flow patterns and flow

makes the most changes in direction while maintaining a constant acceleration into

the pump bell to prevent local flow separation turbulence and vortex formation

Fig 211 represents a sample of various devices to inhibit the different submerged

vortex

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

23

Fig 211 Methods to reduce submerged vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

24

c) Preswirl

The approach flow distributions determine whether flow preswirl exists and a

sufficiently laterally skewed approach flow cause rotating around the intake bell

The most effective way of reducing preswirl is to establish a relatively uniform

approach flow within each pump bay by using the baffling schemes controlling

cross-flow and expanding concentrated flow discussed in the above part

To prevent non-uniform velocity distribution swirls and vortices necessary

modifications are conducted such as fore bay length baffle wall distributors

curtain wall mesh screen guide vane floor splitter wall fillet etc As the above

measures applied modification test will be conducted to check all the flow

conditions and compare with the preliminary results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

25

Chapter 3 Methodology

CFD (Computational fluid dynamics) is the technique which can design cycles to

better performance and reduce costs and time In CFD we solve the governing

equations of given physics (may be differential form or integral form) using some

numerical techniques like Finite Difference Method (FDM) Finite Element Method

(FEM) or Finite Volume Method (FVM) Although CFD plays a great part in the

design of turbomachinery CFD is not a replacement for experimental or analytical

approach In this study the numerical analysis was performed by the commercial

code ANSYS CFX 130 All aspects of the CFD study including modelling

meshing solving and post-processing of the results

31 Numerical modeling

ANSYS CFX solves the unsteady Navier-Stokes equations in their conservation

form The instantaneous equation of mass (continuity) in the stationary frame is

expressed as equation

ρ

δ + nabla ∙ (ρU) = 0 (31)

And the instantaneous equation for momentum is expressed as shown in equation

ρ

δ + nabla ∙ (ρU⨂U) = minusnablap + nabla ∙ τ + S (32)

Where τ is the stress tensor ( including both normal and shear components of the

stress)

These instantaneous equations are averaged for turbulent flows leading to additional

terms that need to be solved While the Navier-Stokes equations describe both

laminar and turbulent flows without addition terms realistic flows involve length

scales much smaller than the smallest finite volume mesh A Direct Numerical

Simulation of these flows would require significantly more computing power than

what is available now or in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

26

Therefore much research has been done to predict the effects of turbulence by

using turbulence models These models account for the effects of turbulence

without the use of a very fine mesh or direct numerical simulation

These turbulence models modify the transport equations by adding averaged and

fluctuating components The transport equations are changed to equations (33) and

(34)

ρ

δ + nabla ∙ (ρU) = 0 (33)

ρ

+ nabla ∙ ρ U ⨂U = minusnablap + nabla ∙ τ minus ρu⨂u + S (34)

The mass equation is not changed but the momentum equation contains extra terms

which are the Reynolds stresses ρ u⨂u and the Reynolds flux ρ u ϕ These

Reynolds stresses need to be modeled by additional equations to obtain closure

Obtaining closure implies that there are a sufficient number of equations to solve for

all the unknowns including the Reynolds stresses and Reynolds fluxes

Various turbulence models provide various ways to obtain closure In this

investigation Shear Stress Transport (SST) model and k-ε model were utilized The

advantage of SST model is that combines the advantages of other turbulence models

(the k-ε Wilcox k-ω) The 3 turbulence models will be discussed briefly

32 Turbulence models

Turbulence models are necessary because we cannot afford big enough computers

to directly capture every scale of motion There are three modelling frameworks

Direct Numerical Simulation (DNS) Reynolds-Averaged Navier-Stokes

Equations(RANS) Large Eddy Simulation(LES)

As RANS method is applied in most productionseveral RANS turbulence models

differ in their usage of wall functions the number of additional variables solved for

and what these variables represent

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

27

Two equation turbulence models

Two-equation turbulence models are widely used in CFD as they give a good

compromise between computational power needed and accuracy The term lsquotwo

equationrsquo refers to the fact that these models solve for the velocity and length scales

using separate transport equations The turbulence velocity scale is obtained by

solving the transport equation The turbulent length scale is estimated from two

properties of the turbulence field namely the turbulent kinetic energy and the

dissipation rate The dissipation rate of the kinetic energy is obtained from its

transport equation The most widely used are k-ε and k-ω two equation models In

the next section the k-ε Wilcox k-ω BSL k-ω and SST models will be briefly

discussed

321 k-ε turbulence model

The k-epsilon model solves for two variables k the turbulent kinetic energy and

epsilon the rate of dissipation of kinetic energy Wall functions are used in this

model so the flow in the buffer region is not simulated The k-epsilon model is very

popular for industrial applications due to its good convergence rate and relatively

low memory requirements It does not very accurately compute flow fields that

exhibit adverse pressure gradients strong curvature to the flow or jet flow It does

perform well for external flow problems around complex geometries

The k-ε model introduces k (m2s2) as the turbulence kinetic energy and ε (m2s3) as

the turbulence eddy dissipation The continuity equation remains the same

ρ

+ nabla ∙ (ρ U) = 0 (35)

The momentum equation changes as shown by equation

δρ

δ + nabla ∙ (ρU⨂U) = minusnablapprime + nabla ∙ μ

nablaU + (nablaU) + S (36)

Where SM is the sum of body forces μeff is the effective viscosity accounting for

turbulence and prsquo is the modified pressure The k-ε model uses the concept of eddy

viscosity giving the equation for effective viscosity as shown by equation

μ

= μ + μ (37)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

28

μt is the turbulence viscosity is linked to the turbulence kinetic energy and

dissipation by the equation

μ = Cμρ

ε (38)

Where Cμ is a constant

The values for k and ε come from the differential transport equations for the

turbulence kinetic energy and the turbulence dissipation rate

The turbulence kinetic energy equation is given as equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus ρε (39)

The turbulence dissipation rate is given by equation

(ρε)

+ nabla ∙ (ρUε) = nabla ∙ μ +

μ

σε nablaε +

ε

(Cε (P + Pε ) minus Cε ρε) (310)

Where Cε1 Cε2 σk σε are constants

Pk is the turbulence production due to viscous forces and is modeled by the

equation

P = μ nablaU ∙ nablaU + nablaU minus

nabla ∙ U 3μ

nabla ∙ U + ρk (311)

If a buoyancy term is added to the previous equation if the full buoyancy model is

used

The buoyancy term Pkb is modeled as

P = minusμ

ρσρg ∙ nablaρ (312)

322 Wilcox k-ω turbulence model

The k-omega model is similar to k-epsilon instead however it solves for omega mdash

the specific rate of dissipation of kinetic energy It also uses wall functions and

therefore has comparable memory requirements It has more difficulty converging

and is quite sensitive to the initial guess at the solution Hence the k-epsilon model

is often used first to find an initial condition for solving the k-omega model The k-

omega model is useful in many cases where the k-epsilon model is not accurate

such as internal flows flows that exhibit strong curvature separated flows and jets

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

29

This model has an advantage over the k-ε model where it does not involve complex

linear damping functions for near wall calculations at low Reynolds numbers

The k-ω model assumes that the turbulence viscosity is related to the turbulence

kinetic energy k and the turbulent frequency ω by the equation

μ = ρ

ω (313)

The transport equation for k is given by the equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus βprimeρ k ω (314)

The transport equation for ω is shown as equation

(ρω)

δ + nabla ∙ (ρ U ω) = nabla ∙ μ +

μ

μω

nablaω + αω

P + Pω minus β ρ ω (315)

The production rate of turbulence (Pk) is calculated as shown previously in the k-ε

section

The recommended values for the constants in the above equations are

βprime = 009

α = 59

β = 0075

σ = 2

σω = 2

The Reynolds stress tensor ρ u⨂u is calculated by

minusρ u⨂u = μ nablaU + (nablaU) minus

part ρk + μ

nabla ∙ U (316)

323 Shear stress transport model

The disadvantage of the Wilcox model is the strong sensitivity to free-stream

conditions Therefore a blending of the k-ω model near the surface and the k-ε in

the outer region was made by Menter [23] which resulted in the formulation of the

BSL k-ω turbulence model It consists of a transformation of the k-ε model to a k-ω

formulation and subsequently adding the resulting equations The Wilcox model is

multiplied by a blending function F1 and the transformed k-ε by another function 1-

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

30

F1 F1 is a function of wall distance (being the value of one near the surface and zero

outside the boundary layer) Outside the boundary and on the edge of the boundary

layer the standard k-ε model is used [24]

SST model is a combination of the k-epsilon in the free stream and the k-omega

models near the walls It does not use wall functions and tends to be most accurate

when solving the flow near the wall The SST model does not always converge to

the solution quickly so the k-epsilon or k-omega models are often solved first to

give good initial conditions

The k-ω based Shear Stress Transport (SST) model of Menter was applied for

turbulence treatment The transport equations for the SST model are expressed

below where the turbulent kinetic energy lsquokrsquo and turbulent frequency or dissipation

per unit turbulent kinetic energy are computed by using the following relations

For Turbulence Kinetic Energy

kkPukt

kTkk Ntilde+Ntilde+-=Ntilde+

para

para)[()( nsnwb (317)

where Pk is the production limiter

For Specific Dissipation Rate

ww

swnsnbwaww

ww NtildeNtilde-+Ntilde+Ntilde+-=Ntilde+para

para

1)1(2])[()( 21

22 kFSut

T (318)

The first blending function F1 is calculated from

iumlthorn

iumlyacute

uuml

iumlicirc

iumliacute

igrave

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

uacuteuacuteucirc

ugrave

ecircecirceuml

eacute

dividedivideoslash

oumlccedilccedilegrave

aelig=

4

22

21

4500maxmintanh

yCD

k

yy

kF

kw

wrs

w

n

wb (319)

y is the distance to the nearest wall and v is the kinematic viscosity

divideoslash

oumlccedilegrave

aeligNtildeNtilde= -10

2 101

2max ww

rsww kCDk (320)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

31

And Kinematic eddy viscosity

)max( 21

1

SF

kT

wa

an = (321)

where F2 is a blending function which restricts the limiter to the wall boundary and

S is the invariant measure of the strain rate

uacuteuacute

ucirc

ugrave

ecircecirc

euml

eacute

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

dividedivideoslash

oumlccedilccedilegrave

aelig=

2

22

5002maxtanh

w

n

wb yy

kF (322)

33 Brief introduction of simulation process

All aspects of CFD study including modelling meshing solving and post-

processing of the results are briefly introduced in this chapter

331 Pre-processing technique

The speed and accuracy of simulation results depending fundamentally on meshing

and the ability to prepare model geometry correctly

The steps in pre-processing are explained as follows

1) Domain Definitions

The domain definition depends mainly on the physics involved in the problem and

objective of the problem

a) Geometry creation or geometry import

b) Geometry cleanup

i) Removing the parts not necessary for simulation under consideration

ii) Closing the gaps in the geometry

iii) Removing small surfaces curves to make meshing process simple

2)Grid generation

To solve the governing equations the numerical methods like Finite Difference

Finite Volume or Finite Element Method to divide the domain into small parts and

solve the equations on one part at a time are applied

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 11: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

viii

Nomenclature

B Distance from the back wall to the pump inlet bell centerline [m]

C Distance between the inlet bell and floor [m]

D Inlet bell design outside diameter [m]

Fr Froude number (=Vb(gD)05) [-]

g Acceleration due to gravity [ms2]

ht Total head [m]

n Number of revolution per min [-]

Pflow Flow power [W]

Pshaft Shaft power [W]

Pν Vapor pressure [Pa]

Q Volume flow rate [m3s]

Q0 Design volume flow rate [m3s]

Re Reynolds number (=VbDν) [-]

S Minimum submergence depth [m]

Vb Velocity at bell mouth [ms]

α Angle of floor slope [deg]

η Pump efficiency []

θ Swirl angle [deg]

λ The scale ratio of model dimensions to prototype [-]

μ Viscosity [Pa s]

ν Kinematic viscosity [m2s]

ρ Specific water density [kgm3]

σ Surface tension [Nm]

τ Stress tensor [-]

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

ix

Abbreviations

AVD Anti vortex device

BEP Best efficiency point

CFD Computational fluid Dynamics

NPSH Net positive suction head

NPSHR Required net positive suction head

PIV Particle Image Velocimetry

RANS Reynolds Averaged Navier- Stokes

RPM Number of revolutions per minute

SST Shear Stress Transport

VOF Volume of Fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

1

Chapter 1 Introduction

11 Background

It is an accepted fact that quite a number of problems faced in a pumping station are

related to the design of sump or intake rather than pump design As an important

part of the pumping station sump is designed to provide uniform swirl and

entrained air free flow to the pump intake Undesirable flow conditions (entrained

air non-uniform flow distribution vortices etc) can reduce the pump efficiency

induce vibration noise and cavitation even lead to excessive bearing loads of

impeller and structure damage Due to quite complex nature of the flow although

the flow conditions causing pump problems are well established there are no

specific solutions to eliminate them There are some design guidelines for the

specific geometrical and hydraulic constraints of the pump sump for any project

the model study is the only tool for solving potential problems in new designs and

rectifying problems observed in existing installations However the scaled model

tests are expensive and time-consuming so alternative Computational Fluid

Dynamics (CFD) methods for evaluating sump performance have been developed

With rapid progress in CFD numerical simulation is regarded as an effective tool in

solving fluid problems in pump sumps

12 Previous study

Various researches show the flow conditions around the sump intake structure are

quite complicated Shyam et al [1] used the commercial code CFX to carry out

two-phase flow simulation to capture air entrainment The air entrainments and its

location were well captured and numerical results were in accordance with the

experimental investigation Matsui J[2] and Okamura T[3] et al conducted a serial

research of the benchmark according to the Turbomachinery Society of Japan (TSJ)

standard including various commercial codes comparison performed in the flow

simulation and comparison analysis between the numerical and experimental results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

2

The purpose of the benchmark is to determine the accuracy and reliability of the

CFD codes used in universities and the industrial field It is also intended to identify

and investigate the formation of air-entrained and submerged vortices near the

pump Tanweer et al[4] have performed a numerical analysis of a multiple intake

pump sump to confirm that the geometry of sump affects the flow within sump

greatly Rajendran et al[5] have made a numerical analysis for the flow

characteristics of a sump model with pump intake and good agreements were

achieved by comparing the numerical results with the experiments Iwano et al[6]

have introduced a numerical method for the submerged vortex by analyzing the

flow in the pump sump with and without baffle plates Lee et al[7] have conducted

the CFD analysis of a multi-intake pump sump model to check the flow uniformity

by predicting the location number and vorticity of the vortex For the CFD

application in mixed flow pump design and performance analysis Oh H W [8]

conducted a mixed flow pump design optimization with specific parameters and

then compared the numerical results with the experiments The procedure presented

can be used as a practical design and analysis guide for the general turbomachinery

Kim et al[9] carried out a mixed flow pump performance analysis with numerical

simulation and improved the hydraulic efficiency by the modification of the

geometry

As an undesirable phenomenon caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage The cavitation

phenomenon has been studied by various investigators Van et al[10] have

investigated the cavitation inception behavior of a mixed-flow pump impellers

model and the calculated results matched well with the experiments Okamura et

al[11] have evaluated two numerical cavitation prediction methods used in the

pump industry manufacturers and attempts were made to improve the cavitation

performance Li et al[12] have conducted the numerical analysis on the basis of the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

3

development liquidvapor interface tracking method to predict the cavitation

characteristics with a centrifugal pump impeller model

Moreover there are a number of correlated studies [13]~[17]

13 Methodology of study

This study will investigate the flow phenomena of two sump models

A scaled sump model is adopted to carry out numerical simulation and PIV

experiment The flow conditions are predicted and the effectiveness of devices for

free surface vortex is evaluated by using numerical code a single intake sump

model with a mixed flow pump installed is modeled and the hydraulic performance

characteristics for the head rise shaft power and pump efficiencies versus flow rate

are studied In addition flow phenomena (vortices cavitation etc) are analyzed

and the effectiveness of an anti-vortex device (AVD) for the submerged vortex is

evaluated

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

4

Chapter 2 Basic theory of sump model

Specific hydraulic phenomena have been confirmed that they can lead to certain

common operational problems The common of the main problems are summarized

as follows

a Free surface vortices

b Submerged vortices

c Swirl of flow entering the pump

d Non-uniform distribution of velocity at the pump impeller

e Entrained air or gas bubbles

These problems encountered in the pump sump will affect the pump

performance and significantly increase the operational and maintenance costs In

order to identify sources of particular problems and find practical solution for

it the usual approach is to conduct the laboratory experiments on a scaled

physical model Typical design objectives are to ensure that a pump station is

designed according to best practice and conforms to the requirements set out in the

American National Standard for Pump Intake Design[18] These standards restrict

the degree to which the above-mentioned undesirable flow patterns may be present

The negative impact of each of these hydraulic conditions varies with pump specific

speed and size Application of the standard to design a sump does not generate a

problem free sump but provides a basis for the initial design

As there are no specific guidelines or criteria for design of trouble free intakes the

most common solution to potential problems in new design and rectification of

problems observed in existing designs is to construct a scaled model in a laboratory

observe and investigate the flow characteristics and propose modifications to the

intake geometry There are additional devices in the form of floor splitters or cones

back wall splitters fillets surface beams guide vanes etc aimed at controlling the

vortex and swirl formation Considerable prior experience and ingenuity are

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

5

required to solve the problems by this method and the cost and time required are

also significant

The intake structure refers to the channel leading from the water source which

may be a river or reservoir and all installations downstream including the pump

column or the intake pipe (the suction tube portion of a vertical intake) the

approach channel (upstream of pump bay) and the pump bay (bounded by the floor

the back wall side walls dividers walls separating adjacent pump column) Usually

the upstream end of the pump column has an inlet attachment called the suction bell

The function of the intake is to supply an evenly distributed flow of water to the

pump suction bell

Several types of pump intake basins are available and ANSIHI 98 provides

design requirements for each Intake structures can be categorized as being

clear liquids or solids-bearing liquids For clear liquids intakes are further

classified into rectangular formed circular and trench types as well as suction

tanks and cans For solid-bearing liquids trench type and rectangular wet wells are

usually considered

In this paper only rectangular intake structure for clear liquid is studied

21 Pump intake design

For pumps to achieve their optimum hydraulic performance across all operating

conditions the flow at the impeller must meet specific hydraulic conditions The

ideal flow entering the pump inlet should be of a uniform velocity distribution

without rotation and stable over time When designing a sump to achieve favorable

inflow to the pump or suction pipe bell various sump dimensions relative to the size

of the bell are required Using these standards of the distances to reduce the

probability of the occurence of strong submerged vortices and free-surface vortices

211 Inlet bell diameter design

The geometry is generally defined in terms of the pump inlet diameter as shown in

Table 21 and Fig 21 The bell diameter can be estimated based on an inlet pipe

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

6

Table 21 Acceptable velocity ranges for inlet bell diameter(ANSIHI 98)

Flow Rate

Qls

Recommended Inlet Bell Design

Velocityms

Acceptable Velocity

Range ms

Q<315 V=17 06leVle27

315 leQ<1260 V=17 09leVle24

Q ge1260 V=17 12leVle21

Fig 21 Recommended inlet bell design diameter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

7

212 Recommended dimensions for a rectangular sump

As the selected bell diameter has been determined the proportions of the inlet

structure can be estimated from Fig 22 and Fig 23 Table 22 shows the

recommended dimensions for a rectangular sump

Fig 22 Recommended intake structure layout

Fig 23 Filler wall details for proper bay width

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

8

Table 22 Recommended dimensions for a rectangular sump

Dimension Variable

Description Recommended value

A Distance from the pump inlet bell

centerline to the entrance A=5D minimum

a Length of constricted bay section near

the pump inlet 25D minimum

B Distance from the back wall to the

pump inlet bell centerline 075D

C Distance between the inlet bell and

floor 03D~05D

D Inlet bell outside diameter refer to section 211

H Minimum liquid depth H=S+C

h Minimum height of constricted bay

section near the pump inlet bell hgtH or 25D

S Minimum pump inlet bell submergence S=D(10+23Fr)

W Pump inlet bay entrance width 2D minimum

w Constricted bay width near the pump

inlet bell w=2D

X Pump inlet bay length 5D minimum

Y Distance from pump inlet bell

centerline to the through-flow traveling screen

4D minimum

Z1 Distance from sump inlet bell centerline to diverging walls

5D minimum

Z2 Distance from inlet bell centerline to

sloping floor 5D minimum

α Angle of floor slope -10deg~ +10deg

β Angle of wall convergence 0deg~ +10deg

empty Angle of convergence from constricted

area to by walls 10deg maximum

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

9

FD=Vb(gD)05 (21)

where

FD Froude number (dimensionless)

Vb velocity at suction inlet=FlowArea based on D

D Outside diameter of bell or pipe inlet

g gravitational acceleration

There is the design sequence summary for rectangular intake structures

middot Consider the flow patterns and boundary geometry of the body of liquid

from which the pump station is to receive flow Compare with the approach

flow condition and determine if a hydraulic physical model study is

required

middot Determine the number and size of pumps required to satisfy the range of

operating conditions likely to be encountered

middot Identify pump inlet bell diameter Refer to the section 211

middot Determine the bell-floor clearance 05D is a good preliminary design value

middot Determine the required bell submergence using the equation 21

middot Determine the minimum allowable liquid depth in the intake structure from

the sum of the floor clearance and the required bell submergence

middot Check bottom elevation near the entrance to the structure and determine if it

is necessary to slope the floor upstream of the bay entrance If the resulting

depth at the entrance to the intake structure is shallow then check to ensure

that gravity-driven flow is not restricted by the entrance condition

middot Check the pump bay velocity for the maximum single-pump flow and

minimum liquid depth with the bay width set to 2D If bay velocity exceeds

05 ms then increase the bay width to reduce the maximum velocity below

05 ms

middot If it is necessary to increase the pump bay width to greater than 2D then

decrease bay width in the vicinity of the pumps according to Fig 23

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

10

middot Compare cross-flow velocity (at maximum system flow) to average pump

bay velocity If cross-flow value exceeds 50 of the bay velocity a

physical hydraulic model study is necessary

middot Determine the length of the structure and dividing walls giving

consideration to minimum allowable distances to a sloping floor screening

equipment and length of dividing walls if dual flow traveling screens or

drum screens are to be used a physical hydraulic model study is required

middot If the final selected pump bell diameter and inlet velocity is within the

range given in section 211 then the sump dimensions (developed based on

the inlet bell design diameter) need not be changed and will comply with

these standards

22 Model test

The approach flow pattern in a pump intake facility is difficult to predict by

traditional mathematics or empirical formulae For large structures or those that

differ significantly from proven designs a model study is the only means to ensure

success A model study is used to identify adverse hydraulic conditions and

derive remedial measures for approach flow patterns generated by structures

upstream of the pump impeller Current model studies are not intended to

investigate flow patterns induced by the pump itself or the flow patterns within the

pump The objective of a model study is to ensure that the pump intake structure

generates favorable flow conditions at the inlet to the pump Evaluation for the

requirement of model test if

a Sump or piping geometry that deviates from this design standard

b Non-uniform or non-symmetric approach flow to the pump sump

exist

c The pumps have flows greater than 2520 ls per pump or the total

station flow with all pumps running would be greater than 6310 ls

d Situations that need ten times cost than the model study

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

11

Adverse hydraulic conditions that can affect pump performance include free and

submerged vortices swirl approaching the pump impeller flow separation at the

pump bell and a non-uniform axial velocity distribution at the suction

The purpose of model test is to examine these adverse hydraulic conditions and to

investigate measure against them

221 The scale effects

One possible difficulty in scale modeling is that it is impossible to reduce all

pertinent forces by the same factor With Froude-scale models the inertial and

gravitational forces are reduced similarly but the viscous and surface tension forces

cannot be simultaneously reduced The extra influence of these forces is known as

a ldquoscale effectrdquo A number of investigations have indicated that the scale effects for

the model are avoided if the Reynolds number based on the inlet flow and

submergence or intake diameter exceeds approximately 3x 104 and the Weber

number exceeds 150 [19]

In most engineering applications involving closed-conduit and open-channel flow

the Reynolds number limits are far exceeded and the flows are fully turbulent

In a sump model Froude number (Fr) and Reynolds number (Re) are the most

important non-dimensional parameters Some reduction in Re should be

introduced to the sump model However flow patterns are generally very similar

at high Re

No specific geometric scale rate is recommended but reasonable large

geometric scale to minimize viscous and surface tension scale effects and reproduce

the flow pattern

Re= uDν (22)

We= u2ρDσ (23)

where

u average axial velocity (ms)

ν kinematic viscosity of the liquid (m2s)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

12

ρ liquid density (kgm3)

σ surface tension of liquidair interface

According to the ANSIHI 98 the viscous scale effects on vortex may be negligible

if the resulting dimensionless numbers must meet these values Re>6times104 and

We>240

To allow practical visual observations of flow patterns accurate measurements of

swirl and velocity distribution and sufficient dimensional control the model scale

shall yield a bay width of at least 300mm a minimum liquid depth of at least

150mm and a pump throat or suction diameter of at least 80mm

222 Principle of similarity

In the similitude analysis the geometric and flow similarity requires the model and

the flow to be identical as the real model and flow This is to ensure the result

obtained from the model study is well presented in order to predict full scale

behavior Model involving a free surface are operated using Froude similarity since

the flow process is controlled by gravity and inertial forces

For similarity of the flow patterns the Froude number shall be equal in model and

prototype

Fr =

=

(24)

and as the scale ratio λ was selected in section 221 the following equation is

obtained

V = V times (

) = V times λ or Q = Q times (

) = Q times λ (25)

where Fr is the Froude number λ is the scale ratio of model dimensions to prototype

The test of 15 times Froude scaled flow should be conducted to ascertain the

potential scale affect on free surface vortices

Models of closed conduit piping system leading to a pump suction are not operated

based on Froude similitude However flow patterns are generally very similar at

high Re A minimum value of 1times105 for the Re at the pump suction is

recommended to ensure the flow patterns are correctly scaled

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

13

223 Preliminary model test

The initial design shall be tested first to identify any hydraulic problems

1) Vortex Formation Mechanisms

There are various investigations that provide insight into the fundamental processes

leading to the development of vortices both in experiment and numerical simulation

Shin et al [20] demonstrated that two basic mechanisms lead to inlet vortex

formation The first mechanism involves the development of an inlet vortex due to

the amplification of ambient vorticity in the approach flow as vortex lines

are convicted into the inlet The second mechanism involves the development of a

trailing vortex in the vicinity of the intake as a result of the variation in

circulation along the inlet For this second case a vortex can develop in a flow that

is irrotational upstream and the vortex development therefore does not depend

on the presence of ambient vorticity Shin et al investigation on kinematic

parameters indicate that the strength of an inlet-vortex or trailing vortex system

increases with decreasing distance from the surface However for an inlet in

an upstream irrotational flow two counter rotating vortices can still trail from

the rear of the inlet

2) Classification of vortex type

As illustrated in Fig 24 vortices in the vicinity of pump intakes may be adjacent to

the channel bottom or a channel wall (ie submerged vortices) or they may appear

adjacent to the free surface (ie free surface vortex)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

14

(a) Schematic diagram of vortices in a sump

(b) Experimental results of different vortex

Fig 24 Typical vortex in a sump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

15

For the model visual observation dye was injected around the intake structure near

the surface and submergence as Fig 24 (b) shows And the model test results of

vortex was classified as Fig 25

(a) Free Surface Vortices

(b) Submerged Vortices

Fig 25 Vortex classifications

There are a number of researches focused on the numerical theory to detect the

inception of a visible vortex As the diameters of vortices are usually much smaller

compared with the computing grid size to compensate for the lack of vortex

resolution in numerical simulation a stretching vortex model is applied to the local

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

16

flow field around semi-analytically identified vortex positions [21][22] The brief

explanations are as follows

a) Free surface vortex

The criteria expressed by equation 26 and 27 is applied to determine the

visible inception of a free surface vortex

≫ 1 (26)

=

gt 1 (27)

where P is the pressure drop at the vortex core and Ph is the static head at the

vortex element and f is the drag force acting at a unit length of water surface

vortex dimple based on Stokes approximation and caused by the down flow

Further f is the force acting on the bubble in the direction from the high

pressure region to the low pressure region like buoyancy

b) Submerged vortex

Equation 28 is applied to determine the visible inception of a submerged vortex

infin =

infin

infin gt 1 (28)

where is the static pressure at the vortex core infin is the atmospheric pressure

and is the critical cavitation inception pressure and nearly equal to the

saturated vapor pressure

3) Swirl angle

Swirl angle predicts the intensity of flow rotation Fig 26 shows experiment

calculation method with swirl meter and attempts of the CFD calculation method

were conducted as Fig 27 A four blade zero-pitch swirl meter which is supported

by a low friction bearing will be installed at about four suction pipe diameter (d)

downstream from the pump suction to measure swirl angle of pump approaching

flow For the swirl meter the tip to tip blade diameter is 075d and the length in

flow direction is 06 and one of the four blade is painted yellow as a reference to

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

17

count revolutions As flow swirl is usually unsteady the observation time of swirl

meter reading should be a continuous period of time

Fig 26 Typical swirl meter

(29)

(210)

(211)

Where qV flow rotational speed at swirl meter (ms)

zV average axial velocity at swirl meter (ms)

n Number of revolution per minute (rpm)

d Diameter of pipe at the swirl meter (m)

q flow rate (m3s)

zV

Vqq 1tan -=

)(60

smnd

Vtimestimes

=p

q

)(4

2

smd

qVz

times=

p

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

18

For the swirl angle calculating method with CFD the key point is to obtain the

average tangential velocity so the swirl check circles were created as illustrated in

Fig 27 The swirl check circle is located at the section of 4d height with diameter

of 025d 05d and 075d respectively

Swirl check circle

Fig 27 Schematic view of swirl angle numerical calculation

23

1 7501 501 250

acute++

=aringaringaring

N

i

N

i

NNVNVNV

Vi

q (212)

where

qV average tangential velocity (ms)

V circumferential velocity of check point in check circle

respectively (ms)

N check point total numbers in single check circle

4) Velocity

Velocity measurement at bell throat by Pitot tube was implemented by ANSIHI

method Typical Pitot tube configuration is shown in Fig 28 Eight measuring

points are defined as Fig 29 Time averaged velocity at eight points was obtained

and these eight data were again averaged to give throat sectional spatial velocity

average

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

19

Fig 28 Example of Pitot tube Fig 29 Measuring points at bell throat

224 Test criteria

There is the criteria for satisfactory performance of the sump model test and

summarized in Table 23

1) Free surface and submerged vortices

Free surface vortices entering the pump must be less severe than vortices with

coherent core Dye core vortices may be acceptable only if they occur less than 10

of the time or only for infrequent pump operating conditions

Submerged vortex which starts from the sump walls induce vibration and noise

problems Dye core submerged vortices extending into the pump bell mouth is not

allowed

2) Swirl angles

Swirl angle is used to check the flow rotation in the suction pipe The calculation

method is referred to section 223

Short-term maximum and long-term average values must be less than 5 degrees

Maximum short-term swirl angles up to 7 degrees may be acceptable only if they

occur less than 10 of the time or only for infrequent pump operating conditions

3)Velocity

Each point time-averaged velocity shall not be larger than 10 of sectional average

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

20

Standard deviation of time-varying velocity at each point was obtained and its

value shall not be larger than 10 of the time-averaged velocity at each point

1) Time-averaged

Variation[] = 100acute-

a

a

V

VV aV Averaged at section

2) Time-varying

Standard deviation= )1(

)(1i

2

-

-aring=

n

xxn

i

Variation[] = (Standard deviation X ) x 100

X Averaged at point

Table 23 Permission criteria of ANSIHI

Vortex occurence Swirl angle

Free surface vortex Submerged vortex

Observation

time Above 10 minutes Above 10 minutes

Above 10

minutes

Permission

criteria

Dye-core

Vortex(F3)

below 110 total

time

Dye-core

Vortex(S2)

below 110 total

time

Below

5degConditional

intermittent

below 7deg

225 Remedial measures for problem intakes

Usually in the preliminary test the worst flow condition is simulated And based on

the results remedial measures for problem intakes are discussed in this part

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

21

1) Overall approach flow patterns

The characteristics of flow approaching an intake structure is one of the foremost

considerations for the designer It is difficult to predict the effects of a given set of

flow conditions upstream from an intake structure Fig 210 depicts several typical

approach flow patterns of pump sump with different operating conditions

Fig 210 Examples of approach flow patterns

Generally speaking for the intake problems especially for the multiple pumps

sump there are the main measures using dividing walls placed between the pumps

as a partitioned structure trash racks with elongated bars guide piers or a number

of structural columns at the bay entrance can provide some assistance in controlling

cross-flow baffles flow distributors and screen mesh etc are applied to improve

the flow distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

22

2) Problems around pump intake

a) Free surface vortices

There are a number of factors that have an effect on the formation of surface

vortices To achieve a higher degree of certainty that objectionable surface vortices

do not form modifications can be made to intake structure to allow operation at

practical depths of submergence and the minimum submergence is referred in Table

22 the use of suction umbrellas vertical curtain walls and horizontal gratings

b) Submerged vortices

In the vicinity of intake bell there are the most complicated flow patterns and flow

makes the most changes in direction while maintaining a constant acceleration into

the pump bell to prevent local flow separation turbulence and vortex formation

Fig 211 represents a sample of various devices to inhibit the different submerged

vortex

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

23

Fig 211 Methods to reduce submerged vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

24

c) Preswirl

The approach flow distributions determine whether flow preswirl exists and a

sufficiently laterally skewed approach flow cause rotating around the intake bell

The most effective way of reducing preswirl is to establish a relatively uniform

approach flow within each pump bay by using the baffling schemes controlling

cross-flow and expanding concentrated flow discussed in the above part

To prevent non-uniform velocity distribution swirls and vortices necessary

modifications are conducted such as fore bay length baffle wall distributors

curtain wall mesh screen guide vane floor splitter wall fillet etc As the above

measures applied modification test will be conducted to check all the flow

conditions and compare with the preliminary results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

25

Chapter 3 Methodology

CFD (Computational fluid dynamics) is the technique which can design cycles to

better performance and reduce costs and time In CFD we solve the governing

equations of given physics (may be differential form or integral form) using some

numerical techniques like Finite Difference Method (FDM) Finite Element Method

(FEM) or Finite Volume Method (FVM) Although CFD plays a great part in the

design of turbomachinery CFD is not a replacement for experimental or analytical

approach In this study the numerical analysis was performed by the commercial

code ANSYS CFX 130 All aspects of the CFD study including modelling

meshing solving and post-processing of the results

31 Numerical modeling

ANSYS CFX solves the unsteady Navier-Stokes equations in their conservation

form The instantaneous equation of mass (continuity) in the stationary frame is

expressed as equation

ρ

δ + nabla ∙ (ρU) = 0 (31)

And the instantaneous equation for momentum is expressed as shown in equation

ρ

δ + nabla ∙ (ρU⨂U) = minusnablap + nabla ∙ τ + S (32)

Where τ is the stress tensor ( including both normal and shear components of the

stress)

These instantaneous equations are averaged for turbulent flows leading to additional

terms that need to be solved While the Navier-Stokes equations describe both

laminar and turbulent flows without addition terms realistic flows involve length

scales much smaller than the smallest finite volume mesh A Direct Numerical

Simulation of these flows would require significantly more computing power than

what is available now or in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

26

Therefore much research has been done to predict the effects of turbulence by

using turbulence models These models account for the effects of turbulence

without the use of a very fine mesh or direct numerical simulation

These turbulence models modify the transport equations by adding averaged and

fluctuating components The transport equations are changed to equations (33) and

(34)

ρ

δ + nabla ∙ (ρU) = 0 (33)

ρ

+ nabla ∙ ρ U ⨂U = minusnablap + nabla ∙ τ minus ρu⨂u + S (34)

The mass equation is not changed but the momentum equation contains extra terms

which are the Reynolds stresses ρ u⨂u and the Reynolds flux ρ u ϕ These

Reynolds stresses need to be modeled by additional equations to obtain closure

Obtaining closure implies that there are a sufficient number of equations to solve for

all the unknowns including the Reynolds stresses and Reynolds fluxes

Various turbulence models provide various ways to obtain closure In this

investigation Shear Stress Transport (SST) model and k-ε model were utilized The

advantage of SST model is that combines the advantages of other turbulence models

(the k-ε Wilcox k-ω) The 3 turbulence models will be discussed briefly

32 Turbulence models

Turbulence models are necessary because we cannot afford big enough computers

to directly capture every scale of motion There are three modelling frameworks

Direct Numerical Simulation (DNS) Reynolds-Averaged Navier-Stokes

Equations(RANS) Large Eddy Simulation(LES)

As RANS method is applied in most productionseveral RANS turbulence models

differ in their usage of wall functions the number of additional variables solved for

and what these variables represent

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

27

Two equation turbulence models

Two-equation turbulence models are widely used in CFD as they give a good

compromise between computational power needed and accuracy The term lsquotwo

equationrsquo refers to the fact that these models solve for the velocity and length scales

using separate transport equations The turbulence velocity scale is obtained by

solving the transport equation The turbulent length scale is estimated from two

properties of the turbulence field namely the turbulent kinetic energy and the

dissipation rate The dissipation rate of the kinetic energy is obtained from its

transport equation The most widely used are k-ε and k-ω two equation models In

the next section the k-ε Wilcox k-ω BSL k-ω and SST models will be briefly

discussed

321 k-ε turbulence model

The k-epsilon model solves for two variables k the turbulent kinetic energy and

epsilon the rate of dissipation of kinetic energy Wall functions are used in this

model so the flow in the buffer region is not simulated The k-epsilon model is very

popular for industrial applications due to its good convergence rate and relatively

low memory requirements It does not very accurately compute flow fields that

exhibit adverse pressure gradients strong curvature to the flow or jet flow It does

perform well for external flow problems around complex geometries

The k-ε model introduces k (m2s2) as the turbulence kinetic energy and ε (m2s3) as

the turbulence eddy dissipation The continuity equation remains the same

ρ

+ nabla ∙ (ρ U) = 0 (35)

The momentum equation changes as shown by equation

δρ

δ + nabla ∙ (ρU⨂U) = minusnablapprime + nabla ∙ μ

nablaU + (nablaU) + S (36)

Where SM is the sum of body forces μeff is the effective viscosity accounting for

turbulence and prsquo is the modified pressure The k-ε model uses the concept of eddy

viscosity giving the equation for effective viscosity as shown by equation

μ

= μ + μ (37)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

28

μt is the turbulence viscosity is linked to the turbulence kinetic energy and

dissipation by the equation

μ = Cμρ

ε (38)

Where Cμ is a constant

The values for k and ε come from the differential transport equations for the

turbulence kinetic energy and the turbulence dissipation rate

The turbulence kinetic energy equation is given as equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus ρε (39)

The turbulence dissipation rate is given by equation

(ρε)

+ nabla ∙ (ρUε) = nabla ∙ μ +

μ

σε nablaε +

ε

(Cε (P + Pε ) minus Cε ρε) (310)

Where Cε1 Cε2 σk σε are constants

Pk is the turbulence production due to viscous forces and is modeled by the

equation

P = μ nablaU ∙ nablaU + nablaU minus

nabla ∙ U 3μ

nabla ∙ U + ρk (311)

If a buoyancy term is added to the previous equation if the full buoyancy model is

used

The buoyancy term Pkb is modeled as

P = minusμ

ρσρg ∙ nablaρ (312)

322 Wilcox k-ω turbulence model

The k-omega model is similar to k-epsilon instead however it solves for omega mdash

the specific rate of dissipation of kinetic energy It also uses wall functions and

therefore has comparable memory requirements It has more difficulty converging

and is quite sensitive to the initial guess at the solution Hence the k-epsilon model

is often used first to find an initial condition for solving the k-omega model The k-

omega model is useful in many cases where the k-epsilon model is not accurate

such as internal flows flows that exhibit strong curvature separated flows and jets

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

29

This model has an advantage over the k-ε model where it does not involve complex

linear damping functions for near wall calculations at low Reynolds numbers

The k-ω model assumes that the turbulence viscosity is related to the turbulence

kinetic energy k and the turbulent frequency ω by the equation

μ = ρ

ω (313)

The transport equation for k is given by the equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus βprimeρ k ω (314)

The transport equation for ω is shown as equation

(ρω)

δ + nabla ∙ (ρ U ω) = nabla ∙ μ +

μ

μω

nablaω + αω

P + Pω minus β ρ ω (315)

The production rate of turbulence (Pk) is calculated as shown previously in the k-ε

section

The recommended values for the constants in the above equations are

βprime = 009

α = 59

β = 0075

σ = 2

σω = 2

The Reynolds stress tensor ρ u⨂u is calculated by

minusρ u⨂u = μ nablaU + (nablaU) minus

part ρk + μ

nabla ∙ U (316)

323 Shear stress transport model

The disadvantage of the Wilcox model is the strong sensitivity to free-stream

conditions Therefore a blending of the k-ω model near the surface and the k-ε in

the outer region was made by Menter [23] which resulted in the formulation of the

BSL k-ω turbulence model It consists of a transformation of the k-ε model to a k-ω

formulation and subsequently adding the resulting equations The Wilcox model is

multiplied by a blending function F1 and the transformed k-ε by another function 1-

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

30

F1 F1 is a function of wall distance (being the value of one near the surface and zero

outside the boundary layer) Outside the boundary and on the edge of the boundary

layer the standard k-ε model is used [24]

SST model is a combination of the k-epsilon in the free stream and the k-omega

models near the walls It does not use wall functions and tends to be most accurate

when solving the flow near the wall The SST model does not always converge to

the solution quickly so the k-epsilon or k-omega models are often solved first to

give good initial conditions

The k-ω based Shear Stress Transport (SST) model of Menter was applied for

turbulence treatment The transport equations for the SST model are expressed

below where the turbulent kinetic energy lsquokrsquo and turbulent frequency or dissipation

per unit turbulent kinetic energy are computed by using the following relations

For Turbulence Kinetic Energy

kkPukt

kTkk Ntilde+Ntilde+-=Ntilde+

para

para)[()( nsnwb (317)

where Pk is the production limiter

For Specific Dissipation Rate

ww

swnsnbwaww

ww NtildeNtilde-+Ntilde+Ntilde+-=Ntilde+para

para

1)1(2])[()( 21

22 kFSut

T (318)

The first blending function F1 is calculated from

iumlthorn

iumlyacute

uuml

iumlicirc

iumliacute

igrave

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

uacuteuacuteucirc

ugrave

ecircecirceuml

eacute

dividedivideoslash

oumlccedilccedilegrave

aelig=

4

22

21

4500maxmintanh

yCD

k

yy

kF

kw

wrs

w

n

wb (319)

y is the distance to the nearest wall and v is the kinematic viscosity

divideoslash

oumlccedilegrave

aeligNtildeNtilde= -10

2 101

2max ww

rsww kCDk (320)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

31

And Kinematic eddy viscosity

)max( 21

1

SF

kT

wa

an = (321)

where F2 is a blending function which restricts the limiter to the wall boundary and

S is the invariant measure of the strain rate

uacuteuacute

ucirc

ugrave

ecircecirc

euml

eacute

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

dividedivideoslash

oumlccedilccedilegrave

aelig=

2

22

5002maxtanh

w

n

wb yy

kF (322)

33 Brief introduction of simulation process

All aspects of CFD study including modelling meshing solving and post-

processing of the results are briefly introduced in this chapter

331 Pre-processing technique

The speed and accuracy of simulation results depending fundamentally on meshing

and the ability to prepare model geometry correctly

The steps in pre-processing are explained as follows

1) Domain Definitions

The domain definition depends mainly on the physics involved in the problem and

objective of the problem

a) Geometry creation or geometry import

b) Geometry cleanup

i) Removing the parts not necessary for simulation under consideration

ii) Closing the gaps in the geometry

iii) Removing small surfaces curves to make meshing process simple

2)Grid generation

To solve the governing equations the numerical methods like Finite Difference

Finite Volume or Finite Element Method to divide the domain into small parts and

solve the equations on one part at a time are applied

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 12: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

ix

Abbreviations

AVD Anti vortex device

BEP Best efficiency point

CFD Computational fluid Dynamics

NPSH Net positive suction head

NPSHR Required net positive suction head

PIV Particle Image Velocimetry

RANS Reynolds Averaged Navier- Stokes

RPM Number of revolutions per minute

SST Shear Stress Transport

VOF Volume of Fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

1

Chapter 1 Introduction

11 Background

It is an accepted fact that quite a number of problems faced in a pumping station are

related to the design of sump or intake rather than pump design As an important

part of the pumping station sump is designed to provide uniform swirl and

entrained air free flow to the pump intake Undesirable flow conditions (entrained

air non-uniform flow distribution vortices etc) can reduce the pump efficiency

induce vibration noise and cavitation even lead to excessive bearing loads of

impeller and structure damage Due to quite complex nature of the flow although

the flow conditions causing pump problems are well established there are no

specific solutions to eliminate them There are some design guidelines for the

specific geometrical and hydraulic constraints of the pump sump for any project

the model study is the only tool for solving potential problems in new designs and

rectifying problems observed in existing installations However the scaled model

tests are expensive and time-consuming so alternative Computational Fluid

Dynamics (CFD) methods for evaluating sump performance have been developed

With rapid progress in CFD numerical simulation is regarded as an effective tool in

solving fluid problems in pump sumps

12 Previous study

Various researches show the flow conditions around the sump intake structure are

quite complicated Shyam et al [1] used the commercial code CFX to carry out

two-phase flow simulation to capture air entrainment The air entrainments and its

location were well captured and numerical results were in accordance with the

experimental investigation Matsui J[2] and Okamura T[3] et al conducted a serial

research of the benchmark according to the Turbomachinery Society of Japan (TSJ)

standard including various commercial codes comparison performed in the flow

simulation and comparison analysis between the numerical and experimental results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

2

The purpose of the benchmark is to determine the accuracy and reliability of the

CFD codes used in universities and the industrial field It is also intended to identify

and investigate the formation of air-entrained and submerged vortices near the

pump Tanweer et al[4] have performed a numerical analysis of a multiple intake

pump sump to confirm that the geometry of sump affects the flow within sump

greatly Rajendran et al[5] have made a numerical analysis for the flow

characteristics of a sump model with pump intake and good agreements were

achieved by comparing the numerical results with the experiments Iwano et al[6]

have introduced a numerical method for the submerged vortex by analyzing the

flow in the pump sump with and without baffle plates Lee et al[7] have conducted

the CFD analysis of a multi-intake pump sump model to check the flow uniformity

by predicting the location number and vorticity of the vortex For the CFD

application in mixed flow pump design and performance analysis Oh H W [8]

conducted a mixed flow pump design optimization with specific parameters and

then compared the numerical results with the experiments The procedure presented

can be used as a practical design and analysis guide for the general turbomachinery

Kim et al[9] carried out a mixed flow pump performance analysis with numerical

simulation and improved the hydraulic efficiency by the modification of the

geometry

As an undesirable phenomenon caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage The cavitation

phenomenon has been studied by various investigators Van et al[10] have

investigated the cavitation inception behavior of a mixed-flow pump impellers

model and the calculated results matched well with the experiments Okamura et

al[11] have evaluated two numerical cavitation prediction methods used in the

pump industry manufacturers and attempts were made to improve the cavitation

performance Li et al[12] have conducted the numerical analysis on the basis of the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

3

development liquidvapor interface tracking method to predict the cavitation

characteristics with a centrifugal pump impeller model

Moreover there are a number of correlated studies [13]~[17]

13 Methodology of study

This study will investigate the flow phenomena of two sump models

A scaled sump model is adopted to carry out numerical simulation and PIV

experiment The flow conditions are predicted and the effectiveness of devices for

free surface vortex is evaluated by using numerical code a single intake sump

model with a mixed flow pump installed is modeled and the hydraulic performance

characteristics for the head rise shaft power and pump efficiencies versus flow rate

are studied In addition flow phenomena (vortices cavitation etc) are analyzed

and the effectiveness of an anti-vortex device (AVD) for the submerged vortex is

evaluated

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

4

Chapter 2 Basic theory of sump model

Specific hydraulic phenomena have been confirmed that they can lead to certain

common operational problems The common of the main problems are summarized

as follows

a Free surface vortices

b Submerged vortices

c Swirl of flow entering the pump

d Non-uniform distribution of velocity at the pump impeller

e Entrained air or gas bubbles

These problems encountered in the pump sump will affect the pump

performance and significantly increase the operational and maintenance costs In

order to identify sources of particular problems and find practical solution for

it the usual approach is to conduct the laboratory experiments on a scaled

physical model Typical design objectives are to ensure that a pump station is

designed according to best practice and conforms to the requirements set out in the

American National Standard for Pump Intake Design[18] These standards restrict

the degree to which the above-mentioned undesirable flow patterns may be present

The negative impact of each of these hydraulic conditions varies with pump specific

speed and size Application of the standard to design a sump does not generate a

problem free sump but provides a basis for the initial design

As there are no specific guidelines or criteria for design of trouble free intakes the

most common solution to potential problems in new design and rectification of

problems observed in existing designs is to construct a scaled model in a laboratory

observe and investigate the flow characteristics and propose modifications to the

intake geometry There are additional devices in the form of floor splitters or cones

back wall splitters fillets surface beams guide vanes etc aimed at controlling the

vortex and swirl formation Considerable prior experience and ingenuity are

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

5

required to solve the problems by this method and the cost and time required are

also significant

The intake structure refers to the channel leading from the water source which

may be a river or reservoir and all installations downstream including the pump

column or the intake pipe (the suction tube portion of a vertical intake) the

approach channel (upstream of pump bay) and the pump bay (bounded by the floor

the back wall side walls dividers walls separating adjacent pump column) Usually

the upstream end of the pump column has an inlet attachment called the suction bell

The function of the intake is to supply an evenly distributed flow of water to the

pump suction bell

Several types of pump intake basins are available and ANSIHI 98 provides

design requirements for each Intake structures can be categorized as being

clear liquids or solids-bearing liquids For clear liquids intakes are further

classified into rectangular formed circular and trench types as well as suction

tanks and cans For solid-bearing liquids trench type and rectangular wet wells are

usually considered

In this paper only rectangular intake structure for clear liquid is studied

21 Pump intake design

For pumps to achieve their optimum hydraulic performance across all operating

conditions the flow at the impeller must meet specific hydraulic conditions The

ideal flow entering the pump inlet should be of a uniform velocity distribution

without rotation and stable over time When designing a sump to achieve favorable

inflow to the pump or suction pipe bell various sump dimensions relative to the size

of the bell are required Using these standards of the distances to reduce the

probability of the occurence of strong submerged vortices and free-surface vortices

211 Inlet bell diameter design

The geometry is generally defined in terms of the pump inlet diameter as shown in

Table 21 and Fig 21 The bell diameter can be estimated based on an inlet pipe

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

6

Table 21 Acceptable velocity ranges for inlet bell diameter(ANSIHI 98)

Flow Rate

Qls

Recommended Inlet Bell Design

Velocityms

Acceptable Velocity

Range ms

Q<315 V=17 06leVle27

315 leQ<1260 V=17 09leVle24

Q ge1260 V=17 12leVle21

Fig 21 Recommended inlet bell design diameter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

7

212 Recommended dimensions for a rectangular sump

As the selected bell diameter has been determined the proportions of the inlet

structure can be estimated from Fig 22 and Fig 23 Table 22 shows the

recommended dimensions for a rectangular sump

Fig 22 Recommended intake structure layout

Fig 23 Filler wall details for proper bay width

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

8

Table 22 Recommended dimensions for a rectangular sump

Dimension Variable

Description Recommended value

A Distance from the pump inlet bell

centerline to the entrance A=5D minimum

a Length of constricted bay section near

the pump inlet 25D minimum

B Distance from the back wall to the

pump inlet bell centerline 075D

C Distance between the inlet bell and

floor 03D~05D

D Inlet bell outside diameter refer to section 211

H Minimum liquid depth H=S+C

h Minimum height of constricted bay

section near the pump inlet bell hgtH or 25D

S Minimum pump inlet bell submergence S=D(10+23Fr)

W Pump inlet bay entrance width 2D minimum

w Constricted bay width near the pump

inlet bell w=2D

X Pump inlet bay length 5D minimum

Y Distance from pump inlet bell

centerline to the through-flow traveling screen

4D minimum

Z1 Distance from sump inlet bell centerline to diverging walls

5D minimum

Z2 Distance from inlet bell centerline to

sloping floor 5D minimum

α Angle of floor slope -10deg~ +10deg

β Angle of wall convergence 0deg~ +10deg

empty Angle of convergence from constricted

area to by walls 10deg maximum

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

9

FD=Vb(gD)05 (21)

where

FD Froude number (dimensionless)

Vb velocity at suction inlet=FlowArea based on D

D Outside diameter of bell or pipe inlet

g gravitational acceleration

There is the design sequence summary for rectangular intake structures

middot Consider the flow patterns and boundary geometry of the body of liquid

from which the pump station is to receive flow Compare with the approach

flow condition and determine if a hydraulic physical model study is

required

middot Determine the number and size of pumps required to satisfy the range of

operating conditions likely to be encountered

middot Identify pump inlet bell diameter Refer to the section 211

middot Determine the bell-floor clearance 05D is a good preliminary design value

middot Determine the required bell submergence using the equation 21

middot Determine the minimum allowable liquid depth in the intake structure from

the sum of the floor clearance and the required bell submergence

middot Check bottom elevation near the entrance to the structure and determine if it

is necessary to slope the floor upstream of the bay entrance If the resulting

depth at the entrance to the intake structure is shallow then check to ensure

that gravity-driven flow is not restricted by the entrance condition

middot Check the pump bay velocity for the maximum single-pump flow and

minimum liquid depth with the bay width set to 2D If bay velocity exceeds

05 ms then increase the bay width to reduce the maximum velocity below

05 ms

middot If it is necessary to increase the pump bay width to greater than 2D then

decrease bay width in the vicinity of the pumps according to Fig 23

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

10

middot Compare cross-flow velocity (at maximum system flow) to average pump

bay velocity If cross-flow value exceeds 50 of the bay velocity a

physical hydraulic model study is necessary

middot Determine the length of the structure and dividing walls giving

consideration to minimum allowable distances to a sloping floor screening

equipment and length of dividing walls if dual flow traveling screens or

drum screens are to be used a physical hydraulic model study is required

middot If the final selected pump bell diameter and inlet velocity is within the

range given in section 211 then the sump dimensions (developed based on

the inlet bell design diameter) need not be changed and will comply with

these standards

22 Model test

The approach flow pattern in a pump intake facility is difficult to predict by

traditional mathematics or empirical formulae For large structures or those that

differ significantly from proven designs a model study is the only means to ensure

success A model study is used to identify adverse hydraulic conditions and

derive remedial measures for approach flow patterns generated by structures

upstream of the pump impeller Current model studies are not intended to

investigate flow patterns induced by the pump itself or the flow patterns within the

pump The objective of a model study is to ensure that the pump intake structure

generates favorable flow conditions at the inlet to the pump Evaluation for the

requirement of model test if

a Sump or piping geometry that deviates from this design standard

b Non-uniform or non-symmetric approach flow to the pump sump

exist

c The pumps have flows greater than 2520 ls per pump or the total

station flow with all pumps running would be greater than 6310 ls

d Situations that need ten times cost than the model study

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

11

Adverse hydraulic conditions that can affect pump performance include free and

submerged vortices swirl approaching the pump impeller flow separation at the

pump bell and a non-uniform axial velocity distribution at the suction

The purpose of model test is to examine these adverse hydraulic conditions and to

investigate measure against them

221 The scale effects

One possible difficulty in scale modeling is that it is impossible to reduce all

pertinent forces by the same factor With Froude-scale models the inertial and

gravitational forces are reduced similarly but the viscous and surface tension forces

cannot be simultaneously reduced The extra influence of these forces is known as

a ldquoscale effectrdquo A number of investigations have indicated that the scale effects for

the model are avoided if the Reynolds number based on the inlet flow and

submergence or intake diameter exceeds approximately 3x 104 and the Weber

number exceeds 150 [19]

In most engineering applications involving closed-conduit and open-channel flow

the Reynolds number limits are far exceeded and the flows are fully turbulent

In a sump model Froude number (Fr) and Reynolds number (Re) are the most

important non-dimensional parameters Some reduction in Re should be

introduced to the sump model However flow patterns are generally very similar

at high Re

No specific geometric scale rate is recommended but reasonable large

geometric scale to minimize viscous and surface tension scale effects and reproduce

the flow pattern

Re= uDν (22)

We= u2ρDσ (23)

where

u average axial velocity (ms)

ν kinematic viscosity of the liquid (m2s)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

12

ρ liquid density (kgm3)

σ surface tension of liquidair interface

According to the ANSIHI 98 the viscous scale effects on vortex may be negligible

if the resulting dimensionless numbers must meet these values Re>6times104 and

We>240

To allow practical visual observations of flow patterns accurate measurements of

swirl and velocity distribution and sufficient dimensional control the model scale

shall yield a bay width of at least 300mm a minimum liquid depth of at least

150mm and a pump throat or suction diameter of at least 80mm

222 Principle of similarity

In the similitude analysis the geometric and flow similarity requires the model and

the flow to be identical as the real model and flow This is to ensure the result

obtained from the model study is well presented in order to predict full scale

behavior Model involving a free surface are operated using Froude similarity since

the flow process is controlled by gravity and inertial forces

For similarity of the flow patterns the Froude number shall be equal in model and

prototype

Fr =

=

(24)

and as the scale ratio λ was selected in section 221 the following equation is

obtained

V = V times (

) = V times λ or Q = Q times (

) = Q times λ (25)

where Fr is the Froude number λ is the scale ratio of model dimensions to prototype

The test of 15 times Froude scaled flow should be conducted to ascertain the

potential scale affect on free surface vortices

Models of closed conduit piping system leading to a pump suction are not operated

based on Froude similitude However flow patterns are generally very similar at

high Re A minimum value of 1times105 for the Re at the pump suction is

recommended to ensure the flow patterns are correctly scaled

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

13

223 Preliminary model test

The initial design shall be tested first to identify any hydraulic problems

1) Vortex Formation Mechanisms

There are various investigations that provide insight into the fundamental processes

leading to the development of vortices both in experiment and numerical simulation

Shin et al [20] demonstrated that two basic mechanisms lead to inlet vortex

formation The first mechanism involves the development of an inlet vortex due to

the amplification of ambient vorticity in the approach flow as vortex lines

are convicted into the inlet The second mechanism involves the development of a

trailing vortex in the vicinity of the intake as a result of the variation in

circulation along the inlet For this second case a vortex can develop in a flow that

is irrotational upstream and the vortex development therefore does not depend

on the presence of ambient vorticity Shin et al investigation on kinematic

parameters indicate that the strength of an inlet-vortex or trailing vortex system

increases with decreasing distance from the surface However for an inlet in

an upstream irrotational flow two counter rotating vortices can still trail from

the rear of the inlet

2) Classification of vortex type

As illustrated in Fig 24 vortices in the vicinity of pump intakes may be adjacent to

the channel bottom or a channel wall (ie submerged vortices) or they may appear

adjacent to the free surface (ie free surface vortex)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

14

(a) Schematic diagram of vortices in a sump

(b) Experimental results of different vortex

Fig 24 Typical vortex in a sump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

15

For the model visual observation dye was injected around the intake structure near

the surface and submergence as Fig 24 (b) shows And the model test results of

vortex was classified as Fig 25

(a) Free Surface Vortices

(b) Submerged Vortices

Fig 25 Vortex classifications

There are a number of researches focused on the numerical theory to detect the

inception of a visible vortex As the diameters of vortices are usually much smaller

compared with the computing grid size to compensate for the lack of vortex

resolution in numerical simulation a stretching vortex model is applied to the local

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

16

flow field around semi-analytically identified vortex positions [21][22] The brief

explanations are as follows

a) Free surface vortex

The criteria expressed by equation 26 and 27 is applied to determine the

visible inception of a free surface vortex

≫ 1 (26)

=

gt 1 (27)

where P is the pressure drop at the vortex core and Ph is the static head at the

vortex element and f is the drag force acting at a unit length of water surface

vortex dimple based on Stokes approximation and caused by the down flow

Further f is the force acting on the bubble in the direction from the high

pressure region to the low pressure region like buoyancy

b) Submerged vortex

Equation 28 is applied to determine the visible inception of a submerged vortex

infin =

infin

infin gt 1 (28)

where is the static pressure at the vortex core infin is the atmospheric pressure

and is the critical cavitation inception pressure and nearly equal to the

saturated vapor pressure

3) Swirl angle

Swirl angle predicts the intensity of flow rotation Fig 26 shows experiment

calculation method with swirl meter and attempts of the CFD calculation method

were conducted as Fig 27 A four blade zero-pitch swirl meter which is supported

by a low friction bearing will be installed at about four suction pipe diameter (d)

downstream from the pump suction to measure swirl angle of pump approaching

flow For the swirl meter the tip to tip blade diameter is 075d and the length in

flow direction is 06 and one of the four blade is painted yellow as a reference to

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

17

count revolutions As flow swirl is usually unsteady the observation time of swirl

meter reading should be a continuous period of time

Fig 26 Typical swirl meter

(29)

(210)

(211)

Where qV flow rotational speed at swirl meter (ms)

zV average axial velocity at swirl meter (ms)

n Number of revolution per minute (rpm)

d Diameter of pipe at the swirl meter (m)

q flow rate (m3s)

zV

Vqq 1tan -=

)(60

smnd

Vtimestimes

=p

q

)(4

2

smd

qVz

times=

p

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

18

For the swirl angle calculating method with CFD the key point is to obtain the

average tangential velocity so the swirl check circles were created as illustrated in

Fig 27 The swirl check circle is located at the section of 4d height with diameter

of 025d 05d and 075d respectively

Swirl check circle

Fig 27 Schematic view of swirl angle numerical calculation

23

1 7501 501 250

acute++

=aringaringaring

N

i

N

i

NNVNVNV

Vi

q (212)

where

qV average tangential velocity (ms)

V circumferential velocity of check point in check circle

respectively (ms)

N check point total numbers in single check circle

4) Velocity

Velocity measurement at bell throat by Pitot tube was implemented by ANSIHI

method Typical Pitot tube configuration is shown in Fig 28 Eight measuring

points are defined as Fig 29 Time averaged velocity at eight points was obtained

and these eight data were again averaged to give throat sectional spatial velocity

average

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

19

Fig 28 Example of Pitot tube Fig 29 Measuring points at bell throat

224 Test criteria

There is the criteria for satisfactory performance of the sump model test and

summarized in Table 23

1) Free surface and submerged vortices

Free surface vortices entering the pump must be less severe than vortices with

coherent core Dye core vortices may be acceptable only if they occur less than 10

of the time or only for infrequent pump operating conditions

Submerged vortex which starts from the sump walls induce vibration and noise

problems Dye core submerged vortices extending into the pump bell mouth is not

allowed

2) Swirl angles

Swirl angle is used to check the flow rotation in the suction pipe The calculation

method is referred to section 223

Short-term maximum and long-term average values must be less than 5 degrees

Maximum short-term swirl angles up to 7 degrees may be acceptable only if they

occur less than 10 of the time or only for infrequent pump operating conditions

3)Velocity

Each point time-averaged velocity shall not be larger than 10 of sectional average

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

20

Standard deviation of time-varying velocity at each point was obtained and its

value shall not be larger than 10 of the time-averaged velocity at each point

1) Time-averaged

Variation[] = 100acute-

a

a

V

VV aV Averaged at section

2) Time-varying

Standard deviation= )1(

)(1i

2

-

-aring=

n

xxn

i

Variation[] = (Standard deviation X ) x 100

X Averaged at point

Table 23 Permission criteria of ANSIHI

Vortex occurence Swirl angle

Free surface vortex Submerged vortex

Observation

time Above 10 minutes Above 10 minutes

Above 10

minutes

Permission

criteria

Dye-core

Vortex(F3)

below 110 total

time

Dye-core

Vortex(S2)

below 110 total

time

Below

5degConditional

intermittent

below 7deg

225 Remedial measures for problem intakes

Usually in the preliminary test the worst flow condition is simulated And based on

the results remedial measures for problem intakes are discussed in this part

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

21

1) Overall approach flow patterns

The characteristics of flow approaching an intake structure is one of the foremost

considerations for the designer It is difficult to predict the effects of a given set of

flow conditions upstream from an intake structure Fig 210 depicts several typical

approach flow patterns of pump sump with different operating conditions

Fig 210 Examples of approach flow patterns

Generally speaking for the intake problems especially for the multiple pumps

sump there are the main measures using dividing walls placed between the pumps

as a partitioned structure trash racks with elongated bars guide piers or a number

of structural columns at the bay entrance can provide some assistance in controlling

cross-flow baffles flow distributors and screen mesh etc are applied to improve

the flow distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

22

2) Problems around pump intake

a) Free surface vortices

There are a number of factors that have an effect on the formation of surface

vortices To achieve a higher degree of certainty that objectionable surface vortices

do not form modifications can be made to intake structure to allow operation at

practical depths of submergence and the minimum submergence is referred in Table

22 the use of suction umbrellas vertical curtain walls and horizontal gratings

b) Submerged vortices

In the vicinity of intake bell there are the most complicated flow patterns and flow

makes the most changes in direction while maintaining a constant acceleration into

the pump bell to prevent local flow separation turbulence and vortex formation

Fig 211 represents a sample of various devices to inhibit the different submerged

vortex

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

23

Fig 211 Methods to reduce submerged vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

24

c) Preswirl

The approach flow distributions determine whether flow preswirl exists and a

sufficiently laterally skewed approach flow cause rotating around the intake bell

The most effective way of reducing preswirl is to establish a relatively uniform

approach flow within each pump bay by using the baffling schemes controlling

cross-flow and expanding concentrated flow discussed in the above part

To prevent non-uniform velocity distribution swirls and vortices necessary

modifications are conducted such as fore bay length baffle wall distributors

curtain wall mesh screen guide vane floor splitter wall fillet etc As the above

measures applied modification test will be conducted to check all the flow

conditions and compare with the preliminary results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

25

Chapter 3 Methodology

CFD (Computational fluid dynamics) is the technique which can design cycles to

better performance and reduce costs and time In CFD we solve the governing

equations of given physics (may be differential form or integral form) using some

numerical techniques like Finite Difference Method (FDM) Finite Element Method

(FEM) or Finite Volume Method (FVM) Although CFD plays a great part in the

design of turbomachinery CFD is not a replacement for experimental or analytical

approach In this study the numerical analysis was performed by the commercial

code ANSYS CFX 130 All aspects of the CFD study including modelling

meshing solving and post-processing of the results

31 Numerical modeling

ANSYS CFX solves the unsteady Navier-Stokes equations in their conservation

form The instantaneous equation of mass (continuity) in the stationary frame is

expressed as equation

ρ

δ + nabla ∙ (ρU) = 0 (31)

And the instantaneous equation for momentum is expressed as shown in equation

ρ

δ + nabla ∙ (ρU⨂U) = minusnablap + nabla ∙ τ + S (32)

Where τ is the stress tensor ( including both normal and shear components of the

stress)

These instantaneous equations are averaged for turbulent flows leading to additional

terms that need to be solved While the Navier-Stokes equations describe both

laminar and turbulent flows without addition terms realistic flows involve length

scales much smaller than the smallest finite volume mesh A Direct Numerical

Simulation of these flows would require significantly more computing power than

what is available now or in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

26

Therefore much research has been done to predict the effects of turbulence by

using turbulence models These models account for the effects of turbulence

without the use of a very fine mesh or direct numerical simulation

These turbulence models modify the transport equations by adding averaged and

fluctuating components The transport equations are changed to equations (33) and

(34)

ρ

δ + nabla ∙ (ρU) = 0 (33)

ρ

+ nabla ∙ ρ U ⨂U = minusnablap + nabla ∙ τ minus ρu⨂u + S (34)

The mass equation is not changed but the momentum equation contains extra terms

which are the Reynolds stresses ρ u⨂u and the Reynolds flux ρ u ϕ These

Reynolds stresses need to be modeled by additional equations to obtain closure

Obtaining closure implies that there are a sufficient number of equations to solve for

all the unknowns including the Reynolds stresses and Reynolds fluxes

Various turbulence models provide various ways to obtain closure In this

investigation Shear Stress Transport (SST) model and k-ε model were utilized The

advantage of SST model is that combines the advantages of other turbulence models

(the k-ε Wilcox k-ω) The 3 turbulence models will be discussed briefly

32 Turbulence models

Turbulence models are necessary because we cannot afford big enough computers

to directly capture every scale of motion There are three modelling frameworks

Direct Numerical Simulation (DNS) Reynolds-Averaged Navier-Stokes

Equations(RANS) Large Eddy Simulation(LES)

As RANS method is applied in most productionseveral RANS turbulence models

differ in their usage of wall functions the number of additional variables solved for

and what these variables represent

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

27

Two equation turbulence models

Two-equation turbulence models are widely used in CFD as they give a good

compromise between computational power needed and accuracy The term lsquotwo

equationrsquo refers to the fact that these models solve for the velocity and length scales

using separate transport equations The turbulence velocity scale is obtained by

solving the transport equation The turbulent length scale is estimated from two

properties of the turbulence field namely the turbulent kinetic energy and the

dissipation rate The dissipation rate of the kinetic energy is obtained from its

transport equation The most widely used are k-ε and k-ω two equation models In

the next section the k-ε Wilcox k-ω BSL k-ω and SST models will be briefly

discussed

321 k-ε turbulence model

The k-epsilon model solves for two variables k the turbulent kinetic energy and

epsilon the rate of dissipation of kinetic energy Wall functions are used in this

model so the flow in the buffer region is not simulated The k-epsilon model is very

popular for industrial applications due to its good convergence rate and relatively

low memory requirements It does not very accurately compute flow fields that

exhibit adverse pressure gradients strong curvature to the flow or jet flow It does

perform well for external flow problems around complex geometries

The k-ε model introduces k (m2s2) as the turbulence kinetic energy and ε (m2s3) as

the turbulence eddy dissipation The continuity equation remains the same

ρ

+ nabla ∙ (ρ U) = 0 (35)

The momentum equation changes as shown by equation

δρ

δ + nabla ∙ (ρU⨂U) = minusnablapprime + nabla ∙ μ

nablaU + (nablaU) + S (36)

Where SM is the sum of body forces μeff is the effective viscosity accounting for

turbulence and prsquo is the modified pressure The k-ε model uses the concept of eddy

viscosity giving the equation for effective viscosity as shown by equation

μ

= μ + μ (37)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

28

μt is the turbulence viscosity is linked to the turbulence kinetic energy and

dissipation by the equation

μ = Cμρ

ε (38)

Where Cμ is a constant

The values for k and ε come from the differential transport equations for the

turbulence kinetic energy and the turbulence dissipation rate

The turbulence kinetic energy equation is given as equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus ρε (39)

The turbulence dissipation rate is given by equation

(ρε)

+ nabla ∙ (ρUε) = nabla ∙ μ +

μ

σε nablaε +

ε

(Cε (P + Pε ) minus Cε ρε) (310)

Where Cε1 Cε2 σk σε are constants

Pk is the turbulence production due to viscous forces and is modeled by the

equation

P = μ nablaU ∙ nablaU + nablaU minus

nabla ∙ U 3μ

nabla ∙ U + ρk (311)

If a buoyancy term is added to the previous equation if the full buoyancy model is

used

The buoyancy term Pkb is modeled as

P = minusμ

ρσρg ∙ nablaρ (312)

322 Wilcox k-ω turbulence model

The k-omega model is similar to k-epsilon instead however it solves for omega mdash

the specific rate of dissipation of kinetic energy It also uses wall functions and

therefore has comparable memory requirements It has more difficulty converging

and is quite sensitive to the initial guess at the solution Hence the k-epsilon model

is often used first to find an initial condition for solving the k-omega model The k-

omega model is useful in many cases where the k-epsilon model is not accurate

such as internal flows flows that exhibit strong curvature separated flows and jets

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

29

This model has an advantage over the k-ε model where it does not involve complex

linear damping functions for near wall calculations at low Reynolds numbers

The k-ω model assumes that the turbulence viscosity is related to the turbulence

kinetic energy k and the turbulent frequency ω by the equation

μ = ρ

ω (313)

The transport equation for k is given by the equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus βprimeρ k ω (314)

The transport equation for ω is shown as equation

(ρω)

δ + nabla ∙ (ρ U ω) = nabla ∙ μ +

μ

μω

nablaω + αω

P + Pω minus β ρ ω (315)

The production rate of turbulence (Pk) is calculated as shown previously in the k-ε

section

The recommended values for the constants in the above equations are

βprime = 009

α = 59

β = 0075

σ = 2

σω = 2

The Reynolds stress tensor ρ u⨂u is calculated by

minusρ u⨂u = μ nablaU + (nablaU) minus

part ρk + μ

nabla ∙ U (316)

323 Shear stress transport model

The disadvantage of the Wilcox model is the strong sensitivity to free-stream

conditions Therefore a blending of the k-ω model near the surface and the k-ε in

the outer region was made by Menter [23] which resulted in the formulation of the

BSL k-ω turbulence model It consists of a transformation of the k-ε model to a k-ω

formulation and subsequently adding the resulting equations The Wilcox model is

multiplied by a blending function F1 and the transformed k-ε by another function 1-

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

30

F1 F1 is a function of wall distance (being the value of one near the surface and zero

outside the boundary layer) Outside the boundary and on the edge of the boundary

layer the standard k-ε model is used [24]

SST model is a combination of the k-epsilon in the free stream and the k-omega

models near the walls It does not use wall functions and tends to be most accurate

when solving the flow near the wall The SST model does not always converge to

the solution quickly so the k-epsilon or k-omega models are often solved first to

give good initial conditions

The k-ω based Shear Stress Transport (SST) model of Menter was applied for

turbulence treatment The transport equations for the SST model are expressed

below where the turbulent kinetic energy lsquokrsquo and turbulent frequency or dissipation

per unit turbulent kinetic energy are computed by using the following relations

For Turbulence Kinetic Energy

kkPukt

kTkk Ntilde+Ntilde+-=Ntilde+

para

para)[()( nsnwb (317)

where Pk is the production limiter

For Specific Dissipation Rate

ww

swnsnbwaww

ww NtildeNtilde-+Ntilde+Ntilde+-=Ntilde+para

para

1)1(2])[()( 21

22 kFSut

T (318)

The first blending function F1 is calculated from

iumlthorn

iumlyacute

uuml

iumlicirc

iumliacute

igrave

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

uacuteuacuteucirc

ugrave

ecircecirceuml

eacute

dividedivideoslash

oumlccedilccedilegrave

aelig=

4

22

21

4500maxmintanh

yCD

k

yy

kF

kw

wrs

w

n

wb (319)

y is the distance to the nearest wall and v is the kinematic viscosity

divideoslash

oumlccedilegrave

aeligNtildeNtilde= -10

2 101

2max ww

rsww kCDk (320)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

31

And Kinematic eddy viscosity

)max( 21

1

SF

kT

wa

an = (321)

where F2 is a blending function which restricts the limiter to the wall boundary and

S is the invariant measure of the strain rate

uacuteuacute

ucirc

ugrave

ecircecirc

euml

eacute

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

dividedivideoslash

oumlccedilccedilegrave

aelig=

2

22

5002maxtanh

w

n

wb yy

kF (322)

33 Brief introduction of simulation process

All aspects of CFD study including modelling meshing solving and post-

processing of the results are briefly introduced in this chapter

331 Pre-processing technique

The speed and accuracy of simulation results depending fundamentally on meshing

and the ability to prepare model geometry correctly

The steps in pre-processing are explained as follows

1) Domain Definitions

The domain definition depends mainly on the physics involved in the problem and

objective of the problem

a) Geometry creation or geometry import

b) Geometry cleanup

i) Removing the parts not necessary for simulation under consideration

ii) Closing the gaps in the geometry

iii) Removing small surfaces curves to make meshing process simple

2)Grid generation

To solve the governing equations the numerical methods like Finite Difference

Finite Volume or Finite Element Method to divide the domain into small parts and

solve the equations on one part at a time are applied

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 13: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

1

Chapter 1 Introduction

11 Background

It is an accepted fact that quite a number of problems faced in a pumping station are

related to the design of sump or intake rather than pump design As an important

part of the pumping station sump is designed to provide uniform swirl and

entrained air free flow to the pump intake Undesirable flow conditions (entrained

air non-uniform flow distribution vortices etc) can reduce the pump efficiency

induce vibration noise and cavitation even lead to excessive bearing loads of

impeller and structure damage Due to quite complex nature of the flow although

the flow conditions causing pump problems are well established there are no

specific solutions to eliminate them There are some design guidelines for the

specific geometrical and hydraulic constraints of the pump sump for any project

the model study is the only tool for solving potential problems in new designs and

rectifying problems observed in existing installations However the scaled model

tests are expensive and time-consuming so alternative Computational Fluid

Dynamics (CFD) methods for evaluating sump performance have been developed

With rapid progress in CFD numerical simulation is regarded as an effective tool in

solving fluid problems in pump sumps

12 Previous study

Various researches show the flow conditions around the sump intake structure are

quite complicated Shyam et al [1] used the commercial code CFX to carry out

two-phase flow simulation to capture air entrainment The air entrainments and its

location were well captured and numerical results were in accordance with the

experimental investigation Matsui J[2] and Okamura T[3] et al conducted a serial

research of the benchmark according to the Turbomachinery Society of Japan (TSJ)

standard including various commercial codes comparison performed in the flow

simulation and comparison analysis between the numerical and experimental results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

2

The purpose of the benchmark is to determine the accuracy and reliability of the

CFD codes used in universities and the industrial field It is also intended to identify

and investigate the formation of air-entrained and submerged vortices near the

pump Tanweer et al[4] have performed a numerical analysis of a multiple intake

pump sump to confirm that the geometry of sump affects the flow within sump

greatly Rajendran et al[5] have made a numerical analysis for the flow

characteristics of a sump model with pump intake and good agreements were

achieved by comparing the numerical results with the experiments Iwano et al[6]

have introduced a numerical method for the submerged vortex by analyzing the

flow in the pump sump with and without baffle plates Lee et al[7] have conducted

the CFD analysis of a multi-intake pump sump model to check the flow uniformity

by predicting the location number and vorticity of the vortex For the CFD

application in mixed flow pump design and performance analysis Oh H W [8]

conducted a mixed flow pump design optimization with specific parameters and

then compared the numerical results with the experiments The procedure presented

can be used as a practical design and analysis guide for the general turbomachinery

Kim et al[9] carried out a mixed flow pump performance analysis with numerical

simulation and improved the hydraulic efficiency by the modification of the

geometry

As an undesirable phenomenon caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage The cavitation

phenomenon has been studied by various investigators Van et al[10] have

investigated the cavitation inception behavior of a mixed-flow pump impellers

model and the calculated results matched well with the experiments Okamura et

al[11] have evaluated two numerical cavitation prediction methods used in the

pump industry manufacturers and attempts were made to improve the cavitation

performance Li et al[12] have conducted the numerical analysis on the basis of the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

3

development liquidvapor interface tracking method to predict the cavitation

characteristics with a centrifugal pump impeller model

Moreover there are a number of correlated studies [13]~[17]

13 Methodology of study

This study will investigate the flow phenomena of two sump models

A scaled sump model is adopted to carry out numerical simulation and PIV

experiment The flow conditions are predicted and the effectiveness of devices for

free surface vortex is evaluated by using numerical code a single intake sump

model with a mixed flow pump installed is modeled and the hydraulic performance

characteristics for the head rise shaft power and pump efficiencies versus flow rate

are studied In addition flow phenomena (vortices cavitation etc) are analyzed

and the effectiveness of an anti-vortex device (AVD) for the submerged vortex is

evaluated

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

4

Chapter 2 Basic theory of sump model

Specific hydraulic phenomena have been confirmed that they can lead to certain

common operational problems The common of the main problems are summarized

as follows

a Free surface vortices

b Submerged vortices

c Swirl of flow entering the pump

d Non-uniform distribution of velocity at the pump impeller

e Entrained air or gas bubbles

These problems encountered in the pump sump will affect the pump

performance and significantly increase the operational and maintenance costs In

order to identify sources of particular problems and find practical solution for

it the usual approach is to conduct the laboratory experiments on a scaled

physical model Typical design objectives are to ensure that a pump station is

designed according to best practice and conforms to the requirements set out in the

American National Standard for Pump Intake Design[18] These standards restrict

the degree to which the above-mentioned undesirable flow patterns may be present

The negative impact of each of these hydraulic conditions varies with pump specific

speed and size Application of the standard to design a sump does not generate a

problem free sump but provides a basis for the initial design

As there are no specific guidelines or criteria for design of trouble free intakes the

most common solution to potential problems in new design and rectification of

problems observed in existing designs is to construct a scaled model in a laboratory

observe and investigate the flow characteristics and propose modifications to the

intake geometry There are additional devices in the form of floor splitters or cones

back wall splitters fillets surface beams guide vanes etc aimed at controlling the

vortex and swirl formation Considerable prior experience and ingenuity are

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

5

required to solve the problems by this method and the cost and time required are

also significant

The intake structure refers to the channel leading from the water source which

may be a river or reservoir and all installations downstream including the pump

column or the intake pipe (the suction tube portion of a vertical intake) the

approach channel (upstream of pump bay) and the pump bay (bounded by the floor

the back wall side walls dividers walls separating adjacent pump column) Usually

the upstream end of the pump column has an inlet attachment called the suction bell

The function of the intake is to supply an evenly distributed flow of water to the

pump suction bell

Several types of pump intake basins are available and ANSIHI 98 provides

design requirements for each Intake structures can be categorized as being

clear liquids or solids-bearing liquids For clear liquids intakes are further

classified into rectangular formed circular and trench types as well as suction

tanks and cans For solid-bearing liquids trench type and rectangular wet wells are

usually considered

In this paper only rectangular intake structure for clear liquid is studied

21 Pump intake design

For pumps to achieve their optimum hydraulic performance across all operating

conditions the flow at the impeller must meet specific hydraulic conditions The

ideal flow entering the pump inlet should be of a uniform velocity distribution

without rotation and stable over time When designing a sump to achieve favorable

inflow to the pump or suction pipe bell various sump dimensions relative to the size

of the bell are required Using these standards of the distances to reduce the

probability of the occurence of strong submerged vortices and free-surface vortices

211 Inlet bell diameter design

The geometry is generally defined in terms of the pump inlet diameter as shown in

Table 21 and Fig 21 The bell diameter can be estimated based on an inlet pipe

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

6

Table 21 Acceptable velocity ranges for inlet bell diameter(ANSIHI 98)

Flow Rate

Qls

Recommended Inlet Bell Design

Velocityms

Acceptable Velocity

Range ms

Q<315 V=17 06leVle27

315 leQ<1260 V=17 09leVle24

Q ge1260 V=17 12leVle21

Fig 21 Recommended inlet bell design diameter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

7

212 Recommended dimensions for a rectangular sump

As the selected bell diameter has been determined the proportions of the inlet

structure can be estimated from Fig 22 and Fig 23 Table 22 shows the

recommended dimensions for a rectangular sump

Fig 22 Recommended intake structure layout

Fig 23 Filler wall details for proper bay width

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

8

Table 22 Recommended dimensions for a rectangular sump

Dimension Variable

Description Recommended value

A Distance from the pump inlet bell

centerline to the entrance A=5D minimum

a Length of constricted bay section near

the pump inlet 25D minimum

B Distance from the back wall to the

pump inlet bell centerline 075D

C Distance between the inlet bell and

floor 03D~05D

D Inlet bell outside diameter refer to section 211

H Minimum liquid depth H=S+C

h Minimum height of constricted bay

section near the pump inlet bell hgtH or 25D

S Minimum pump inlet bell submergence S=D(10+23Fr)

W Pump inlet bay entrance width 2D minimum

w Constricted bay width near the pump

inlet bell w=2D

X Pump inlet bay length 5D minimum

Y Distance from pump inlet bell

centerline to the through-flow traveling screen

4D minimum

Z1 Distance from sump inlet bell centerline to diverging walls

5D minimum

Z2 Distance from inlet bell centerline to

sloping floor 5D minimum

α Angle of floor slope -10deg~ +10deg

β Angle of wall convergence 0deg~ +10deg

empty Angle of convergence from constricted

area to by walls 10deg maximum

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

9

FD=Vb(gD)05 (21)

where

FD Froude number (dimensionless)

Vb velocity at suction inlet=FlowArea based on D

D Outside diameter of bell or pipe inlet

g gravitational acceleration

There is the design sequence summary for rectangular intake structures

middot Consider the flow patterns and boundary geometry of the body of liquid

from which the pump station is to receive flow Compare with the approach

flow condition and determine if a hydraulic physical model study is

required

middot Determine the number and size of pumps required to satisfy the range of

operating conditions likely to be encountered

middot Identify pump inlet bell diameter Refer to the section 211

middot Determine the bell-floor clearance 05D is a good preliminary design value

middot Determine the required bell submergence using the equation 21

middot Determine the minimum allowable liquid depth in the intake structure from

the sum of the floor clearance and the required bell submergence

middot Check bottom elevation near the entrance to the structure and determine if it

is necessary to slope the floor upstream of the bay entrance If the resulting

depth at the entrance to the intake structure is shallow then check to ensure

that gravity-driven flow is not restricted by the entrance condition

middot Check the pump bay velocity for the maximum single-pump flow and

minimum liquid depth with the bay width set to 2D If bay velocity exceeds

05 ms then increase the bay width to reduce the maximum velocity below

05 ms

middot If it is necessary to increase the pump bay width to greater than 2D then

decrease bay width in the vicinity of the pumps according to Fig 23

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

10

middot Compare cross-flow velocity (at maximum system flow) to average pump

bay velocity If cross-flow value exceeds 50 of the bay velocity a

physical hydraulic model study is necessary

middot Determine the length of the structure and dividing walls giving

consideration to minimum allowable distances to a sloping floor screening

equipment and length of dividing walls if dual flow traveling screens or

drum screens are to be used a physical hydraulic model study is required

middot If the final selected pump bell diameter and inlet velocity is within the

range given in section 211 then the sump dimensions (developed based on

the inlet bell design diameter) need not be changed and will comply with

these standards

22 Model test

The approach flow pattern in a pump intake facility is difficult to predict by

traditional mathematics or empirical formulae For large structures or those that

differ significantly from proven designs a model study is the only means to ensure

success A model study is used to identify adverse hydraulic conditions and

derive remedial measures for approach flow patterns generated by structures

upstream of the pump impeller Current model studies are not intended to

investigate flow patterns induced by the pump itself or the flow patterns within the

pump The objective of a model study is to ensure that the pump intake structure

generates favorable flow conditions at the inlet to the pump Evaluation for the

requirement of model test if

a Sump or piping geometry that deviates from this design standard

b Non-uniform or non-symmetric approach flow to the pump sump

exist

c The pumps have flows greater than 2520 ls per pump or the total

station flow with all pumps running would be greater than 6310 ls

d Situations that need ten times cost than the model study

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

11

Adverse hydraulic conditions that can affect pump performance include free and

submerged vortices swirl approaching the pump impeller flow separation at the

pump bell and a non-uniform axial velocity distribution at the suction

The purpose of model test is to examine these adverse hydraulic conditions and to

investigate measure against them

221 The scale effects

One possible difficulty in scale modeling is that it is impossible to reduce all

pertinent forces by the same factor With Froude-scale models the inertial and

gravitational forces are reduced similarly but the viscous and surface tension forces

cannot be simultaneously reduced The extra influence of these forces is known as

a ldquoscale effectrdquo A number of investigations have indicated that the scale effects for

the model are avoided if the Reynolds number based on the inlet flow and

submergence or intake diameter exceeds approximately 3x 104 and the Weber

number exceeds 150 [19]

In most engineering applications involving closed-conduit and open-channel flow

the Reynolds number limits are far exceeded and the flows are fully turbulent

In a sump model Froude number (Fr) and Reynolds number (Re) are the most

important non-dimensional parameters Some reduction in Re should be

introduced to the sump model However flow patterns are generally very similar

at high Re

No specific geometric scale rate is recommended but reasonable large

geometric scale to minimize viscous and surface tension scale effects and reproduce

the flow pattern

Re= uDν (22)

We= u2ρDσ (23)

where

u average axial velocity (ms)

ν kinematic viscosity of the liquid (m2s)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

12

ρ liquid density (kgm3)

σ surface tension of liquidair interface

According to the ANSIHI 98 the viscous scale effects on vortex may be negligible

if the resulting dimensionless numbers must meet these values Re>6times104 and

We>240

To allow practical visual observations of flow patterns accurate measurements of

swirl and velocity distribution and sufficient dimensional control the model scale

shall yield a bay width of at least 300mm a minimum liquid depth of at least

150mm and a pump throat or suction diameter of at least 80mm

222 Principle of similarity

In the similitude analysis the geometric and flow similarity requires the model and

the flow to be identical as the real model and flow This is to ensure the result

obtained from the model study is well presented in order to predict full scale

behavior Model involving a free surface are operated using Froude similarity since

the flow process is controlled by gravity and inertial forces

For similarity of the flow patterns the Froude number shall be equal in model and

prototype

Fr =

=

(24)

and as the scale ratio λ was selected in section 221 the following equation is

obtained

V = V times (

) = V times λ or Q = Q times (

) = Q times λ (25)

where Fr is the Froude number λ is the scale ratio of model dimensions to prototype

The test of 15 times Froude scaled flow should be conducted to ascertain the

potential scale affect on free surface vortices

Models of closed conduit piping system leading to a pump suction are not operated

based on Froude similitude However flow patterns are generally very similar at

high Re A minimum value of 1times105 for the Re at the pump suction is

recommended to ensure the flow patterns are correctly scaled

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

13

223 Preliminary model test

The initial design shall be tested first to identify any hydraulic problems

1) Vortex Formation Mechanisms

There are various investigations that provide insight into the fundamental processes

leading to the development of vortices both in experiment and numerical simulation

Shin et al [20] demonstrated that two basic mechanisms lead to inlet vortex

formation The first mechanism involves the development of an inlet vortex due to

the amplification of ambient vorticity in the approach flow as vortex lines

are convicted into the inlet The second mechanism involves the development of a

trailing vortex in the vicinity of the intake as a result of the variation in

circulation along the inlet For this second case a vortex can develop in a flow that

is irrotational upstream and the vortex development therefore does not depend

on the presence of ambient vorticity Shin et al investigation on kinematic

parameters indicate that the strength of an inlet-vortex or trailing vortex system

increases with decreasing distance from the surface However for an inlet in

an upstream irrotational flow two counter rotating vortices can still trail from

the rear of the inlet

2) Classification of vortex type

As illustrated in Fig 24 vortices in the vicinity of pump intakes may be adjacent to

the channel bottom or a channel wall (ie submerged vortices) or they may appear

adjacent to the free surface (ie free surface vortex)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

14

(a) Schematic diagram of vortices in a sump

(b) Experimental results of different vortex

Fig 24 Typical vortex in a sump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

15

For the model visual observation dye was injected around the intake structure near

the surface and submergence as Fig 24 (b) shows And the model test results of

vortex was classified as Fig 25

(a) Free Surface Vortices

(b) Submerged Vortices

Fig 25 Vortex classifications

There are a number of researches focused on the numerical theory to detect the

inception of a visible vortex As the diameters of vortices are usually much smaller

compared with the computing grid size to compensate for the lack of vortex

resolution in numerical simulation a stretching vortex model is applied to the local

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

16

flow field around semi-analytically identified vortex positions [21][22] The brief

explanations are as follows

a) Free surface vortex

The criteria expressed by equation 26 and 27 is applied to determine the

visible inception of a free surface vortex

≫ 1 (26)

=

gt 1 (27)

where P is the pressure drop at the vortex core and Ph is the static head at the

vortex element and f is the drag force acting at a unit length of water surface

vortex dimple based on Stokes approximation and caused by the down flow

Further f is the force acting on the bubble in the direction from the high

pressure region to the low pressure region like buoyancy

b) Submerged vortex

Equation 28 is applied to determine the visible inception of a submerged vortex

infin =

infin

infin gt 1 (28)

where is the static pressure at the vortex core infin is the atmospheric pressure

and is the critical cavitation inception pressure and nearly equal to the

saturated vapor pressure

3) Swirl angle

Swirl angle predicts the intensity of flow rotation Fig 26 shows experiment

calculation method with swirl meter and attempts of the CFD calculation method

were conducted as Fig 27 A four blade zero-pitch swirl meter which is supported

by a low friction bearing will be installed at about four suction pipe diameter (d)

downstream from the pump suction to measure swirl angle of pump approaching

flow For the swirl meter the tip to tip blade diameter is 075d and the length in

flow direction is 06 and one of the four blade is painted yellow as a reference to

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

17

count revolutions As flow swirl is usually unsteady the observation time of swirl

meter reading should be a continuous period of time

Fig 26 Typical swirl meter

(29)

(210)

(211)

Where qV flow rotational speed at swirl meter (ms)

zV average axial velocity at swirl meter (ms)

n Number of revolution per minute (rpm)

d Diameter of pipe at the swirl meter (m)

q flow rate (m3s)

zV

Vqq 1tan -=

)(60

smnd

Vtimestimes

=p

q

)(4

2

smd

qVz

times=

p

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

18

For the swirl angle calculating method with CFD the key point is to obtain the

average tangential velocity so the swirl check circles were created as illustrated in

Fig 27 The swirl check circle is located at the section of 4d height with diameter

of 025d 05d and 075d respectively

Swirl check circle

Fig 27 Schematic view of swirl angle numerical calculation

23

1 7501 501 250

acute++

=aringaringaring

N

i

N

i

NNVNVNV

Vi

q (212)

where

qV average tangential velocity (ms)

V circumferential velocity of check point in check circle

respectively (ms)

N check point total numbers in single check circle

4) Velocity

Velocity measurement at bell throat by Pitot tube was implemented by ANSIHI

method Typical Pitot tube configuration is shown in Fig 28 Eight measuring

points are defined as Fig 29 Time averaged velocity at eight points was obtained

and these eight data were again averaged to give throat sectional spatial velocity

average

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

19

Fig 28 Example of Pitot tube Fig 29 Measuring points at bell throat

224 Test criteria

There is the criteria for satisfactory performance of the sump model test and

summarized in Table 23

1) Free surface and submerged vortices

Free surface vortices entering the pump must be less severe than vortices with

coherent core Dye core vortices may be acceptable only if they occur less than 10

of the time or only for infrequent pump operating conditions

Submerged vortex which starts from the sump walls induce vibration and noise

problems Dye core submerged vortices extending into the pump bell mouth is not

allowed

2) Swirl angles

Swirl angle is used to check the flow rotation in the suction pipe The calculation

method is referred to section 223

Short-term maximum and long-term average values must be less than 5 degrees

Maximum short-term swirl angles up to 7 degrees may be acceptable only if they

occur less than 10 of the time or only for infrequent pump operating conditions

3)Velocity

Each point time-averaged velocity shall not be larger than 10 of sectional average

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

20

Standard deviation of time-varying velocity at each point was obtained and its

value shall not be larger than 10 of the time-averaged velocity at each point

1) Time-averaged

Variation[] = 100acute-

a

a

V

VV aV Averaged at section

2) Time-varying

Standard deviation= )1(

)(1i

2

-

-aring=

n

xxn

i

Variation[] = (Standard deviation X ) x 100

X Averaged at point

Table 23 Permission criteria of ANSIHI

Vortex occurence Swirl angle

Free surface vortex Submerged vortex

Observation

time Above 10 minutes Above 10 minutes

Above 10

minutes

Permission

criteria

Dye-core

Vortex(F3)

below 110 total

time

Dye-core

Vortex(S2)

below 110 total

time

Below

5degConditional

intermittent

below 7deg

225 Remedial measures for problem intakes

Usually in the preliminary test the worst flow condition is simulated And based on

the results remedial measures for problem intakes are discussed in this part

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

21

1) Overall approach flow patterns

The characteristics of flow approaching an intake structure is one of the foremost

considerations for the designer It is difficult to predict the effects of a given set of

flow conditions upstream from an intake structure Fig 210 depicts several typical

approach flow patterns of pump sump with different operating conditions

Fig 210 Examples of approach flow patterns

Generally speaking for the intake problems especially for the multiple pumps

sump there are the main measures using dividing walls placed between the pumps

as a partitioned structure trash racks with elongated bars guide piers or a number

of structural columns at the bay entrance can provide some assistance in controlling

cross-flow baffles flow distributors and screen mesh etc are applied to improve

the flow distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

22

2) Problems around pump intake

a) Free surface vortices

There are a number of factors that have an effect on the formation of surface

vortices To achieve a higher degree of certainty that objectionable surface vortices

do not form modifications can be made to intake structure to allow operation at

practical depths of submergence and the minimum submergence is referred in Table

22 the use of suction umbrellas vertical curtain walls and horizontal gratings

b) Submerged vortices

In the vicinity of intake bell there are the most complicated flow patterns and flow

makes the most changes in direction while maintaining a constant acceleration into

the pump bell to prevent local flow separation turbulence and vortex formation

Fig 211 represents a sample of various devices to inhibit the different submerged

vortex

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

23

Fig 211 Methods to reduce submerged vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

24

c) Preswirl

The approach flow distributions determine whether flow preswirl exists and a

sufficiently laterally skewed approach flow cause rotating around the intake bell

The most effective way of reducing preswirl is to establish a relatively uniform

approach flow within each pump bay by using the baffling schemes controlling

cross-flow and expanding concentrated flow discussed in the above part

To prevent non-uniform velocity distribution swirls and vortices necessary

modifications are conducted such as fore bay length baffle wall distributors

curtain wall mesh screen guide vane floor splitter wall fillet etc As the above

measures applied modification test will be conducted to check all the flow

conditions and compare with the preliminary results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

25

Chapter 3 Methodology

CFD (Computational fluid dynamics) is the technique which can design cycles to

better performance and reduce costs and time In CFD we solve the governing

equations of given physics (may be differential form or integral form) using some

numerical techniques like Finite Difference Method (FDM) Finite Element Method

(FEM) or Finite Volume Method (FVM) Although CFD plays a great part in the

design of turbomachinery CFD is not a replacement for experimental or analytical

approach In this study the numerical analysis was performed by the commercial

code ANSYS CFX 130 All aspects of the CFD study including modelling

meshing solving and post-processing of the results

31 Numerical modeling

ANSYS CFX solves the unsteady Navier-Stokes equations in their conservation

form The instantaneous equation of mass (continuity) in the stationary frame is

expressed as equation

ρ

δ + nabla ∙ (ρU) = 0 (31)

And the instantaneous equation for momentum is expressed as shown in equation

ρ

δ + nabla ∙ (ρU⨂U) = minusnablap + nabla ∙ τ + S (32)

Where τ is the stress tensor ( including both normal and shear components of the

stress)

These instantaneous equations are averaged for turbulent flows leading to additional

terms that need to be solved While the Navier-Stokes equations describe both

laminar and turbulent flows without addition terms realistic flows involve length

scales much smaller than the smallest finite volume mesh A Direct Numerical

Simulation of these flows would require significantly more computing power than

what is available now or in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

26

Therefore much research has been done to predict the effects of turbulence by

using turbulence models These models account for the effects of turbulence

without the use of a very fine mesh or direct numerical simulation

These turbulence models modify the transport equations by adding averaged and

fluctuating components The transport equations are changed to equations (33) and

(34)

ρ

δ + nabla ∙ (ρU) = 0 (33)

ρ

+ nabla ∙ ρ U ⨂U = minusnablap + nabla ∙ τ minus ρu⨂u + S (34)

The mass equation is not changed but the momentum equation contains extra terms

which are the Reynolds stresses ρ u⨂u and the Reynolds flux ρ u ϕ These

Reynolds stresses need to be modeled by additional equations to obtain closure

Obtaining closure implies that there are a sufficient number of equations to solve for

all the unknowns including the Reynolds stresses and Reynolds fluxes

Various turbulence models provide various ways to obtain closure In this

investigation Shear Stress Transport (SST) model and k-ε model were utilized The

advantage of SST model is that combines the advantages of other turbulence models

(the k-ε Wilcox k-ω) The 3 turbulence models will be discussed briefly

32 Turbulence models

Turbulence models are necessary because we cannot afford big enough computers

to directly capture every scale of motion There are three modelling frameworks

Direct Numerical Simulation (DNS) Reynolds-Averaged Navier-Stokes

Equations(RANS) Large Eddy Simulation(LES)

As RANS method is applied in most productionseveral RANS turbulence models

differ in their usage of wall functions the number of additional variables solved for

and what these variables represent

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

27

Two equation turbulence models

Two-equation turbulence models are widely used in CFD as they give a good

compromise between computational power needed and accuracy The term lsquotwo

equationrsquo refers to the fact that these models solve for the velocity and length scales

using separate transport equations The turbulence velocity scale is obtained by

solving the transport equation The turbulent length scale is estimated from two

properties of the turbulence field namely the turbulent kinetic energy and the

dissipation rate The dissipation rate of the kinetic energy is obtained from its

transport equation The most widely used are k-ε and k-ω two equation models In

the next section the k-ε Wilcox k-ω BSL k-ω and SST models will be briefly

discussed

321 k-ε turbulence model

The k-epsilon model solves for two variables k the turbulent kinetic energy and

epsilon the rate of dissipation of kinetic energy Wall functions are used in this

model so the flow in the buffer region is not simulated The k-epsilon model is very

popular for industrial applications due to its good convergence rate and relatively

low memory requirements It does not very accurately compute flow fields that

exhibit adverse pressure gradients strong curvature to the flow or jet flow It does

perform well for external flow problems around complex geometries

The k-ε model introduces k (m2s2) as the turbulence kinetic energy and ε (m2s3) as

the turbulence eddy dissipation The continuity equation remains the same

ρ

+ nabla ∙ (ρ U) = 0 (35)

The momentum equation changes as shown by equation

δρ

δ + nabla ∙ (ρU⨂U) = minusnablapprime + nabla ∙ μ

nablaU + (nablaU) + S (36)

Where SM is the sum of body forces μeff is the effective viscosity accounting for

turbulence and prsquo is the modified pressure The k-ε model uses the concept of eddy

viscosity giving the equation for effective viscosity as shown by equation

μ

= μ + μ (37)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

28

μt is the turbulence viscosity is linked to the turbulence kinetic energy and

dissipation by the equation

μ = Cμρ

ε (38)

Where Cμ is a constant

The values for k and ε come from the differential transport equations for the

turbulence kinetic energy and the turbulence dissipation rate

The turbulence kinetic energy equation is given as equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus ρε (39)

The turbulence dissipation rate is given by equation

(ρε)

+ nabla ∙ (ρUε) = nabla ∙ μ +

μ

σε nablaε +

ε

(Cε (P + Pε ) minus Cε ρε) (310)

Where Cε1 Cε2 σk σε are constants

Pk is the turbulence production due to viscous forces and is modeled by the

equation

P = μ nablaU ∙ nablaU + nablaU minus

nabla ∙ U 3μ

nabla ∙ U + ρk (311)

If a buoyancy term is added to the previous equation if the full buoyancy model is

used

The buoyancy term Pkb is modeled as

P = minusμ

ρσρg ∙ nablaρ (312)

322 Wilcox k-ω turbulence model

The k-omega model is similar to k-epsilon instead however it solves for omega mdash

the specific rate of dissipation of kinetic energy It also uses wall functions and

therefore has comparable memory requirements It has more difficulty converging

and is quite sensitive to the initial guess at the solution Hence the k-epsilon model

is often used first to find an initial condition for solving the k-omega model The k-

omega model is useful in many cases where the k-epsilon model is not accurate

such as internal flows flows that exhibit strong curvature separated flows and jets

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

29

This model has an advantage over the k-ε model where it does not involve complex

linear damping functions for near wall calculations at low Reynolds numbers

The k-ω model assumes that the turbulence viscosity is related to the turbulence

kinetic energy k and the turbulent frequency ω by the equation

μ = ρ

ω (313)

The transport equation for k is given by the equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus βprimeρ k ω (314)

The transport equation for ω is shown as equation

(ρω)

δ + nabla ∙ (ρ U ω) = nabla ∙ μ +

μ

μω

nablaω + αω

P + Pω minus β ρ ω (315)

The production rate of turbulence (Pk) is calculated as shown previously in the k-ε

section

The recommended values for the constants in the above equations are

βprime = 009

α = 59

β = 0075

σ = 2

σω = 2

The Reynolds stress tensor ρ u⨂u is calculated by

minusρ u⨂u = μ nablaU + (nablaU) minus

part ρk + μ

nabla ∙ U (316)

323 Shear stress transport model

The disadvantage of the Wilcox model is the strong sensitivity to free-stream

conditions Therefore a blending of the k-ω model near the surface and the k-ε in

the outer region was made by Menter [23] which resulted in the formulation of the

BSL k-ω turbulence model It consists of a transformation of the k-ε model to a k-ω

formulation and subsequently adding the resulting equations The Wilcox model is

multiplied by a blending function F1 and the transformed k-ε by another function 1-

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

30

F1 F1 is a function of wall distance (being the value of one near the surface and zero

outside the boundary layer) Outside the boundary and on the edge of the boundary

layer the standard k-ε model is used [24]

SST model is a combination of the k-epsilon in the free stream and the k-omega

models near the walls It does not use wall functions and tends to be most accurate

when solving the flow near the wall The SST model does not always converge to

the solution quickly so the k-epsilon or k-omega models are often solved first to

give good initial conditions

The k-ω based Shear Stress Transport (SST) model of Menter was applied for

turbulence treatment The transport equations for the SST model are expressed

below where the turbulent kinetic energy lsquokrsquo and turbulent frequency or dissipation

per unit turbulent kinetic energy are computed by using the following relations

For Turbulence Kinetic Energy

kkPukt

kTkk Ntilde+Ntilde+-=Ntilde+

para

para)[()( nsnwb (317)

where Pk is the production limiter

For Specific Dissipation Rate

ww

swnsnbwaww

ww NtildeNtilde-+Ntilde+Ntilde+-=Ntilde+para

para

1)1(2])[()( 21

22 kFSut

T (318)

The first blending function F1 is calculated from

iumlthorn

iumlyacute

uuml

iumlicirc

iumliacute

igrave

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

uacuteuacuteucirc

ugrave

ecircecirceuml

eacute

dividedivideoslash

oumlccedilccedilegrave

aelig=

4

22

21

4500maxmintanh

yCD

k

yy

kF

kw

wrs

w

n

wb (319)

y is the distance to the nearest wall and v is the kinematic viscosity

divideoslash

oumlccedilegrave

aeligNtildeNtilde= -10

2 101

2max ww

rsww kCDk (320)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

31

And Kinematic eddy viscosity

)max( 21

1

SF

kT

wa

an = (321)

where F2 is a blending function which restricts the limiter to the wall boundary and

S is the invariant measure of the strain rate

uacuteuacute

ucirc

ugrave

ecircecirc

euml

eacute

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

dividedivideoslash

oumlccedilccedilegrave

aelig=

2

22

5002maxtanh

w

n

wb yy

kF (322)

33 Brief introduction of simulation process

All aspects of CFD study including modelling meshing solving and post-

processing of the results are briefly introduced in this chapter

331 Pre-processing technique

The speed and accuracy of simulation results depending fundamentally on meshing

and the ability to prepare model geometry correctly

The steps in pre-processing are explained as follows

1) Domain Definitions

The domain definition depends mainly on the physics involved in the problem and

objective of the problem

a) Geometry creation or geometry import

b) Geometry cleanup

i) Removing the parts not necessary for simulation under consideration

ii) Closing the gaps in the geometry

iii) Removing small surfaces curves to make meshing process simple

2)Grid generation

To solve the governing equations the numerical methods like Finite Difference

Finite Volume or Finite Element Method to divide the domain into small parts and

solve the equations on one part at a time are applied

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 14: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

2

The purpose of the benchmark is to determine the accuracy and reliability of the

CFD codes used in universities and the industrial field It is also intended to identify

and investigate the formation of air-entrained and submerged vortices near the

pump Tanweer et al[4] have performed a numerical analysis of a multiple intake

pump sump to confirm that the geometry of sump affects the flow within sump

greatly Rajendran et al[5] have made a numerical analysis for the flow

characteristics of a sump model with pump intake and good agreements were

achieved by comparing the numerical results with the experiments Iwano et al[6]

have introduced a numerical method for the submerged vortex by analyzing the

flow in the pump sump with and without baffle plates Lee et al[7] have conducted

the CFD analysis of a multi-intake pump sump model to check the flow uniformity

by predicting the location number and vorticity of the vortex For the CFD

application in mixed flow pump design and performance analysis Oh H W [8]

conducted a mixed flow pump design optimization with specific parameters and

then compared the numerical results with the experiments The procedure presented

can be used as a practical design and analysis guide for the general turbomachinery

Kim et al[9] carried out a mixed flow pump performance analysis with numerical

simulation and improved the hydraulic efficiency by the modification of the

geometry

As an undesirable phenomenon caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage The cavitation

phenomenon has been studied by various investigators Van et al[10] have

investigated the cavitation inception behavior of a mixed-flow pump impellers

model and the calculated results matched well with the experiments Okamura et

al[11] have evaluated two numerical cavitation prediction methods used in the

pump industry manufacturers and attempts were made to improve the cavitation

performance Li et al[12] have conducted the numerical analysis on the basis of the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

3

development liquidvapor interface tracking method to predict the cavitation

characteristics with a centrifugal pump impeller model

Moreover there are a number of correlated studies [13]~[17]

13 Methodology of study

This study will investigate the flow phenomena of two sump models

A scaled sump model is adopted to carry out numerical simulation and PIV

experiment The flow conditions are predicted and the effectiveness of devices for

free surface vortex is evaluated by using numerical code a single intake sump

model with a mixed flow pump installed is modeled and the hydraulic performance

characteristics for the head rise shaft power and pump efficiencies versus flow rate

are studied In addition flow phenomena (vortices cavitation etc) are analyzed

and the effectiveness of an anti-vortex device (AVD) for the submerged vortex is

evaluated

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

4

Chapter 2 Basic theory of sump model

Specific hydraulic phenomena have been confirmed that they can lead to certain

common operational problems The common of the main problems are summarized

as follows

a Free surface vortices

b Submerged vortices

c Swirl of flow entering the pump

d Non-uniform distribution of velocity at the pump impeller

e Entrained air or gas bubbles

These problems encountered in the pump sump will affect the pump

performance and significantly increase the operational and maintenance costs In

order to identify sources of particular problems and find practical solution for

it the usual approach is to conduct the laboratory experiments on a scaled

physical model Typical design objectives are to ensure that a pump station is

designed according to best practice and conforms to the requirements set out in the

American National Standard for Pump Intake Design[18] These standards restrict

the degree to which the above-mentioned undesirable flow patterns may be present

The negative impact of each of these hydraulic conditions varies with pump specific

speed and size Application of the standard to design a sump does not generate a

problem free sump but provides a basis for the initial design

As there are no specific guidelines or criteria for design of trouble free intakes the

most common solution to potential problems in new design and rectification of

problems observed in existing designs is to construct a scaled model in a laboratory

observe and investigate the flow characteristics and propose modifications to the

intake geometry There are additional devices in the form of floor splitters or cones

back wall splitters fillets surface beams guide vanes etc aimed at controlling the

vortex and swirl formation Considerable prior experience and ingenuity are

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

5

required to solve the problems by this method and the cost and time required are

also significant

The intake structure refers to the channel leading from the water source which

may be a river or reservoir and all installations downstream including the pump

column or the intake pipe (the suction tube portion of a vertical intake) the

approach channel (upstream of pump bay) and the pump bay (bounded by the floor

the back wall side walls dividers walls separating adjacent pump column) Usually

the upstream end of the pump column has an inlet attachment called the suction bell

The function of the intake is to supply an evenly distributed flow of water to the

pump suction bell

Several types of pump intake basins are available and ANSIHI 98 provides

design requirements for each Intake structures can be categorized as being

clear liquids or solids-bearing liquids For clear liquids intakes are further

classified into rectangular formed circular and trench types as well as suction

tanks and cans For solid-bearing liquids trench type and rectangular wet wells are

usually considered

In this paper only rectangular intake structure for clear liquid is studied

21 Pump intake design

For pumps to achieve their optimum hydraulic performance across all operating

conditions the flow at the impeller must meet specific hydraulic conditions The

ideal flow entering the pump inlet should be of a uniform velocity distribution

without rotation and stable over time When designing a sump to achieve favorable

inflow to the pump or suction pipe bell various sump dimensions relative to the size

of the bell are required Using these standards of the distances to reduce the

probability of the occurence of strong submerged vortices and free-surface vortices

211 Inlet bell diameter design

The geometry is generally defined in terms of the pump inlet diameter as shown in

Table 21 and Fig 21 The bell diameter can be estimated based on an inlet pipe

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

6

Table 21 Acceptable velocity ranges for inlet bell diameter(ANSIHI 98)

Flow Rate

Qls

Recommended Inlet Bell Design

Velocityms

Acceptable Velocity

Range ms

Q<315 V=17 06leVle27

315 leQ<1260 V=17 09leVle24

Q ge1260 V=17 12leVle21

Fig 21 Recommended inlet bell design diameter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

7

212 Recommended dimensions for a rectangular sump

As the selected bell diameter has been determined the proportions of the inlet

structure can be estimated from Fig 22 and Fig 23 Table 22 shows the

recommended dimensions for a rectangular sump

Fig 22 Recommended intake structure layout

Fig 23 Filler wall details for proper bay width

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

8

Table 22 Recommended dimensions for a rectangular sump

Dimension Variable

Description Recommended value

A Distance from the pump inlet bell

centerline to the entrance A=5D minimum

a Length of constricted bay section near

the pump inlet 25D minimum

B Distance from the back wall to the

pump inlet bell centerline 075D

C Distance between the inlet bell and

floor 03D~05D

D Inlet bell outside diameter refer to section 211

H Minimum liquid depth H=S+C

h Minimum height of constricted bay

section near the pump inlet bell hgtH or 25D

S Minimum pump inlet bell submergence S=D(10+23Fr)

W Pump inlet bay entrance width 2D minimum

w Constricted bay width near the pump

inlet bell w=2D

X Pump inlet bay length 5D minimum

Y Distance from pump inlet bell

centerline to the through-flow traveling screen

4D minimum

Z1 Distance from sump inlet bell centerline to diverging walls

5D minimum

Z2 Distance from inlet bell centerline to

sloping floor 5D minimum

α Angle of floor slope -10deg~ +10deg

β Angle of wall convergence 0deg~ +10deg

empty Angle of convergence from constricted

area to by walls 10deg maximum

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

9

FD=Vb(gD)05 (21)

where

FD Froude number (dimensionless)

Vb velocity at suction inlet=FlowArea based on D

D Outside diameter of bell or pipe inlet

g gravitational acceleration

There is the design sequence summary for rectangular intake structures

middot Consider the flow patterns and boundary geometry of the body of liquid

from which the pump station is to receive flow Compare with the approach

flow condition and determine if a hydraulic physical model study is

required

middot Determine the number and size of pumps required to satisfy the range of

operating conditions likely to be encountered

middot Identify pump inlet bell diameter Refer to the section 211

middot Determine the bell-floor clearance 05D is a good preliminary design value

middot Determine the required bell submergence using the equation 21

middot Determine the minimum allowable liquid depth in the intake structure from

the sum of the floor clearance and the required bell submergence

middot Check bottom elevation near the entrance to the structure and determine if it

is necessary to slope the floor upstream of the bay entrance If the resulting

depth at the entrance to the intake structure is shallow then check to ensure

that gravity-driven flow is not restricted by the entrance condition

middot Check the pump bay velocity for the maximum single-pump flow and

minimum liquid depth with the bay width set to 2D If bay velocity exceeds

05 ms then increase the bay width to reduce the maximum velocity below

05 ms

middot If it is necessary to increase the pump bay width to greater than 2D then

decrease bay width in the vicinity of the pumps according to Fig 23

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

10

middot Compare cross-flow velocity (at maximum system flow) to average pump

bay velocity If cross-flow value exceeds 50 of the bay velocity a

physical hydraulic model study is necessary

middot Determine the length of the structure and dividing walls giving

consideration to minimum allowable distances to a sloping floor screening

equipment and length of dividing walls if dual flow traveling screens or

drum screens are to be used a physical hydraulic model study is required

middot If the final selected pump bell diameter and inlet velocity is within the

range given in section 211 then the sump dimensions (developed based on

the inlet bell design diameter) need not be changed and will comply with

these standards

22 Model test

The approach flow pattern in a pump intake facility is difficult to predict by

traditional mathematics or empirical formulae For large structures or those that

differ significantly from proven designs a model study is the only means to ensure

success A model study is used to identify adverse hydraulic conditions and

derive remedial measures for approach flow patterns generated by structures

upstream of the pump impeller Current model studies are not intended to

investigate flow patterns induced by the pump itself or the flow patterns within the

pump The objective of a model study is to ensure that the pump intake structure

generates favorable flow conditions at the inlet to the pump Evaluation for the

requirement of model test if

a Sump or piping geometry that deviates from this design standard

b Non-uniform or non-symmetric approach flow to the pump sump

exist

c The pumps have flows greater than 2520 ls per pump or the total

station flow with all pumps running would be greater than 6310 ls

d Situations that need ten times cost than the model study

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

11

Adverse hydraulic conditions that can affect pump performance include free and

submerged vortices swirl approaching the pump impeller flow separation at the

pump bell and a non-uniform axial velocity distribution at the suction

The purpose of model test is to examine these adverse hydraulic conditions and to

investigate measure against them

221 The scale effects

One possible difficulty in scale modeling is that it is impossible to reduce all

pertinent forces by the same factor With Froude-scale models the inertial and

gravitational forces are reduced similarly but the viscous and surface tension forces

cannot be simultaneously reduced The extra influence of these forces is known as

a ldquoscale effectrdquo A number of investigations have indicated that the scale effects for

the model are avoided if the Reynolds number based on the inlet flow and

submergence or intake diameter exceeds approximately 3x 104 and the Weber

number exceeds 150 [19]

In most engineering applications involving closed-conduit and open-channel flow

the Reynolds number limits are far exceeded and the flows are fully turbulent

In a sump model Froude number (Fr) and Reynolds number (Re) are the most

important non-dimensional parameters Some reduction in Re should be

introduced to the sump model However flow patterns are generally very similar

at high Re

No specific geometric scale rate is recommended but reasonable large

geometric scale to minimize viscous and surface tension scale effects and reproduce

the flow pattern

Re= uDν (22)

We= u2ρDσ (23)

where

u average axial velocity (ms)

ν kinematic viscosity of the liquid (m2s)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

12

ρ liquid density (kgm3)

σ surface tension of liquidair interface

According to the ANSIHI 98 the viscous scale effects on vortex may be negligible

if the resulting dimensionless numbers must meet these values Re>6times104 and

We>240

To allow practical visual observations of flow patterns accurate measurements of

swirl and velocity distribution and sufficient dimensional control the model scale

shall yield a bay width of at least 300mm a minimum liquid depth of at least

150mm and a pump throat or suction diameter of at least 80mm

222 Principle of similarity

In the similitude analysis the geometric and flow similarity requires the model and

the flow to be identical as the real model and flow This is to ensure the result

obtained from the model study is well presented in order to predict full scale

behavior Model involving a free surface are operated using Froude similarity since

the flow process is controlled by gravity and inertial forces

For similarity of the flow patterns the Froude number shall be equal in model and

prototype

Fr =

=

(24)

and as the scale ratio λ was selected in section 221 the following equation is

obtained

V = V times (

) = V times λ or Q = Q times (

) = Q times λ (25)

where Fr is the Froude number λ is the scale ratio of model dimensions to prototype

The test of 15 times Froude scaled flow should be conducted to ascertain the

potential scale affect on free surface vortices

Models of closed conduit piping system leading to a pump suction are not operated

based on Froude similitude However flow patterns are generally very similar at

high Re A minimum value of 1times105 for the Re at the pump suction is

recommended to ensure the flow patterns are correctly scaled

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

13

223 Preliminary model test

The initial design shall be tested first to identify any hydraulic problems

1) Vortex Formation Mechanisms

There are various investigations that provide insight into the fundamental processes

leading to the development of vortices both in experiment and numerical simulation

Shin et al [20] demonstrated that two basic mechanisms lead to inlet vortex

formation The first mechanism involves the development of an inlet vortex due to

the amplification of ambient vorticity in the approach flow as vortex lines

are convicted into the inlet The second mechanism involves the development of a

trailing vortex in the vicinity of the intake as a result of the variation in

circulation along the inlet For this second case a vortex can develop in a flow that

is irrotational upstream and the vortex development therefore does not depend

on the presence of ambient vorticity Shin et al investigation on kinematic

parameters indicate that the strength of an inlet-vortex or trailing vortex system

increases with decreasing distance from the surface However for an inlet in

an upstream irrotational flow two counter rotating vortices can still trail from

the rear of the inlet

2) Classification of vortex type

As illustrated in Fig 24 vortices in the vicinity of pump intakes may be adjacent to

the channel bottom or a channel wall (ie submerged vortices) or they may appear

adjacent to the free surface (ie free surface vortex)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

14

(a) Schematic diagram of vortices in a sump

(b) Experimental results of different vortex

Fig 24 Typical vortex in a sump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

15

For the model visual observation dye was injected around the intake structure near

the surface and submergence as Fig 24 (b) shows And the model test results of

vortex was classified as Fig 25

(a) Free Surface Vortices

(b) Submerged Vortices

Fig 25 Vortex classifications

There are a number of researches focused on the numerical theory to detect the

inception of a visible vortex As the diameters of vortices are usually much smaller

compared with the computing grid size to compensate for the lack of vortex

resolution in numerical simulation a stretching vortex model is applied to the local

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

16

flow field around semi-analytically identified vortex positions [21][22] The brief

explanations are as follows

a) Free surface vortex

The criteria expressed by equation 26 and 27 is applied to determine the

visible inception of a free surface vortex

≫ 1 (26)

=

gt 1 (27)

where P is the pressure drop at the vortex core and Ph is the static head at the

vortex element and f is the drag force acting at a unit length of water surface

vortex dimple based on Stokes approximation and caused by the down flow

Further f is the force acting on the bubble in the direction from the high

pressure region to the low pressure region like buoyancy

b) Submerged vortex

Equation 28 is applied to determine the visible inception of a submerged vortex

infin =

infin

infin gt 1 (28)

where is the static pressure at the vortex core infin is the atmospheric pressure

and is the critical cavitation inception pressure and nearly equal to the

saturated vapor pressure

3) Swirl angle

Swirl angle predicts the intensity of flow rotation Fig 26 shows experiment

calculation method with swirl meter and attempts of the CFD calculation method

were conducted as Fig 27 A four blade zero-pitch swirl meter which is supported

by a low friction bearing will be installed at about four suction pipe diameter (d)

downstream from the pump suction to measure swirl angle of pump approaching

flow For the swirl meter the tip to tip blade diameter is 075d and the length in

flow direction is 06 and one of the four blade is painted yellow as a reference to

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

17

count revolutions As flow swirl is usually unsteady the observation time of swirl

meter reading should be a continuous period of time

Fig 26 Typical swirl meter

(29)

(210)

(211)

Where qV flow rotational speed at swirl meter (ms)

zV average axial velocity at swirl meter (ms)

n Number of revolution per minute (rpm)

d Diameter of pipe at the swirl meter (m)

q flow rate (m3s)

zV

Vqq 1tan -=

)(60

smnd

Vtimestimes

=p

q

)(4

2

smd

qVz

times=

p

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

18

For the swirl angle calculating method with CFD the key point is to obtain the

average tangential velocity so the swirl check circles were created as illustrated in

Fig 27 The swirl check circle is located at the section of 4d height with diameter

of 025d 05d and 075d respectively

Swirl check circle

Fig 27 Schematic view of swirl angle numerical calculation

23

1 7501 501 250

acute++

=aringaringaring

N

i

N

i

NNVNVNV

Vi

q (212)

where

qV average tangential velocity (ms)

V circumferential velocity of check point in check circle

respectively (ms)

N check point total numbers in single check circle

4) Velocity

Velocity measurement at bell throat by Pitot tube was implemented by ANSIHI

method Typical Pitot tube configuration is shown in Fig 28 Eight measuring

points are defined as Fig 29 Time averaged velocity at eight points was obtained

and these eight data were again averaged to give throat sectional spatial velocity

average

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

19

Fig 28 Example of Pitot tube Fig 29 Measuring points at bell throat

224 Test criteria

There is the criteria for satisfactory performance of the sump model test and

summarized in Table 23

1) Free surface and submerged vortices

Free surface vortices entering the pump must be less severe than vortices with

coherent core Dye core vortices may be acceptable only if they occur less than 10

of the time or only for infrequent pump operating conditions

Submerged vortex which starts from the sump walls induce vibration and noise

problems Dye core submerged vortices extending into the pump bell mouth is not

allowed

2) Swirl angles

Swirl angle is used to check the flow rotation in the suction pipe The calculation

method is referred to section 223

Short-term maximum and long-term average values must be less than 5 degrees

Maximum short-term swirl angles up to 7 degrees may be acceptable only if they

occur less than 10 of the time or only for infrequent pump operating conditions

3)Velocity

Each point time-averaged velocity shall not be larger than 10 of sectional average

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

20

Standard deviation of time-varying velocity at each point was obtained and its

value shall not be larger than 10 of the time-averaged velocity at each point

1) Time-averaged

Variation[] = 100acute-

a

a

V

VV aV Averaged at section

2) Time-varying

Standard deviation= )1(

)(1i

2

-

-aring=

n

xxn

i

Variation[] = (Standard deviation X ) x 100

X Averaged at point

Table 23 Permission criteria of ANSIHI

Vortex occurence Swirl angle

Free surface vortex Submerged vortex

Observation

time Above 10 minutes Above 10 minutes

Above 10

minutes

Permission

criteria

Dye-core

Vortex(F3)

below 110 total

time

Dye-core

Vortex(S2)

below 110 total

time

Below

5degConditional

intermittent

below 7deg

225 Remedial measures for problem intakes

Usually in the preliminary test the worst flow condition is simulated And based on

the results remedial measures for problem intakes are discussed in this part

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

21

1) Overall approach flow patterns

The characteristics of flow approaching an intake structure is one of the foremost

considerations for the designer It is difficult to predict the effects of a given set of

flow conditions upstream from an intake structure Fig 210 depicts several typical

approach flow patterns of pump sump with different operating conditions

Fig 210 Examples of approach flow patterns

Generally speaking for the intake problems especially for the multiple pumps

sump there are the main measures using dividing walls placed between the pumps

as a partitioned structure trash racks with elongated bars guide piers or a number

of structural columns at the bay entrance can provide some assistance in controlling

cross-flow baffles flow distributors and screen mesh etc are applied to improve

the flow distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

22

2) Problems around pump intake

a) Free surface vortices

There are a number of factors that have an effect on the formation of surface

vortices To achieve a higher degree of certainty that objectionable surface vortices

do not form modifications can be made to intake structure to allow operation at

practical depths of submergence and the minimum submergence is referred in Table

22 the use of suction umbrellas vertical curtain walls and horizontal gratings

b) Submerged vortices

In the vicinity of intake bell there are the most complicated flow patterns and flow

makes the most changes in direction while maintaining a constant acceleration into

the pump bell to prevent local flow separation turbulence and vortex formation

Fig 211 represents a sample of various devices to inhibit the different submerged

vortex

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

23

Fig 211 Methods to reduce submerged vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

24

c) Preswirl

The approach flow distributions determine whether flow preswirl exists and a

sufficiently laterally skewed approach flow cause rotating around the intake bell

The most effective way of reducing preswirl is to establish a relatively uniform

approach flow within each pump bay by using the baffling schemes controlling

cross-flow and expanding concentrated flow discussed in the above part

To prevent non-uniform velocity distribution swirls and vortices necessary

modifications are conducted such as fore bay length baffle wall distributors

curtain wall mesh screen guide vane floor splitter wall fillet etc As the above

measures applied modification test will be conducted to check all the flow

conditions and compare with the preliminary results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

25

Chapter 3 Methodology

CFD (Computational fluid dynamics) is the technique which can design cycles to

better performance and reduce costs and time In CFD we solve the governing

equations of given physics (may be differential form or integral form) using some

numerical techniques like Finite Difference Method (FDM) Finite Element Method

(FEM) or Finite Volume Method (FVM) Although CFD plays a great part in the

design of turbomachinery CFD is not a replacement for experimental or analytical

approach In this study the numerical analysis was performed by the commercial

code ANSYS CFX 130 All aspects of the CFD study including modelling

meshing solving and post-processing of the results

31 Numerical modeling

ANSYS CFX solves the unsteady Navier-Stokes equations in their conservation

form The instantaneous equation of mass (continuity) in the stationary frame is

expressed as equation

ρ

δ + nabla ∙ (ρU) = 0 (31)

And the instantaneous equation for momentum is expressed as shown in equation

ρ

δ + nabla ∙ (ρU⨂U) = minusnablap + nabla ∙ τ + S (32)

Where τ is the stress tensor ( including both normal and shear components of the

stress)

These instantaneous equations are averaged for turbulent flows leading to additional

terms that need to be solved While the Navier-Stokes equations describe both

laminar and turbulent flows without addition terms realistic flows involve length

scales much smaller than the smallest finite volume mesh A Direct Numerical

Simulation of these flows would require significantly more computing power than

what is available now or in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

26

Therefore much research has been done to predict the effects of turbulence by

using turbulence models These models account for the effects of turbulence

without the use of a very fine mesh or direct numerical simulation

These turbulence models modify the transport equations by adding averaged and

fluctuating components The transport equations are changed to equations (33) and

(34)

ρ

δ + nabla ∙ (ρU) = 0 (33)

ρ

+ nabla ∙ ρ U ⨂U = minusnablap + nabla ∙ τ minus ρu⨂u + S (34)

The mass equation is not changed but the momentum equation contains extra terms

which are the Reynolds stresses ρ u⨂u and the Reynolds flux ρ u ϕ These

Reynolds stresses need to be modeled by additional equations to obtain closure

Obtaining closure implies that there are a sufficient number of equations to solve for

all the unknowns including the Reynolds stresses and Reynolds fluxes

Various turbulence models provide various ways to obtain closure In this

investigation Shear Stress Transport (SST) model and k-ε model were utilized The

advantage of SST model is that combines the advantages of other turbulence models

(the k-ε Wilcox k-ω) The 3 turbulence models will be discussed briefly

32 Turbulence models

Turbulence models are necessary because we cannot afford big enough computers

to directly capture every scale of motion There are three modelling frameworks

Direct Numerical Simulation (DNS) Reynolds-Averaged Navier-Stokes

Equations(RANS) Large Eddy Simulation(LES)

As RANS method is applied in most productionseveral RANS turbulence models

differ in their usage of wall functions the number of additional variables solved for

and what these variables represent

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

27

Two equation turbulence models

Two-equation turbulence models are widely used in CFD as they give a good

compromise between computational power needed and accuracy The term lsquotwo

equationrsquo refers to the fact that these models solve for the velocity and length scales

using separate transport equations The turbulence velocity scale is obtained by

solving the transport equation The turbulent length scale is estimated from two

properties of the turbulence field namely the turbulent kinetic energy and the

dissipation rate The dissipation rate of the kinetic energy is obtained from its

transport equation The most widely used are k-ε and k-ω two equation models In

the next section the k-ε Wilcox k-ω BSL k-ω and SST models will be briefly

discussed

321 k-ε turbulence model

The k-epsilon model solves for two variables k the turbulent kinetic energy and

epsilon the rate of dissipation of kinetic energy Wall functions are used in this

model so the flow in the buffer region is not simulated The k-epsilon model is very

popular for industrial applications due to its good convergence rate and relatively

low memory requirements It does not very accurately compute flow fields that

exhibit adverse pressure gradients strong curvature to the flow or jet flow It does

perform well for external flow problems around complex geometries

The k-ε model introduces k (m2s2) as the turbulence kinetic energy and ε (m2s3) as

the turbulence eddy dissipation The continuity equation remains the same

ρ

+ nabla ∙ (ρ U) = 0 (35)

The momentum equation changes as shown by equation

δρ

δ + nabla ∙ (ρU⨂U) = minusnablapprime + nabla ∙ μ

nablaU + (nablaU) + S (36)

Where SM is the sum of body forces μeff is the effective viscosity accounting for

turbulence and prsquo is the modified pressure The k-ε model uses the concept of eddy

viscosity giving the equation for effective viscosity as shown by equation

μ

= μ + μ (37)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

28

μt is the turbulence viscosity is linked to the turbulence kinetic energy and

dissipation by the equation

μ = Cμρ

ε (38)

Where Cμ is a constant

The values for k and ε come from the differential transport equations for the

turbulence kinetic energy and the turbulence dissipation rate

The turbulence kinetic energy equation is given as equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus ρε (39)

The turbulence dissipation rate is given by equation

(ρε)

+ nabla ∙ (ρUε) = nabla ∙ μ +

μ

σε nablaε +

ε

(Cε (P + Pε ) minus Cε ρε) (310)

Where Cε1 Cε2 σk σε are constants

Pk is the turbulence production due to viscous forces and is modeled by the

equation

P = μ nablaU ∙ nablaU + nablaU minus

nabla ∙ U 3μ

nabla ∙ U + ρk (311)

If a buoyancy term is added to the previous equation if the full buoyancy model is

used

The buoyancy term Pkb is modeled as

P = minusμ

ρσρg ∙ nablaρ (312)

322 Wilcox k-ω turbulence model

The k-omega model is similar to k-epsilon instead however it solves for omega mdash

the specific rate of dissipation of kinetic energy It also uses wall functions and

therefore has comparable memory requirements It has more difficulty converging

and is quite sensitive to the initial guess at the solution Hence the k-epsilon model

is often used first to find an initial condition for solving the k-omega model The k-

omega model is useful in many cases where the k-epsilon model is not accurate

such as internal flows flows that exhibit strong curvature separated flows and jets

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

29

This model has an advantage over the k-ε model where it does not involve complex

linear damping functions for near wall calculations at low Reynolds numbers

The k-ω model assumes that the turbulence viscosity is related to the turbulence

kinetic energy k and the turbulent frequency ω by the equation

μ = ρ

ω (313)

The transport equation for k is given by the equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus βprimeρ k ω (314)

The transport equation for ω is shown as equation

(ρω)

δ + nabla ∙ (ρ U ω) = nabla ∙ μ +

μ

μω

nablaω + αω

P + Pω minus β ρ ω (315)

The production rate of turbulence (Pk) is calculated as shown previously in the k-ε

section

The recommended values for the constants in the above equations are

βprime = 009

α = 59

β = 0075

σ = 2

σω = 2

The Reynolds stress tensor ρ u⨂u is calculated by

minusρ u⨂u = μ nablaU + (nablaU) minus

part ρk + μ

nabla ∙ U (316)

323 Shear stress transport model

The disadvantage of the Wilcox model is the strong sensitivity to free-stream

conditions Therefore a blending of the k-ω model near the surface and the k-ε in

the outer region was made by Menter [23] which resulted in the formulation of the

BSL k-ω turbulence model It consists of a transformation of the k-ε model to a k-ω

formulation and subsequently adding the resulting equations The Wilcox model is

multiplied by a blending function F1 and the transformed k-ε by another function 1-

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

30

F1 F1 is a function of wall distance (being the value of one near the surface and zero

outside the boundary layer) Outside the boundary and on the edge of the boundary

layer the standard k-ε model is used [24]

SST model is a combination of the k-epsilon in the free stream and the k-omega

models near the walls It does not use wall functions and tends to be most accurate

when solving the flow near the wall The SST model does not always converge to

the solution quickly so the k-epsilon or k-omega models are often solved first to

give good initial conditions

The k-ω based Shear Stress Transport (SST) model of Menter was applied for

turbulence treatment The transport equations for the SST model are expressed

below where the turbulent kinetic energy lsquokrsquo and turbulent frequency or dissipation

per unit turbulent kinetic energy are computed by using the following relations

For Turbulence Kinetic Energy

kkPukt

kTkk Ntilde+Ntilde+-=Ntilde+

para

para)[()( nsnwb (317)

where Pk is the production limiter

For Specific Dissipation Rate

ww

swnsnbwaww

ww NtildeNtilde-+Ntilde+Ntilde+-=Ntilde+para

para

1)1(2])[()( 21

22 kFSut

T (318)

The first blending function F1 is calculated from

iumlthorn

iumlyacute

uuml

iumlicirc

iumliacute

igrave

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

uacuteuacuteucirc

ugrave

ecircecirceuml

eacute

dividedivideoslash

oumlccedilccedilegrave

aelig=

4

22

21

4500maxmintanh

yCD

k

yy

kF

kw

wrs

w

n

wb (319)

y is the distance to the nearest wall and v is the kinematic viscosity

divideoslash

oumlccedilegrave

aeligNtildeNtilde= -10

2 101

2max ww

rsww kCDk (320)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

31

And Kinematic eddy viscosity

)max( 21

1

SF

kT

wa

an = (321)

where F2 is a blending function which restricts the limiter to the wall boundary and

S is the invariant measure of the strain rate

uacuteuacute

ucirc

ugrave

ecircecirc

euml

eacute

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

dividedivideoslash

oumlccedilccedilegrave

aelig=

2

22

5002maxtanh

w

n

wb yy

kF (322)

33 Brief introduction of simulation process

All aspects of CFD study including modelling meshing solving and post-

processing of the results are briefly introduced in this chapter

331 Pre-processing technique

The speed and accuracy of simulation results depending fundamentally on meshing

and the ability to prepare model geometry correctly

The steps in pre-processing are explained as follows

1) Domain Definitions

The domain definition depends mainly on the physics involved in the problem and

objective of the problem

a) Geometry creation or geometry import

b) Geometry cleanup

i) Removing the parts not necessary for simulation under consideration

ii) Closing the gaps in the geometry

iii) Removing small surfaces curves to make meshing process simple

2)Grid generation

To solve the governing equations the numerical methods like Finite Difference

Finite Volume or Finite Element Method to divide the domain into small parts and

solve the equations on one part at a time are applied

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 15: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

3

development liquidvapor interface tracking method to predict the cavitation

characteristics with a centrifugal pump impeller model

Moreover there are a number of correlated studies [13]~[17]

13 Methodology of study

This study will investigate the flow phenomena of two sump models

A scaled sump model is adopted to carry out numerical simulation and PIV

experiment The flow conditions are predicted and the effectiveness of devices for

free surface vortex is evaluated by using numerical code a single intake sump

model with a mixed flow pump installed is modeled and the hydraulic performance

characteristics for the head rise shaft power and pump efficiencies versus flow rate

are studied In addition flow phenomena (vortices cavitation etc) are analyzed

and the effectiveness of an anti-vortex device (AVD) for the submerged vortex is

evaluated

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

4

Chapter 2 Basic theory of sump model

Specific hydraulic phenomena have been confirmed that they can lead to certain

common operational problems The common of the main problems are summarized

as follows

a Free surface vortices

b Submerged vortices

c Swirl of flow entering the pump

d Non-uniform distribution of velocity at the pump impeller

e Entrained air or gas bubbles

These problems encountered in the pump sump will affect the pump

performance and significantly increase the operational and maintenance costs In

order to identify sources of particular problems and find practical solution for

it the usual approach is to conduct the laboratory experiments on a scaled

physical model Typical design objectives are to ensure that a pump station is

designed according to best practice and conforms to the requirements set out in the

American National Standard for Pump Intake Design[18] These standards restrict

the degree to which the above-mentioned undesirable flow patterns may be present

The negative impact of each of these hydraulic conditions varies with pump specific

speed and size Application of the standard to design a sump does not generate a

problem free sump but provides a basis for the initial design

As there are no specific guidelines or criteria for design of trouble free intakes the

most common solution to potential problems in new design and rectification of

problems observed in existing designs is to construct a scaled model in a laboratory

observe and investigate the flow characteristics and propose modifications to the

intake geometry There are additional devices in the form of floor splitters or cones

back wall splitters fillets surface beams guide vanes etc aimed at controlling the

vortex and swirl formation Considerable prior experience and ingenuity are

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

5

required to solve the problems by this method and the cost and time required are

also significant

The intake structure refers to the channel leading from the water source which

may be a river or reservoir and all installations downstream including the pump

column or the intake pipe (the suction tube portion of a vertical intake) the

approach channel (upstream of pump bay) and the pump bay (bounded by the floor

the back wall side walls dividers walls separating adjacent pump column) Usually

the upstream end of the pump column has an inlet attachment called the suction bell

The function of the intake is to supply an evenly distributed flow of water to the

pump suction bell

Several types of pump intake basins are available and ANSIHI 98 provides

design requirements for each Intake structures can be categorized as being

clear liquids or solids-bearing liquids For clear liquids intakes are further

classified into rectangular formed circular and trench types as well as suction

tanks and cans For solid-bearing liquids trench type and rectangular wet wells are

usually considered

In this paper only rectangular intake structure for clear liquid is studied

21 Pump intake design

For pumps to achieve their optimum hydraulic performance across all operating

conditions the flow at the impeller must meet specific hydraulic conditions The

ideal flow entering the pump inlet should be of a uniform velocity distribution

without rotation and stable over time When designing a sump to achieve favorable

inflow to the pump or suction pipe bell various sump dimensions relative to the size

of the bell are required Using these standards of the distances to reduce the

probability of the occurence of strong submerged vortices and free-surface vortices

211 Inlet bell diameter design

The geometry is generally defined in terms of the pump inlet diameter as shown in

Table 21 and Fig 21 The bell diameter can be estimated based on an inlet pipe

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

6

Table 21 Acceptable velocity ranges for inlet bell diameter(ANSIHI 98)

Flow Rate

Qls

Recommended Inlet Bell Design

Velocityms

Acceptable Velocity

Range ms

Q<315 V=17 06leVle27

315 leQ<1260 V=17 09leVle24

Q ge1260 V=17 12leVle21

Fig 21 Recommended inlet bell design diameter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

7

212 Recommended dimensions for a rectangular sump

As the selected bell diameter has been determined the proportions of the inlet

structure can be estimated from Fig 22 and Fig 23 Table 22 shows the

recommended dimensions for a rectangular sump

Fig 22 Recommended intake structure layout

Fig 23 Filler wall details for proper bay width

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

8

Table 22 Recommended dimensions for a rectangular sump

Dimension Variable

Description Recommended value

A Distance from the pump inlet bell

centerline to the entrance A=5D minimum

a Length of constricted bay section near

the pump inlet 25D minimum

B Distance from the back wall to the

pump inlet bell centerline 075D

C Distance between the inlet bell and

floor 03D~05D

D Inlet bell outside diameter refer to section 211

H Minimum liquid depth H=S+C

h Minimum height of constricted bay

section near the pump inlet bell hgtH or 25D

S Minimum pump inlet bell submergence S=D(10+23Fr)

W Pump inlet bay entrance width 2D minimum

w Constricted bay width near the pump

inlet bell w=2D

X Pump inlet bay length 5D minimum

Y Distance from pump inlet bell

centerline to the through-flow traveling screen

4D minimum

Z1 Distance from sump inlet bell centerline to diverging walls

5D minimum

Z2 Distance from inlet bell centerline to

sloping floor 5D minimum

α Angle of floor slope -10deg~ +10deg

β Angle of wall convergence 0deg~ +10deg

empty Angle of convergence from constricted

area to by walls 10deg maximum

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

9

FD=Vb(gD)05 (21)

where

FD Froude number (dimensionless)

Vb velocity at suction inlet=FlowArea based on D

D Outside diameter of bell or pipe inlet

g gravitational acceleration

There is the design sequence summary for rectangular intake structures

middot Consider the flow patterns and boundary geometry of the body of liquid

from which the pump station is to receive flow Compare with the approach

flow condition and determine if a hydraulic physical model study is

required

middot Determine the number and size of pumps required to satisfy the range of

operating conditions likely to be encountered

middot Identify pump inlet bell diameter Refer to the section 211

middot Determine the bell-floor clearance 05D is a good preliminary design value

middot Determine the required bell submergence using the equation 21

middot Determine the minimum allowable liquid depth in the intake structure from

the sum of the floor clearance and the required bell submergence

middot Check bottom elevation near the entrance to the structure and determine if it

is necessary to slope the floor upstream of the bay entrance If the resulting

depth at the entrance to the intake structure is shallow then check to ensure

that gravity-driven flow is not restricted by the entrance condition

middot Check the pump bay velocity for the maximum single-pump flow and

minimum liquid depth with the bay width set to 2D If bay velocity exceeds

05 ms then increase the bay width to reduce the maximum velocity below

05 ms

middot If it is necessary to increase the pump bay width to greater than 2D then

decrease bay width in the vicinity of the pumps according to Fig 23

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

10

middot Compare cross-flow velocity (at maximum system flow) to average pump

bay velocity If cross-flow value exceeds 50 of the bay velocity a

physical hydraulic model study is necessary

middot Determine the length of the structure and dividing walls giving

consideration to minimum allowable distances to a sloping floor screening

equipment and length of dividing walls if dual flow traveling screens or

drum screens are to be used a physical hydraulic model study is required

middot If the final selected pump bell diameter and inlet velocity is within the

range given in section 211 then the sump dimensions (developed based on

the inlet bell design diameter) need not be changed and will comply with

these standards

22 Model test

The approach flow pattern in a pump intake facility is difficult to predict by

traditional mathematics or empirical formulae For large structures or those that

differ significantly from proven designs a model study is the only means to ensure

success A model study is used to identify adverse hydraulic conditions and

derive remedial measures for approach flow patterns generated by structures

upstream of the pump impeller Current model studies are not intended to

investigate flow patterns induced by the pump itself or the flow patterns within the

pump The objective of a model study is to ensure that the pump intake structure

generates favorable flow conditions at the inlet to the pump Evaluation for the

requirement of model test if

a Sump or piping geometry that deviates from this design standard

b Non-uniform or non-symmetric approach flow to the pump sump

exist

c The pumps have flows greater than 2520 ls per pump or the total

station flow with all pumps running would be greater than 6310 ls

d Situations that need ten times cost than the model study

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

11

Adverse hydraulic conditions that can affect pump performance include free and

submerged vortices swirl approaching the pump impeller flow separation at the

pump bell and a non-uniform axial velocity distribution at the suction

The purpose of model test is to examine these adverse hydraulic conditions and to

investigate measure against them

221 The scale effects

One possible difficulty in scale modeling is that it is impossible to reduce all

pertinent forces by the same factor With Froude-scale models the inertial and

gravitational forces are reduced similarly but the viscous and surface tension forces

cannot be simultaneously reduced The extra influence of these forces is known as

a ldquoscale effectrdquo A number of investigations have indicated that the scale effects for

the model are avoided if the Reynolds number based on the inlet flow and

submergence or intake diameter exceeds approximately 3x 104 and the Weber

number exceeds 150 [19]

In most engineering applications involving closed-conduit and open-channel flow

the Reynolds number limits are far exceeded and the flows are fully turbulent

In a sump model Froude number (Fr) and Reynolds number (Re) are the most

important non-dimensional parameters Some reduction in Re should be

introduced to the sump model However flow patterns are generally very similar

at high Re

No specific geometric scale rate is recommended but reasonable large

geometric scale to minimize viscous and surface tension scale effects and reproduce

the flow pattern

Re= uDν (22)

We= u2ρDσ (23)

where

u average axial velocity (ms)

ν kinematic viscosity of the liquid (m2s)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

12

ρ liquid density (kgm3)

σ surface tension of liquidair interface

According to the ANSIHI 98 the viscous scale effects on vortex may be negligible

if the resulting dimensionless numbers must meet these values Re>6times104 and

We>240

To allow practical visual observations of flow patterns accurate measurements of

swirl and velocity distribution and sufficient dimensional control the model scale

shall yield a bay width of at least 300mm a minimum liquid depth of at least

150mm and a pump throat or suction diameter of at least 80mm

222 Principle of similarity

In the similitude analysis the geometric and flow similarity requires the model and

the flow to be identical as the real model and flow This is to ensure the result

obtained from the model study is well presented in order to predict full scale

behavior Model involving a free surface are operated using Froude similarity since

the flow process is controlled by gravity and inertial forces

For similarity of the flow patterns the Froude number shall be equal in model and

prototype

Fr =

=

(24)

and as the scale ratio λ was selected in section 221 the following equation is

obtained

V = V times (

) = V times λ or Q = Q times (

) = Q times λ (25)

where Fr is the Froude number λ is the scale ratio of model dimensions to prototype

The test of 15 times Froude scaled flow should be conducted to ascertain the

potential scale affect on free surface vortices

Models of closed conduit piping system leading to a pump suction are not operated

based on Froude similitude However flow patterns are generally very similar at

high Re A minimum value of 1times105 for the Re at the pump suction is

recommended to ensure the flow patterns are correctly scaled

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

13

223 Preliminary model test

The initial design shall be tested first to identify any hydraulic problems

1) Vortex Formation Mechanisms

There are various investigations that provide insight into the fundamental processes

leading to the development of vortices both in experiment and numerical simulation

Shin et al [20] demonstrated that two basic mechanisms lead to inlet vortex

formation The first mechanism involves the development of an inlet vortex due to

the amplification of ambient vorticity in the approach flow as vortex lines

are convicted into the inlet The second mechanism involves the development of a

trailing vortex in the vicinity of the intake as a result of the variation in

circulation along the inlet For this second case a vortex can develop in a flow that

is irrotational upstream and the vortex development therefore does not depend

on the presence of ambient vorticity Shin et al investigation on kinematic

parameters indicate that the strength of an inlet-vortex or trailing vortex system

increases with decreasing distance from the surface However for an inlet in

an upstream irrotational flow two counter rotating vortices can still trail from

the rear of the inlet

2) Classification of vortex type

As illustrated in Fig 24 vortices in the vicinity of pump intakes may be adjacent to

the channel bottom or a channel wall (ie submerged vortices) or they may appear

adjacent to the free surface (ie free surface vortex)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

14

(a) Schematic diagram of vortices in a sump

(b) Experimental results of different vortex

Fig 24 Typical vortex in a sump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

15

For the model visual observation dye was injected around the intake structure near

the surface and submergence as Fig 24 (b) shows And the model test results of

vortex was classified as Fig 25

(a) Free Surface Vortices

(b) Submerged Vortices

Fig 25 Vortex classifications

There are a number of researches focused on the numerical theory to detect the

inception of a visible vortex As the diameters of vortices are usually much smaller

compared with the computing grid size to compensate for the lack of vortex

resolution in numerical simulation a stretching vortex model is applied to the local

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

16

flow field around semi-analytically identified vortex positions [21][22] The brief

explanations are as follows

a) Free surface vortex

The criteria expressed by equation 26 and 27 is applied to determine the

visible inception of a free surface vortex

≫ 1 (26)

=

gt 1 (27)

where P is the pressure drop at the vortex core and Ph is the static head at the

vortex element and f is the drag force acting at a unit length of water surface

vortex dimple based on Stokes approximation and caused by the down flow

Further f is the force acting on the bubble in the direction from the high

pressure region to the low pressure region like buoyancy

b) Submerged vortex

Equation 28 is applied to determine the visible inception of a submerged vortex

infin =

infin

infin gt 1 (28)

where is the static pressure at the vortex core infin is the atmospheric pressure

and is the critical cavitation inception pressure and nearly equal to the

saturated vapor pressure

3) Swirl angle

Swirl angle predicts the intensity of flow rotation Fig 26 shows experiment

calculation method with swirl meter and attempts of the CFD calculation method

were conducted as Fig 27 A four blade zero-pitch swirl meter which is supported

by a low friction bearing will be installed at about four suction pipe diameter (d)

downstream from the pump suction to measure swirl angle of pump approaching

flow For the swirl meter the tip to tip blade diameter is 075d and the length in

flow direction is 06 and one of the four blade is painted yellow as a reference to

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

17

count revolutions As flow swirl is usually unsteady the observation time of swirl

meter reading should be a continuous period of time

Fig 26 Typical swirl meter

(29)

(210)

(211)

Where qV flow rotational speed at swirl meter (ms)

zV average axial velocity at swirl meter (ms)

n Number of revolution per minute (rpm)

d Diameter of pipe at the swirl meter (m)

q flow rate (m3s)

zV

Vqq 1tan -=

)(60

smnd

Vtimestimes

=p

q

)(4

2

smd

qVz

times=

p

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

18

For the swirl angle calculating method with CFD the key point is to obtain the

average tangential velocity so the swirl check circles were created as illustrated in

Fig 27 The swirl check circle is located at the section of 4d height with diameter

of 025d 05d and 075d respectively

Swirl check circle

Fig 27 Schematic view of swirl angle numerical calculation

23

1 7501 501 250

acute++

=aringaringaring

N

i

N

i

NNVNVNV

Vi

q (212)

where

qV average tangential velocity (ms)

V circumferential velocity of check point in check circle

respectively (ms)

N check point total numbers in single check circle

4) Velocity

Velocity measurement at bell throat by Pitot tube was implemented by ANSIHI

method Typical Pitot tube configuration is shown in Fig 28 Eight measuring

points are defined as Fig 29 Time averaged velocity at eight points was obtained

and these eight data were again averaged to give throat sectional spatial velocity

average

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

19

Fig 28 Example of Pitot tube Fig 29 Measuring points at bell throat

224 Test criteria

There is the criteria for satisfactory performance of the sump model test and

summarized in Table 23

1) Free surface and submerged vortices

Free surface vortices entering the pump must be less severe than vortices with

coherent core Dye core vortices may be acceptable only if they occur less than 10

of the time or only for infrequent pump operating conditions

Submerged vortex which starts from the sump walls induce vibration and noise

problems Dye core submerged vortices extending into the pump bell mouth is not

allowed

2) Swirl angles

Swirl angle is used to check the flow rotation in the suction pipe The calculation

method is referred to section 223

Short-term maximum and long-term average values must be less than 5 degrees

Maximum short-term swirl angles up to 7 degrees may be acceptable only if they

occur less than 10 of the time or only for infrequent pump operating conditions

3)Velocity

Each point time-averaged velocity shall not be larger than 10 of sectional average

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

20

Standard deviation of time-varying velocity at each point was obtained and its

value shall not be larger than 10 of the time-averaged velocity at each point

1) Time-averaged

Variation[] = 100acute-

a

a

V

VV aV Averaged at section

2) Time-varying

Standard deviation= )1(

)(1i

2

-

-aring=

n

xxn

i

Variation[] = (Standard deviation X ) x 100

X Averaged at point

Table 23 Permission criteria of ANSIHI

Vortex occurence Swirl angle

Free surface vortex Submerged vortex

Observation

time Above 10 minutes Above 10 minutes

Above 10

minutes

Permission

criteria

Dye-core

Vortex(F3)

below 110 total

time

Dye-core

Vortex(S2)

below 110 total

time

Below

5degConditional

intermittent

below 7deg

225 Remedial measures for problem intakes

Usually in the preliminary test the worst flow condition is simulated And based on

the results remedial measures for problem intakes are discussed in this part

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

21

1) Overall approach flow patterns

The characteristics of flow approaching an intake structure is one of the foremost

considerations for the designer It is difficult to predict the effects of a given set of

flow conditions upstream from an intake structure Fig 210 depicts several typical

approach flow patterns of pump sump with different operating conditions

Fig 210 Examples of approach flow patterns

Generally speaking for the intake problems especially for the multiple pumps

sump there are the main measures using dividing walls placed between the pumps

as a partitioned structure trash racks with elongated bars guide piers or a number

of structural columns at the bay entrance can provide some assistance in controlling

cross-flow baffles flow distributors and screen mesh etc are applied to improve

the flow distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

22

2) Problems around pump intake

a) Free surface vortices

There are a number of factors that have an effect on the formation of surface

vortices To achieve a higher degree of certainty that objectionable surface vortices

do not form modifications can be made to intake structure to allow operation at

practical depths of submergence and the minimum submergence is referred in Table

22 the use of suction umbrellas vertical curtain walls and horizontal gratings

b) Submerged vortices

In the vicinity of intake bell there are the most complicated flow patterns and flow

makes the most changes in direction while maintaining a constant acceleration into

the pump bell to prevent local flow separation turbulence and vortex formation

Fig 211 represents a sample of various devices to inhibit the different submerged

vortex

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

23

Fig 211 Methods to reduce submerged vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

24

c) Preswirl

The approach flow distributions determine whether flow preswirl exists and a

sufficiently laterally skewed approach flow cause rotating around the intake bell

The most effective way of reducing preswirl is to establish a relatively uniform

approach flow within each pump bay by using the baffling schemes controlling

cross-flow and expanding concentrated flow discussed in the above part

To prevent non-uniform velocity distribution swirls and vortices necessary

modifications are conducted such as fore bay length baffle wall distributors

curtain wall mesh screen guide vane floor splitter wall fillet etc As the above

measures applied modification test will be conducted to check all the flow

conditions and compare with the preliminary results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

25

Chapter 3 Methodology

CFD (Computational fluid dynamics) is the technique which can design cycles to

better performance and reduce costs and time In CFD we solve the governing

equations of given physics (may be differential form or integral form) using some

numerical techniques like Finite Difference Method (FDM) Finite Element Method

(FEM) or Finite Volume Method (FVM) Although CFD plays a great part in the

design of turbomachinery CFD is not a replacement for experimental or analytical

approach In this study the numerical analysis was performed by the commercial

code ANSYS CFX 130 All aspects of the CFD study including modelling

meshing solving and post-processing of the results

31 Numerical modeling

ANSYS CFX solves the unsteady Navier-Stokes equations in their conservation

form The instantaneous equation of mass (continuity) in the stationary frame is

expressed as equation

ρ

δ + nabla ∙ (ρU) = 0 (31)

And the instantaneous equation for momentum is expressed as shown in equation

ρ

δ + nabla ∙ (ρU⨂U) = minusnablap + nabla ∙ τ + S (32)

Where τ is the stress tensor ( including both normal and shear components of the

stress)

These instantaneous equations are averaged for turbulent flows leading to additional

terms that need to be solved While the Navier-Stokes equations describe both

laminar and turbulent flows without addition terms realistic flows involve length

scales much smaller than the smallest finite volume mesh A Direct Numerical

Simulation of these flows would require significantly more computing power than

what is available now or in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

26

Therefore much research has been done to predict the effects of turbulence by

using turbulence models These models account for the effects of turbulence

without the use of a very fine mesh or direct numerical simulation

These turbulence models modify the transport equations by adding averaged and

fluctuating components The transport equations are changed to equations (33) and

(34)

ρ

δ + nabla ∙ (ρU) = 0 (33)

ρ

+ nabla ∙ ρ U ⨂U = minusnablap + nabla ∙ τ minus ρu⨂u + S (34)

The mass equation is not changed but the momentum equation contains extra terms

which are the Reynolds stresses ρ u⨂u and the Reynolds flux ρ u ϕ These

Reynolds stresses need to be modeled by additional equations to obtain closure

Obtaining closure implies that there are a sufficient number of equations to solve for

all the unknowns including the Reynolds stresses and Reynolds fluxes

Various turbulence models provide various ways to obtain closure In this

investigation Shear Stress Transport (SST) model and k-ε model were utilized The

advantage of SST model is that combines the advantages of other turbulence models

(the k-ε Wilcox k-ω) The 3 turbulence models will be discussed briefly

32 Turbulence models

Turbulence models are necessary because we cannot afford big enough computers

to directly capture every scale of motion There are three modelling frameworks

Direct Numerical Simulation (DNS) Reynolds-Averaged Navier-Stokes

Equations(RANS) Large Eddy Simulation(LES)

As RANS method is applied in most productionseveral RANS turbulence models

differ in their usage of wall functions the number of additional variables solved for

and what these variables represent

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

27

Two equation turbulence models

Two-equation turbulence models are widely used in CFD as they give a good

compromise between computational power needed and accuracy The term lsquotwo

equationrsquo refers to the fact that these models solve for the velocity and length scales

using separate transport equations The turbulence velocity scale is obtained by

solving the transport equation The turbulent length scale is estimated from two

properties of the turbulence field namely the turbulent kinetic energy and the

dissipation rate The dissipation rate of the kinetic energy is obtained from its

transport equation The most widely used are k-ε and k-ω two equation models In

the next section the k-ε Wilcox k-ω BSL k-ω and SST models will be briefly

discussed

321 k-ε turbulence model

The k-epsilon model solves for two variables k the turbulent kinetic energy and

epsilon the rate of dissipation of kinetic energy Wall functions are used in this

model so the flow in the buffer region is not simulated The k-epsilon model is very

popular for industrial applications due to its good convergence rate and relatively

low memory requirements It does not very accurately compute flow fields that

exhibit adverse pressure gradients strong curvature to the flow or jet flow It does

perform well for external flow problems around complex geometries

The k-ε model introduces k (m2s2) as the turbulence kinetic energy and ε (m2s3) as

the turbulence eddy dissipation The continuity equation remains the same

ρ

+ nabla ∙ (ρ U) = 0 (35)

The momentum equation changes as shown by equation

δρ

δ + nabla ∙ (ρU⨂U) = minusnablapprime + nabla ∙ μ

nablaU + (nablaU) + S (36)

Where SM is the sum of body forces μeff is the effective viscosity accounting for

turbulence and prsquo is the modified pressure The k-ε model uses the concept of eddy

viscosity giving the equation for effective viscosity as shown by equation

μ

= μ + μ (37)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

28

μt is the turbulence viscosity is linked to the turbulence kinetic energy and

dissipation by the equation

μ = Cμρ

ε (38)

Where Cμ is a constant

The values for k and ε come from the differential transport equations for the

turbulence kinetic energy and the turbulence dissipation rate

The turbulence kinetic energy equation is given as equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus ρε (39)

The turbulence dissipation rate is given by equation

(ρε)

+ nabla ∙ (ρUε) = nabla ∙ μ +

μ

σε nablaε +

ε

(Cε (P + Pε ) minus Cε ρε) (310)

Where Cε1 Cε2 σk σε are constants

Pk is the turbulence production due to viscous forces and is modeled by the

equation

P = μ nablaU ∙ nablaU + nablaU minus

nabla ∙ U 3μ

nabla ∙ U + ρk (311)

If a buoyancy term is added to the previous equation if the full buoyancy model is

used

The buoyancy term Pkb is modeled as

P = minusμ

ρσρg ∙ nablaρ (312)

322 Wilcox k-ω turbulence model

The k-omega model is similar to k-epsilon instead however it solves for omega mdash

the specific rate of dissipation of kinetic energy It also uses wall functions and

therefore has comparable memory requirements It has more difficulty converging

and is quite sensitive to the initial guess at the solution Hence the k-epsilon model

is often used first to find an initial condition for solving the k-omega model The k-

omega model is useful in many cases where the k-epsilon model is not accurate

such as internal flows flows that exhibit strong curvature separated flows and jets

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

29

This model has an advantage over the k-ε model where it does not involve complex

linear damping functions for near wall calculations at low Reynolds numbers

The k-ω model assumes that the turbulence viscosity is related to the turbulence

kinetic energy k and the turbulent frequency ω by the equation

μ = ρ

ω (313)

The transport equation for k is given by the equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus βprimeρ k ω (314)

The transport equation for ω is shown as equation

(ρω)

δ + nabla ∙ (ρ U ω) = nabla ∙ μ +

μ

μω

nablaω + αω

P + Pω minus β ρ ω (315)

The production rate of turbulence (Pk) is calculated as shown previously in the k-ε

section

The recommended values for the constants in the above equations are

βprime = 009

α = 59

β = 0075

σ = 2

σω = 2

The Reynolds stress tensor ρ u⨂u is calculated by

minusρ u⨂u = μ nablaU + (nablaU) minus

part ρk + μ

nabla ∙ U (316)

323 Shear stress transport model

The disadvantage of the Wilcox model is the strong sensitivity to free-stream

conditions Therefore a blending of the k-ω model near the surface and the k-ε in

the outer region was made by Menter [23] which resulted in the formulation of the

BSL k-ω turbulence model It consists of a transformation of the k-ε model to a k-ω

formulation and subsequently adding the resulting equations The Wilcox model is

multiplied by a blending function F1 and the transformed k-ε by another function 1-

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

30

F1 F1 is a function of wall distance (being the value of one near the surface and zero

outside the boundary layer) Outside the boundary and on the edge of the boundary

layer the standard k-ε model is used [24]

SST model is a combination of the k-epsilon in the free stream and the k-omega

models near the walls It does not use wall functions and tends to be most accurate

when solving the flow near the wall The SST model does not always converge to

the solution quickly so the k-epsilon or k-omega models are often solved first to

give good initial conditions

The k-ω based Shear Stress Transport (SST) model of Menter was applied for

turbulence treatment The transport equations for the SST model are expressed

below where the turbulent kinetic energy lsquokrsquo and turbulent frequency or dissipation

per unit turbulent kinetic energy are computed by using the following relations

For Turbulence Kinetic Energy

kkPukt

kTkk Ntilde+Ntilde+-=Ntilde+

para

para)[()( nsnwb (317)

where Pk is the production limiter

For Specific Dissipation Rate

ww

swnsnbwaww

ww NtildeNtilde-+Ntilde+Ntilde+-=Ntilde+para

para

1)1(2])[()( 21

22 kFSut

T (318)

The first blending function F1 is calculated from

iumlthorn

iumlyacute

uuml

iumlicirc

iumliacute

igrave

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

uacuteuacuteucirc

ugrave

ecircecirceuml

eacute

dividedivideoslash

oumlccedilccedilegrave

aelig=

4

22

21

4500maxmintanh

yCD

k

yy

kF

kw

wrs

w

n

wb (319)

y is the distance to the nearest wall and v is the kinematic viscosity

divideoslash

oumlccedilegrave

aeligNtildeNtilde= -10

2 101

2max ww

rsww kCDk (320)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

31

And Kinematic eddy viscosity

)max( 21

1

SF

kT

wa

an = (321)

where F2 is a blending function which restricts the limiter to the wall boundary and

S is the invariant measure of the strain rate

uacuteuacute

ucirc

ugrave

ecircecirc

euml

eacute

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

dividedivideoslash

oumlccedilccedilegrave

aelig=

2

22

5002maxtanh

w

n

wb yy

kF (322)

33 Brief introduction of simulation process

All aspects of CFD study including modelling meshing solving and post-

processing of the results are briefly introduced in this chapter

331 Pre-processing technique

The speed and accuracy of simulation results depending fundamentally on meshing

and the ability to prepare model geometry correctly

The steps in pre-processing are explained as follows

1) Domain Definitions

The domain definition depends mainly on the physics involved in the problem and

objective of the problem

a) Geometry creation or geometry import

b) Geometry cleanup

i) Removing the parts not necessary for simulation under consideration

ii) Closing the gaps in the geometry

iii) Removing small surfaces curves to make meshing process simple

2)Grid generation

To solve the governing equations the numerical methods like Finite Difference

Finite Volume or Finite Element Method to divide the domain into small parts and

solve the equations on one part at a time are applied

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 16: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

4

Chapter 2 Basic theory of sump model

Specific hydraulic phenomena have been confirmed that they can lead to certain

common operational problems The common of the main problems are summarized

as follows

a Free surface vortices

b Submerged vortices

c Swirl of flow entering the pump

d Non-uniform distribution of velocity at the pump impeller

e Entrained air or gas bubbles

These problems encountered in the pump sump will affect the pump

performance and significantly increase the operational and maintenance costs In

order to identify sources of particular problems and find practical solution for

it the usual approach is to conduct the laboratory experiments on a scaled

physical model Typical design objectives are to ensure that a pump station is

designed according to best practice and conforms to the requirements set out in the

American National Standard for Pump Intake Design[18] These standards restrict

the degree to which the above-mentioned undesirable flow patterns may be present

The negative impact of each of these hydraulic conditions varies with pump specific

speed and size Application of the standard to design a sump does not generate a

problem free sump but provides a basis for the initial design

As there are no specific guidelines or criteria for design of trouble free intakes the

most common solution to potential problems in new design and rectification of

problems observed in existing designs is to construct a scaled model in a laboratory

observe and investigate the flow characteristics and propose modifications to the

intake geometry There are additional devices in the form of floor splitters or cones

back wall splitters fillets surface beams guide vanes etc aimed at controlling the

vortex and swirl formation Considerable prior experience and ingenuity are

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

5

required to solve the problems by this method and the cost and time required are

also significant

The intake structure refers to the channel leading from the water source which

may be a river or reservoir and all installations downstream including the pump

column or the intake pipe (the suction tube portion of a vertical intake) the

approach channel (upstream of pump bay) and the pump bay (bounded by the floor

the back wall side walls dividers walls separating adjacent pump column) Usually

the upstream end of the pump column has an inlet attachment called the suction bell

The function of the intake is to supply an evenly distributed flow of water to the

pump suction bell

Several types of pump intake basins are available and ANSIHI 98 provides

design requirements for each Intake structures can be categorized as being

clear liquids or solids-bearing liquids For clear liquids intakes are further

classified into rectangular formed circular and trench types as well as suction

tanks and cans For solid-bearing liquids trench type and rectangular wet wells are

usually considered

In this paper only rectangular intake structure for clear liquid is studied

21 Pump intake design

For pumps to achieve their optimum hydraulic performance across all operating

conditions the flow at the impeller must meet specific hydraulic conditions The

ideal flow entering the pump inlet should be of a uniform velocity distribution

without rotation and stable over time When designing a sump to achieve favorable

inflow to the pump or suction pipe bell various sump dimensions relative to the size

of the bell are required Using these standards of the distances to reduce the

probability of the occurence of strong submerged vortices and free-surface vortices

211 Inlet bell diameter design

The geometry is generally defined in terms of the pump inlet diameter as shown in

Table 21 and Fig 21 The bell diameter can be estimated based on an inlet pipe

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

6

Table 21 Acceptable velocity ranges for inlet bell diameter(ANSIHI 98)

Flow Rate

Qls

Recommended Inlet Bell Design

Velocityms

Acceptable Velocity

Range ms

Q<315 V=17 06leVle27

315 leQ<1260 V=17 09leVle24

Q ge1260 V=17 12leVle21

Fig 21 Recommended inlet bell design diameter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

7

212 Recommended dimensions for a rectangular sump

As the selected bell diameter has been determined the proportions of the inlet

structure can be estimated from Fig 22 and Fig 23 Table 22 shows the

recommended dimensions for a rectangular sump

Fig 22 Recommended intake structure layout

Fig 23 Filler wall details for proper bay width

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

8

Table 22 Recommended dimensions for a rectangular sump

Dimension Variable

Description Recommended value

A Distance from the pump inlet bell

centerline to the entrance A=5D minimum

a Length of constricted bay section near

the pump inlet 25D minimum

B Distance from the back wall to the

pump inlet bell centerline 075D

C Distance between the inlet bell and

floor 03D~05D

D Inlet bell outside diameter refer to section 211

H Minimum liquid depth H=S+C

h Minimum height of constricted bay

section near the pump inlet bell hgtH or 25D

S Minimum pump inlet bell submergence S=D(10+23Fr)

W Pump inlet bay entrance width 2D minimum

w Constricted bay width near the pump

inlet bell w=2D

X Pump inlet bay length 5D minimum

Y Distance from pump inlet bell

centerline to the through-flow traveling screen

4D minimum

Z1 Distance from sump inlet bell centerline to diverging walls

5D minimum

Z2 Distance from inlet bell centerline to

sloping floor 5D minimum

α Angle of floor slope -10deg~ +10deg

β Angle of wall convergence 0deg~ +10deg

empty Angle of convergence from constricted

area to by walls 10deg maximum

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

9

FD=Vb(gD)05 (21)

where

FD Froude number (dimensionless)

Vb velocity at suction inlet=FlowArea based on D

D Outside diameter of bell or pipe inlet

g gravitational acceleration

There is the design sequence summary for rectangular intake structures

middot Consider the flow patterns and boundary geometry of the body of liquid

from which the pump station is to receive flow Compare with the approach

flow condition and determine if a hydraulic physical model study is

required

middot Determine the number and size of pumps required to satisfy the range of

operating conditions likely to be encountered

middot Identify pump inlet bell diameter Refer to the section 211

middot Determine the bell-floor clearance 05D is a good preliminary design value

middot Determine the required bell submergence using the equation 21

middot Determine the minimum allowable liquid depth in the intake structure from

the sum of the floor clearance and the required bell submergence

middot Check bottom elevation near the entrance to the structure and determine if it

is necessary to slope the floor upstream of the bay entrance If the resulting

depth at the entrance to the intake structure is shallow then check to ensure

that gravity-driven flow is not restricted by the entrance condition

middot Check the pump bay velocity for the maximum single-pump flow and

minimum liquid depth with the bay width set to 2D If bay velocity exceeds

05 ms then increase the bay width to reduce the maximum velocity below

05 ms

middot If it is necessary to increase the pump bay width to greater than 2D then

decrease bay width in the vicinity of the pumps according to Fig 23

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

10

middot Compare cross-flow velocity (at maximum system flow) to average pump

bay velocity If cross-flow value exceeds 50 of the bay velocity a

physical hydraulic model study is necessary

middot Determine the length of the structure and dividing walls giving

consideration to minimum allowable distances to a sloping floor screening

equipment and length of dividing walls if dual flow traveling screens or

drum screens are to be used a physical hydraulic model study is required

middot If the final selected pump bell diameter and inlet velocity is within the

range given in section 211 then the sump dimensions (developed based on

the inlet bell design diameter) need not be changed and will comply with

these standards

22 Model test

The approach flow pattern in a pump intake facility is difficult to predict by

traditional mathematics or empirical formulae For large structures or those that

differ significantly from proven designs a model study is the only means to ensure

success A model study is used to identify adverse hydraulic conditions and

derive remedial measures for approach flow patterns generated by structures

upstream of the pump impeller Current model studies are not intended to

investigate flow patterns induced by the pump itself or the flow patterns within the

pump The objective of a model study is to ensure that the pump intake structure

generates favorable flow conditions at the inlet to the pump Evaluation for the

requirement of model test if

a Sump or piping geometry that deviates from this design standard

b Non-uniform or non-symmetric approach flow to the pump sump

exist

c The pumps have flows greater than 2520 ls per pump or the total

station flow with all pumps running would be greater than 6310 ls

d Situations that need ten times cost than the model study

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

11

Adverse hydraulic conditions that can affect pump performance include free and

submerged vortices swirl approaching the pump impeller flow separation at the

pump bell and a non-uniform axial velocity distribution at the suction

The purpose of model test is to examine these adverse hydraulic conditions and to

investigate measure against them

221 The scale effects

One possible difficulty in scale modeling is that it is impossible to reduce all

pertinent forces by the same factor With Froude-scale models the inertial and

gravitational forces are reduced similarly but the viscous and surface tension forces

cannot be simultaneously reduced The extra influence of these forces is known as

a ldquoscale effectrdquo A number of investigations have indicated that the scale effects for

the model are avoided if the Reynolds number based on the inlet flow and

submergence or intake diameter exceeds approximately 3x 104 and the Weber

number exceeds 150 [19]

In most engineering applications involving closed-conduit and open-channel flow

the Reynolds number limits are far exceeded and the flows are fully turbulent

In a sump model Froude number (Fr) and Reynolds number (Re) are the most

important non-dimensional parameters Some reduction in Re should be

introduced to the sump model However flow patterns are generally very similar

at high Re

No specific geometric scale rate is recommended but reasonable large

geometric scale to minimize viscous and surface tension scale effects and reproduce

the flow pattern

Re= uDν (22)

We= u2ρDσ (23)

where

u average axial velocity (ms)

ν kinematic viscosity of the liquid (m2s)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

12

ρ liquid density (kgm3)

σ surface tension of liquidair interface

According to the ANSIHI 98 the viscous scale effects on vortex may be negligible

if the resulting dimensionless numbers must meet these values Re>6times104 and

We>240

To allow practical visual observations of flow patterns accurate measurements of

swirl and velocity distribution and sufficient dimensional control the model scale

shall yield a bay width of at least 300mm a minimum liquid depth of at least

150mm and a pump throat or suction diameter of at least 80mm

222 Principle of similarity

In the similitude analysis the geometric and flow similarity requires the model and

the flow to be identical as the real model and flow This is to ensure the result

obtained from the model study is well presented in order to predict full scale

behavior Model involving a free surface are operated using Froude similarity since

the flow process is controlled by gravity and inertial forces

For similarity of the flow patterns the Froude number shall be equal in model and

prototype

Fr =

=

(24)

and as the scale ratio λ was selected in section 221 the following equation is

obtained

V = V times (

) = V times λ or Q = Q times (

) = Q times λ (25)

where Fr is the Froude number λ is the scale ratio of model dimensions to prototype

The test of 15 times Froude scaled flow should be conducted to ascertain the

potential scale affect on free surface vortices

Models of closed conduit piping system leading to a pump suction are not operated

based on Froude similitude However flow patterns are generally very similar at

high Re A minimum value of 1times105 for the Re at the pump suction is

recommended to ensure the flow patterns are correctly scaled

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

13

223 Preliminary model test

The initial design shall be tested first to identify any hydraulic problems

1) Vortex Formation Mechanisms

There are various investigations that provide insight into the fundamental processes

leading to the development of vortices both in experiment and numerical simulation

Shin et al [20] demonstrated that two basic mechanisms lead to inlet vortex

formation The first mechanism involves the development of an inlet vortex due to

the amplification of ambient vorticity in the approach flow as vortex lines

are convicted into the inlet The second mechanism involves the development of a

trailing vortex in the vicinity of the intake as a result of the variation in

circulation along the inlet For this second case a vortex can develop in a flow that

is irrotational upstream and the vortex development therefore does not depend

on the presence of ambient vorticity Shin et al investigation on kinematic

parameters indicate that the strength of an inlet-vortex or trailing vortex system

increases with decreasing distance from the surface However for an inlet in

an upstream irrotational flow two counter rotating vortices can still trail from

the rear of the inlet

2) Classification of vortex type

As illustrated in Fig 24 vortices in the vicinity of pump intakes may be adjacent to

the channel bottom or a channel wall (ie submerged vortices) or they may appear

adjacent to the free surface (ie free surface vortex)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

14

(a) Schematic diagram of vortices in a sump

(b) Experimental results of different vortex

Fig 24 Typical vortex in a sump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

15

For the model visual observation dye was injected around the intake structure near

the surface and submergence as Fig 24 (b) shows And the model test results of

vortex was classified as Fig 25

(a) Free Surface Vortices

(b) Submerged Vortices

Fig 25 Vortex classifications

There are a number of researches focused on the numerical theory to detect the

inception of a visible vortex As the diameters of vortices are usually much smaller

compared with the computing grid size to compensate for the lack of vortex

resolution in numerical simulation a stretching vortex model is applied to the local

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

16

flow field around semi-analytically identified vortex positions [21][22] The brief

explanations are as follows

a) Free surface vortex

The criteria expressed by equation 26 and 27 is applied to determine the

visible inception of a free surface vortex

≫ 1 (26)

=

gt 1 (27)

where P is the pressure drop at the vortex core and Ph is the static head at the

vortex element and f is the drag force acting at a unit length of water surface

vortex dimple based on Stokes approximation and caused by the down flow

Further f is the force acting on the bubble in the direction from the high

pressure region to the low pressure region like buoyancy

b) Submerged vortex

Equation 28 is applied to determine the visible inception of a submerged vortex

infin =

infin

infin gt 1 (28)

where is the static pressure at the vortex core infin is the atmospheric pressure

and is the critical cavitation inception pressure and nearly equal to the

saturated vapor pressure

3) Swirl angle

Swirl angle predicts the intensity of flow rotation Fig 26 shows experiment

calculation method with swirl meter and attempts of the CFD calculation method

were conducted as Fig 27 A four blade zero-pitch swirl meter which is supported

by a low friction bearing will be installed at about four suction pipe diameter (d)

downstream from the pump suction to measure swirl angle of pump approaching

flow For the swirl meter the tip to tip blade diameter is 075d and the length in

flow direction is 06 and one of the four blade is painted yellow as a reference to

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

17

count revolutions As flow swirl is usually unsteady the observation time of swirl

meter reading should be a continuous period of time

Fig 26 Typical swirl meter

(29)

(210)

(211)

Where qV flow rotational speed at swirl meter (ms)

zV average axial velocity at swirl meter (ms)

n Number of revolution per minute (rpm)

d Diameter of pipe at the swirl meter (m)

q flow rate (m3s)

zV

Vqq 1tan -=

)(60

smnd

Vtimestimes

=p

q

)(4

2

smd

qVz

times=

p

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

18

For the swirl angle calculating method with CFD the key point is to obtain the

average tangential velocity so the swirl check circles were created as illustrated in

Fig 27 The swirl check circle is located at the section of 4d height with diameter

of 025d 05d and 075d respectively

Swirl check circle

Fig 27 Schematic view of swirl angle numerical calculation

23

1 7501 501 250

acute++

=aringaringaring

N

i

N

i

NNVNVNV

Vi

q (212)

where

qV average tangential velocity (ms)

V circumferential velocity of check point in check circle

respectively (ms)

N check point total numbers in single check circle

4) Velocity

Velocity measurement at bell throat by Pitot tube was implemented by ANSIHI

method Typical Pitot tube configuration is shown in Fig 28 Eight measuring

points are defined as Fig 29 Time averaged velocity at eight points was obtained

and these eight data were again averaged to give throat sectional spatial velocity

average

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

19

Fig 28 Example of Pitot tube Fig 29 Measuring points at bell throat

224 Test criteria

There is the criteria for satisfactory performance of the sump model test and

summarized in Table 23

1) Free surface and submerged vortices

Free surface vortices entering the pump must be less severe than vortices with

coherent core Dye core vortices may be acceptable only if they occur less than 10

of the time or only for infrequent pump operating conditions

Submerged vortex which starts from the sump walls induce vibration and noise

problems Dye core submerged vortices extending into the pump bell mouth is not

allowed

2) Swirl angles

Swirl angle is used to check the flow rotation in the suction pipe The calculation

method is referred to section 223

Short-term maximum and long-term average values must be less than 5 degrees

Maximum short-term swirl angles up to 7 degrees may be acceptable only if they

occur less than 10 of the time or only for infrequent pump operating conditions

3)Velocity

Each point time-averaged velocity shall not be larger than 10 of sectional average

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

20

Standard deviation of time-varying velocity at each point was obtained and its

value shall not be larger than 10 of the time-averaged velocity at each point

1) Time-averaged

Variation[] = 100acute-

a

a

V

VV aV Averaged at section

2) Time-varying

Standard deviation= )1(

)(1i

2

-

-aring=

n

xxn

i

Variation[] = (Standard deviation X ) x 100

X Averaged at point

Table 23 Permission criteria of ANSIHI

Vortex occurence Swirl angle

Free surface vortex Submerged vortex

Observation

time Above 10 minutes Above 10 minutes

Above 10

minutes

Permission

criteria

Dye-core

Vortex(F3)

below 110 total

time

Dye-core

Vortex(S2)

below 110 total

time

Below

5degConditional

intermittent

below 7deg

225 Remedial measures for problem intakes

Usually in the preliminary test the worst flow condition is simulated And based on

the results remedial measures for problem intakes are discussed in this part

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

21

1) Overall approach flow patterns

The characteristics of flow approaching an intake structure is one of the foremost

considerations for the designer It is difficult to predict the effects of a given set of

flow conditions upstream from an intake structure Fig 210 depicts several typical

approach flow patterns of pump sump with different operating conditions

Fig 210 Examples of approach flow patterns

Generally speaking for the intake problems especially for the multiple pumps

sump there are the main measures using dividing walls placed between the pumps

as a partitioned structure trash racks with elongated bars guide piers or a number

of structural columns at the bay entrance can provide some assistance in controlling

cross-flow baffles flow distributors and screen mesh etc are applied to improve

the flow distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

22

2) Problems around pump intake

a) Free surface vortices

There are a number of factors that have an effect on the formation of surface

vortices To achieve a higher degree of certainty that objectionable surface vortices

do not form modifications can be made to intake structure to allow operation at

practical depths of submergence and the minimum submergence is referred in Table

22 the use of suction umbrellas vertical curtain walls and horizontal gratings

b) Submerged vortices

In the vicinity of intake bell there are the most complicated flow patterns and flow

makes the most changes in direction while maintaining a constant acceleration into

the pump bell to prevent local flow separation turbulence and vortex formation

Fig 211 represents a sample of various devices to inhibit the different submerged

vortex

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

23

Fig 211 Methods to reduce submerged vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

24

c) Preswirl

The approach flow distributions determine whether flow preswirl exists and a

sufficiently laterally skewed approach flow cause rotating around the intake bell

The most effective way of reducing preswirl is to establish a relatively uniform

approach flow within each pump bay by using the baffling schemes controlling

cross-flow and expanding concentrated flow discussed in the above part

To prevent non-uniform velocity distribution swirls and vortices necessary

modifications are conducted such as fore bay length baffle wall distributors

curtain wall mesh screen guide vane floor splitter wall fillet etc As the above

measures applied modification test will be conducted to check all the flow

conditions and compare with the preliminary results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

25

Chapter 3 Methodology

CFD (Computational fluid dynamics) is the technique which can design cycles to

better performance and reduce costs and time In CFD we solve the governing

equations of given physics (may be differential form or integral form) using some

numerical techniques like Finite Difference Method (FDM) Finite Element Method

(FEM) or Finite Volume Method (FVM) Although CFD plays a great part in the

design of turbomachinery CFD is not a replacement for experimental or analytical

approach In this study the numerical analysis was performed by the commercial

code ANSYS CFX 130 All aspects of the CFD study including modelling

meshing solving and post-processing of the results

31 Numerical modeling

ANSYS CFX solves the unsteady Navier-Stokes equations in their conservation

form The instantaneous equation of mass (continuity) in the stationary frame is

expressed as equation

ρ

δ + nabla ∙ (ρU) = 0 (31)

And the instantaneous equation for momentum is expressed as shown in equation

ρ

δ + nabla ∙ (ρU⨂U) = minusnablap + nabla ∙ τ + S (32)

Where τ is the stress tensor ( including both normal and shear components of the

stress)

These instantaneous equations are averaged for turbulent flows leading to additional

terms that need to be solved While the Navier-Stokes equations describe both

laminar and turbulent flows without addition terms realistic flows involve length

scales much smaller than the smallest finite volume mesh A Direct Numerical

Simulation of these flows would require significantly more computing power than

what is available now or in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

26

Therefore much research has been done to predict the effects of turbulence by

using turbulence models These models account for the effects of turbulence

without the use of a very fine mesh or direct numerical simulation

These turbulence models modify the transport equations by adding averaged and

fluctuating components The transport equations are changed to equations (33) and

(34)

ρ

δ + nabla ∙ (ρU) = 0 (33)

ρ

+ nabla ∙ ρ U ⨂U = minusnablap + nabla ∙ τ minus ρu⨂u + S (34)

The mass equation is not changed but the momentum equation contains extra terms

which are the Reynolds stresses ρ u⨂u and the Reynolds flux ρ u ϕ These

Reynolds stresses need to be modeled by additional equations to obtain closure

Obtaining closure implies that there are a sufficient number of equations to solve for

all the unknowns including the Reynolds stresses and Reynolds fluxes

Various turbulence models provide various ways to obtain closure In this

investigation Shear Stress Transport (SST) model and k-ε model were utilized The

advantage of SST model is that combines the advantages of other turbulence models

(the k-ε Wilcox k-ω) The 3 turbulence models will be discussed briefly

32 Turbulence models

Turbulence models are necessary because we cannot afford big enough computers

to directly capture every scale of motion There are three modelling frameworks

Direct Numerical Simulation (DNS) Reynolds-Averaged Navier-Stokes

Equations(RANS) Large Eddy Simulation(LES)

As RANS method is applied in most productionseveral RANS turbulence models

differ in their usage of wall functions the number of additional variables solved for

and what these variables represent

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

27

Two equation turbulence models

Two-equation turbulence models are widely used in CFD as they give a good

compromise between computational power needed and accuracy The term lsquotwo

equationrsquo refers to the fact that these models solve for the velocity and length scales

using separate transport equations The turbulence velocity scale is obtained by

solving the transport equation The turbulent length scale is estimated from two

properties of the turbulence field namely the turbulent kinetic energy and the

dissipation rate The dissipation rate of the kinetic energy is obtained from its

transport equation The most widely used are k-ε and k-ω two equation models In

the next section the k-ε Wilcox k-ω BSL k-ω and SST models will be briefly

discussed

321 k-ε turbulence model

The k-epsilon model solves for two variables k the turbulent kinetic energy and

epsilon the rate of dissipation of kinetic energy Wall functions are used in this

model so the flow in the buffer region is not simulated The k-epsilon model is very

popular for industrial applications due to its good convergence rate and relatively

low memory requirements It does not very accurately compute flow fields that

exhibit adverse pressure gradients strong curvature to the flow or jet flow It does

perform well for external flow problems around complex geometries

The k-ε model introduces k (m2s2) as the turbulence kinetic energy and ε (m2s3) as

the turbulence eddy dissipation The continuity equation remains the same

ρ

+ nabla ∙ (ρ U) = 0 (35)

The momentum equation changes as shown by equation

δρ

δ + nabla ∙ (ρU⨂U) = minusnablapprime + nabla ∙ μ

nablaU + (nablaU) + S (36)

Where SM is the sum of body forces μeff is the effective viscosity accounting for

turbulence and prsquo is the modified pressure The k-ε model uses the concept of eddy

viscosity giving the equation for effective viscosity as shown by equation

μ

= μ + μ (37)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

28

μt is the turbulence viscosity is linked to the turbulence kinetic energy and

dissipation by the equation

μ = Cμρ

ε (38)

Where Cμ is a constant

The values for k and ε come from the differential transport equations for the

turbulence kinetic energy and the turbulence dissipation rate

The turbulence kinetic energy equation is given as equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus ρε (39)

The turbulence dissipation rate is given by equation

(ρε)

+ nabla ∙ (ρUε) = nabla ∙ μ +

μ

σε nablaε +

ε

(Cε (P + Pε ) minus Cε ρε) (310)

Where Cε1 Cε2 σk σε are constants

Pk is the turbulence production due to viscous forces and is modeled by the

equation

P = μ nablaU ∙ nablaU + nablaU minus

nabla ∙ U 3μ

nabla ∙ U + ρk (311)

If a buoyancy term is added to the previous equation if the full buoyancy model is

used

The buoyancy term Pkb is modeled as

P = minusμ

ρσρg ∙ nablaρ (312)

322 Wilcox k-ω turbulence model

The k-omega model is similar to k-epsilon instead however it solves for omega mdash

the specific rate of dissipation of kinetic energy It also uses wall functions and

therefore has comparable memory requirements It has more difficulty converging

and is quite sensitive to the initial guess at the solution Hence the k-epsilon model

is often used first to find an initial condition for solving the k-omega model The k-

omega model is useful in many cases where the k-epsilon model is not accurate

such as internal flows flows that exhibit strong curvature separated flows and jets

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

29

This model has an advantage over the k-ε model where it does not involve complex

linear damping functions for near wall calculations at low Reynolds numbers

The k-ω model assumes that the turbulence viscosity is related to the turbulence

kinetic energy k and the turbulent frequency ω by the equation

μ = ρ

ω (313)

The transport equation for k is given by the equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus βprimeρ k ω (314)

The transport equation for ω is shown as equation

(ρω)

δ + nabla ∙ (ρ U ω) = nabla ∙ μ +

μ

μω

nablaω + αω

P + Pω minus β ρ ω (315)

The production rate of turbulence (Pk) is calculated as shown previously in the k-ε

section

The recommended values for the constants in the above equations are

βprime = 009

α = 59

β = 0075

σ = 2

σω = 2

The Reynolds stress tensor ρ u⨂u is calculated by

minusρ u⨂u = μ nablaU + (nablaU) minus

part ρk + μ

nabla ∙ U (316)

323 Shear stress transport model

The disadvantage of the Wilcox model is the strong sensitivity to free-stream

conditions Therefore a blending of the k-ω model near the surface and the k-ε in

the outer region was made by Menter [23] which resulted in the formulation of the

BSL k-ω turbulence model It consists of a transformation of the k-ε model to a k-ω

formulation and subsequently adding the resulting equations The Wilcox model is

multiplied by a blending function F1 and the transformed k-ε by another function 1-

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

30

F1 F1 is a function of wall distance (being the value of one near the surface and zero

outside the boundary layer) Outside the boundary and on the edge of the boundary

layer the standard k-ε model is used [24]

SST model is a combination of the k-epsilon in the free stream and the k-omega

models near the walls It does not use wall functions and tends to be most accurate

when solving the flow near the wall The SST model does not always converge to

the solution quickly so the k-epsilon or k-omega models are often solved first to

give good initial conditions

The k-ω based Shear Stress Transport (SST) model of Menter was applied for

turbulence treatment The transport equations for the SST model are expressed

below where the turbulent kinetic energy lsquokrsquo and turbulent frequency or dissipation

per unit turbulent kinetic energy are computed by using the following relations

For Turbulence Kinetic Energy

kkPukt

kTkk Ntilde+Ntilde+-=Ntilde+

para

para)[()( nsnwb (317)

where Pk is the production limiter

For Specific Dissipation Rate

ww

swnsnbwaww

ww NtildeNtilde-+Ntilde+Ntilde+-=Ntilde+para

para

1)1(2])[()( 21

22 kFSut

T (318)

The first blending function F1 is calculated from

iumlthorn

iumlyacute

uuml

iumlicirc

iumliacute

igrave

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

uacuteuacuteucirc

ugrave

ecircecirceuml

eacute

dividedivideoslash

oumlccedilccedilegrave

aelig=

4

22

21

4500maxmintanh

yCD

k

yy

kF

kw

wrs

w

n

wb (319)

y is the distance to the nearest wall and v is the kinematic viscosity

divideoslash

oumlccedilegrave

aeligNtildeNtilde= -10

2 101

2max ww

rsww kCDk (320)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

31

And Kinematic eddy viscosity

)max( 21

1

SF

kT

wa

an = (321)

where F2 is a blending function which restricts the limiter to the wall boundary and

S is the invariant measure of the strain rate

uacuteuacute

ucirc

ugrave

ecircecirc

euml

eacute

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

dividedivideoslash

oumlccedilccedilegrave

aelig=

2

22

5002maxtanh

w

n

wb yy

kF (322)

33 Brief introduction of simulation process

All aspects of CFD study including modelling meshing solving and post-

processing of the results are briefly introduced in this chapter

331 Pre-processing technique

The speed and accuracy of simulation results depending fundamentally on meshing

and the ability to prepare model geometry correctly

The steps in pre-processing are explained as follows

1) Domain Definitions

The domain definition depends mainly on the physics involved in the problem and

objective of the problem

a) Geometry creation or geometry import

b) Geometry cleanup

i) Removing the parts not necessary for simulation under consideration

ii) Closing the gaps in the geometry

iii) Removing small surfaces curves to make meshing process simple

2)Grid generation

To solve the governing equations the numerical methods like Finite Difference

Finite Volume or Finite Element Method to divide the domain into small parts and

solve the equations on one part at a time are applied

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 17: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

5

required to solve the problems by this method and the cost and time required are

also significant

The intake structure refers to the channel leading from the water source which

may be a river or reservoir and all installations downstream including the pump

column or the intake pipe (the suction tube portion of a vertical intake) the

approach channel (upstream of pump bay) and the pump bay (bounded by the floor

the back wall side walls dividers walls separating adjacent pump column) Usually

the upstream end of the pump column has an inlet attachment called the suction bell

The function of the intake is to supply an evenly distributed flow of water to the

pump suction bell

Several types of pump intake basins are available and ANSIHI 98 provides

design requirements for each Intake structures can be categorized as being

clear liquids or solids-bearing liquids For clear liquids intakes are further

classified into rectangular formed circular and trench types as well as suction

tanks and cans For solid-bearing liquids trench type and rectangular wet wells are

usually considered

In this paper only rectangular intake structure for clear liquid is studied

21 Pump intake design

For pumps to achieve their optimum hydraulic performance across all operating

conditions the flow at the impeller must meet specific hydraulic conditions The

ideal flow entering the pump inlet should be of a uniform velocity distribution

without rotation and stable over time When designing a sump to achieve favorable

inflow to the pump or suction pipe bell various sump dimensions relative to the size

of the bell are required Using these standards of the distances to reduce the

probability of the occurence of strong submerged vortices and free-surface vortices

211 Inlet bell diameter design

The geometry is generally defined in terms of the pump inlet diameter as shown in

Table 21 and Fig 21 The bell diameter can be estimated based on an inlet pipe

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

6

Table 21 Acceptable velocity ranges for inlet bell diameter(ANSIHI 98)

Flow Rate

Qls

Recommended Inlet Bell Design

Velocityms

Acceptable Velocity

Range ms

Q<315 V=17 06leVle27

315 leQ<1260 V=17 09leVle24

Q ge1260 V=17 12leVle21

Fig 21 Recommended inlet bell design diameter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

7

212 Recommended dimensions for a rectangular sump

As the selected bell diameter has been determined the proportions of the inlet

structure can be estimated from Fig 22 and Fig 23 Table 22 shows the

recommended dimensions for a rectangular sump

Fig 22 Recommended intake structure layout

Fig 23 Filler wall details for proper bay width

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

8

Table 22 Recommended dimensions for a rectangular sump

Dimension Variable

Description Recommended value

A Distance from the pump inlet bell

centerline to the entrance A=5D minimum

a Length of constricted bay section near

the pump inlet 25D minimum

B Distance from the back wall to the

pump inlet bell centerline 075D

C Distance between the inlet bell and

floor 03D~05D

D Inlet bell outside diameter refer to section 211

H Minimum liquid depth H=S+C

h Minimum height of constricted bay

section near the pump inlet bell hgtH or 25D

S Minimum pump inlet bell submergence S=D(10+23Fr)

W Pump inlet bay entrance width 2D minimum

w Constricted bay width near the pump

inlet bell w=2D

X Pump inlet bay length 5D minimum

Y Distance from pump inlet bell

centerline to the through-flow traveling screen

4D minimum

Z1 Distance from sump inlet bell centerline to diverging walls

5D minimum

Z2 Distance from inlet bell centerline to

sloping floor 5D minimum

α Angle of floor slope -10deg~ +10deg

β Angle of wall convergence 0deg~ +10deg

empty Angle of convergence from constricted

area to by walls 10deg maximum

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

9

FD=Vb(gD)05 (21)

where

FD Froude number (dimensionless)

Vb velocity at suction inlet=FlowArea based on D

D Outside diameter of bell or pipe inlet

g gravitational acceleration

There is the design sequence summary for rectangular intake structures

middot Consider the flow patterns and boundary geometry of the body of liquid

from which the pump station is to receive flow Compare with the approach

flow condition and determine if a hydraulic physical model study is

required

middot Determine the number and size of pumps required to satisfy the range of

operating conditions likely to be encountered

middot Identify pump inlet bell diameter Refer to the section 211

middot Determine the bell-floor clearance 05D is a good preliminary design value

middot Determine the required bell submergence using the equation 21

middot Determine the minimum allowable liquid depth in the intake structure from

the sum of the floor clearance and the required bell submergence

middot Check bottom elevation near the entrance to the structure and determine if it

is necessary to slope the floor upstream of the bay entrance If the resulting

depth at the entrance to the intake structure is shallow then check to ensure

that gravity-driven flow is not restricted by the entrance condition

middot Check the pump bay velocity for the maximum single-pump flow and

minimum liquid depth with the bay width set to 2D If bay velocity exceeds

05 ms then increase the bay width to reduce the maximum velocity below

05 ms

middot If it is necessary to increase the pump bay width to greater than 2D then

decrease bay width in the vicinity of the pumps according to Fig 23

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

10

middot Compare cross-flow velocity (at maximum system flow) to average pump

bay velocity If cross-flow value exceeds 50 of the bay velocity a

physical hydraulic model study is necessary

middot Determine the length of the structure and dividing walls giving

consideration to minimum allowable distances to a sloping floor screening

equipment and length of dividing walls if dual flow traveling screens or

drum screens are to be used a physical hydraulic model study is required

middot If the final selected pump bell diameter and inlet velocity is within the

range given in section 211 then the sump dimensions (developed based on

the inlet bell design diameter) need not be changed and will comply with

these standards

22 Model test

The approach flow pattern in a pump intake facility is difficult to predict by

traditional mathematics or empirical formulae For large structures or those that

differ significantly from proven designs a model study is the only means to ensure

success A model study is used to identify adverse hydraulic conditions and

derive remedial measures for approach flow patterns generated by structures

upstream of the pump impeller Current model studies are not intended to

investigate flow patterns induced by the pump itself or the flow patterns within the

pump The objective of a model study is to ensure that the pump intake structure

generates favorable flow conditions at the inlet to the pump Evaluation for the

requirement of model test if

a Sump or piping geometry that deviates from this design standard

b Non-uniform or non-symmetric approach flow to the pump sump

exist

c The pumps have flows greater than 2520 ls per pump or the total

station flow with all pumps running would be greater than 6310 ls

d Situations that need ten times cost than the model study

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

11

Adverse hydraulic conditions that can affect pump performance include free and

submerged vortices swirl approaching the pump impeller flow separation at the

pump bell and a non-uniform axial velocity distribution at the suction

The purpose of model test is to examine these adverse hydraulic conditions and to

investigate measure against them

221 The scale effects

One possible difficulty in scale modeling is that it is impossible to reduce all

pertinent forces by the same factor With Froude-scale models the inertial and

gravitational forces are reduced similarly but the viscous and surface tension forces

cannot be simultaneously reduced The extra influence of these forces is known as

a ldquoscale effectrdquo A number of investigations have indicated that the scale effects for

the model are avoided if the Reynolds number based on the inlet flow and

submergence or intake diameter exceeds approximately 3x 104 and the Weber

number exceeds 150 [19]

In most engineering applications involving closed-conduit and open-channel flow

the Reynolds number limits are far exceeded and the flows are fully turbulent

In a sump model Froude number (Fr) and Reynolds number (Re) are the most

important non-dimensional parameters Some reduction in Re should be

introduced to the sump model However flow patterns are generally very similar

at high Re

No specific geometric scale rate is recommended but reasonable large

geometric scale to minimize viscous and surface tension scale effects and reproduce

the flow pattern

Re= uDν (22)

We= u2ρDσ (23)

where

u average axial velocity (ms)

ν kinematic viscosity of the liquid (m2s)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

12

ρ liquid density (kgm3)

σ surface tension of liquidair interface

According to the ANSIHI 98 the viscous scale effects on vortex may be negligible

if the resulting dimensionless numbers must meet these values Re>6times104 and

We>240

To allow practical visual observations of flow patterns accurate measurements of

swirl and velocity distribution and sufficient dimensional control the model scale

shall yield a bay width of at least 300mm a minimum liquid depth of at least

150mm and a pump throat or suction diameter of at least 80mm

222 Principle of similarity

In the similitude analysis the geometric and flow similarity requires the model and

the flow to be identical as the real model and flow This is to ensure the result

obtained from the model study is well presented in order to predict full scale

behavior Model involving a free surface are operated using Froude similarity since

the flow process is controlled by gravity and inertial forces

For similarity of the flow patterns the Froude number shall be equal in model and

prototype

Fr =

=

(24)

and as the scale ratio λ was selected in section 221 the following equation is

obtained

V = V times (

) = V times λ or Q = Q times (

) = Q times λ (25)

where Fr is the Froude number λ is the scale ratio of model dimensions to prototype

The test of 15 times Froude scaled flow should be conducted to ascertain the

potential scale affect on free surface vortices

Models of closed conduit piping system leading to a pump suction are not operated

based on Froude similitude However flow patterns are generally very similar at

high Re A minimum value of 1times105 for the Re at the pump suction is

recommended to ensure the flow patterns are correctly scaled

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

13

223 Preliminary model test

The initial design shall be tested first to identify any hydraulic problems

1) Vortex Formation Mechanisms

There are various investigations that provide insight into the fundamental processes

leading to the development of vortices both in experiment and numerical simulation

Shin et al [20] demonstrated that two basic mechanisms lead to inlet vortex

formation The first mechanism involves the development of an inlet vortex due to

the amplification of ambient vorticity in the approach flow as vortex lines

are convicted into the inlet The second mechanism involves the development of a

trailing vortex in the vicinity of the intake as a result of the variation in

circulation along the inlet For this second case a vortex can develop in a flow that

is irrotational upstream and the vortex development therefore does not depend

on the presence of ambient vorticity Shin et al investigation on kinematic

parameters indicate that the strength of an inlet-vortex or trailing vortex system

increases with decreasing distance from the surface However for an inlet in

an upstream irrotational flow two counter rotating vortices can still trail from

the rear of the inlet

2) Classification of vortex type

As illustrated in Fig 24 vortices in the vicinity of pump intakes may be adjacent to

the channel bottom or a channel wall (ie submerged vortices) or they may appear

adjacent to the free surface (ie free surface vortex)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

14

(a) Schematic diagram of vortices in a sump

(b) Experimental results of different vortex

Fig 24 Typical vortex in a sump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

15

For the model visual observation dye was injected around the intake structure near

the surface and submergence as Fig 24 (b) shows And the model test results of

vortex was classified as Fig 25

(a) Free Surface Vortices

(b) Submerged Vortices

Fig 25 Vortex classifications

There are a number of researches focused on the numerical theory to detect the

inception of a visible vortex As the diameters of vortices are usually much smaller

compared with the computing grid size to compensate for the lack of vortex

resolution in numerical simulation a stretching vortex model is applied to the local

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

16

flow field around semi-analytically identified vortex positions [21][22] The brief

explanations are as follows

a) Free surface vortex

The criteria expressed by equation 26 and 27 is applied to determine the

visible inception of a free surface vortex

≫ 1 (26)

=

gt 1 (27)

where P is the pressure drop at the vortex core and Ph is the static head at the

vortex element and f is the drag force acting at a unit length of water surface

vortex dimple based on Stokes approximation and caused by the down flow

Further f is the force acting on the bubble in the direction from the high

pressure region to the low pressure region like buoyancy

b) Submerged vortex

Equation 28 is applied to determine the visible inception of a submerged vortex

infin =

infin

infin gt 1 (28)

where is the static pressure at the vortex core infin is the atmospheric pressure

and is the critical cavitation inception pressure and nearly equal to the

saturated vapor pressure

3) Swirl angle

Swirl angle predicts the intensity of flow rotation Fig 26 shows experiment

calculation method with swirl meter and attempts of the CFD calculation method

were conducted as Fig 27 A four blade zero-pitch swirl meter which is supported

by a low friction bearing will be installed at about four suction pipe diameter (d)

downstream from the pump suction to measure swirl angle of pump approaching

flow For the swirl meter the tip to tip blade diameter is 075d and the length in

flow direction is 06 and one of the four blade is painted yellow as a reference to

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

17

count revolutions As flow swirl is usually unsteady the observation time of swirl

meter reading should be a continuous period of time

Fig 26 Typical swirl meter

(29)

(210)

(211)

Where qV flow rotational speed at swirl meter (ms)

zV average axial velocity at swirl meter (ms)

n Number of revolution per minute (rpm)

d Diameter of pipe at the swirl meter (m)

q flow rate (m3s)

zV

Vqq 1tan -=

)(60

smnd

Vtimestimes

=p

q

)(4

2

smd

qVz

times=

p

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

18

For the swirl angle calculating method with CFD the key point is to obtain the

average tangential velocity so the swirl check circles were created as illustrated in

Fig 27 The swirl check circle is located at the section of 4d height with diameter

of 025d 05d and 075d respectively

Swirl check circle

Fig 27 Schematic view of swirl angle numerical calculation

23

1 7501 501 250

acute++

=aringaringaring

N

i

N

i

NNVNVNV

Vi

q (212)

where

qV average tangential velocity (ms)

V circumferential velocity of check point in check circle

respectively (ms)

N check point total numbers in single check circle

4) Velocity

Velocity measurement at bell throat by Pitot tube was implemented by ANSIHI

method Typical Pitot tube configuration is shown in Fig 28 Eight measuring

points are defined as Fig 29 Time averaged velocity at eight points was obtained

and these eight data were again averaged to give throat sectional spatial velocity

average

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

19

Fig 28 Example of Pitot tube Fig 29 Measuring points at bell throat

224 Test criteria

There is the criteria for satisfactory performance of the sump model test and

summarized in Table 23

1) Free surface and submerged vortices

Free surface vortices entering the pump must be less severe than vortices with

coherent core Dye core vortices may be acceptable only if they occur less than 10

of the time or only for infrequent pump operating conditions

Submerged vortex which starts from the sump walls induce vibration and noise

problems Dye core submerged vortices extending into the pump bell mouth is not

allowed

2) Swirl angles

Swirl angle is used to check the flow rotation in the suction pipe The calculation

method is referred to section 223

Short-term maximum and long-term average values must be less than 5 degrees

Maximum short-term swirl angles up to 7 degrees may be acceptable only if they

occur less than 10 of the time or only for infrequent pump operating conditions

3)Velocity

Each point time-averaged velocity shall not be larger than 10 of sectional average

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

20

Standard deviation of time-varying velocity at each point was obtained and its

value shall not be larger than 10 of the time-averaged velocity at each point

1) Time-averaged

Variation[] = 100acute-

a

a

V

VV aV Averaged at section

2) Time-varying

Standard deviation= )1(

)(1i

2

-

-aring=

n

xxn

i

Variation[] = (Standard deviation X ) x 100

X Averaged at point

Table 23 Permission criteria of ANSIHI

Vortex occurence Swirl angle

Free surface vortex Submerged vortex

Observation

time Above 10 minutes Above 10 minutes

Above 10

minutes

Permission

criteria

Dye-core

Vortex(F3)

below 110 total

time

Dye-core

Vortex(S2)

below 110 total

time

Below

5degConditional

intermittent

below 7deg

225 Remedial measures for problem intakes

Usually in the preliminary test the worst flow condition is simulated And based on

the results remedial measures for problem intakes are discussed in this part

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

21

1) Overall approach flow patterns

The characteristics of flow approaching an intake structure is one of the foremost

considerations for the designer It is difficult to predict the effects of a given set of

flow conditions upstream from an intake structure Fig 210 depicts several typical

approach flow patterns of pump sump with different operating conditions

Fig 210 Examples of approach flow patterns

Generally speaking for the intake problems especially for the multiple pumps

sump there are the main measures using dividing walls placed between the pumps

as a partitioned structure trash racks with elongated bars guide piers or a number

of structural columns at the bay entrance can provide some assistance in controlling

cross-flow baffles flow distributors and screen mesh etc are applied to improve

the flow distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

22

2) Problems around pump intake

a) Free surface vortices

There are a number of factors that have an effect on the formation of surface

vortices To achieve a higher degree of certainty that objectionable surface vortices

do not form modifications can be made to intake structure to allow operation at

practical depths of submergence and the minimum submergence is referred in Table

22 the use of suction umbrellas vertical curtain walls and horizontal gratings

b) Submerged vortices

In the vicinity of intake bell there are the most complicated flow patterns and flow

makes the most changes in direction while maintaining a constant acceleration into

the pump bell to prevent local flow separation turbulence and vortex formation

Fig 211 represents a sample of various devices to inhibit the different submerged

vortex

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

23

Fig 211 Methods to reduce submerged vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

24

c) Preswirl

The approach flow distributions determine whether flow preswirl exists and a

sufficiently laterally skewed approach flow cause rotating around the intake bell

The most effective way of reducing preswirl is to establish a relatively uniform

approach flow within each pump bay by using the baffling schemes controlling

cross-flow and expanding concentrated flow discussed in the above part

To prevent non-uniform velocity distribution swirls and vortices necessary

modifications are conducted such as fore bay length baffle wall distributors

curtain wall mesh screen guide vane floor splitter wall fillet etc As the above

measures applied modification test will be conducted to check all the flow

conditions and compare with the preliminary results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

25

Chapter 3 Methodology

CFD (Computational fluid dynamics) is the technique which can design cycles to

better performance and reduce costs and time In CFD we solve the governing

equations of given physics (may be differential form or integral form) using some

numerical techniques like Finite Difference Method (FDM) Finite Element Method

(FEM) or Finite Volume Method (FVM) Although CFD plays a great part in the

design of turbomachinery CFD is not a replacement for experimental or analytical

approach In this study the numerical analysis was performed by the commercial

code ANSYS CFX 130 All aspects of the CFD study including modelling

meshing solving and post-processing of the results

31 Numerical modeling

ANSYS CFX solves the unsteady Navier-Stokes equations in their conservation

form The instantaneous equation of mass (continuity) in the stationary frame is

expressed as equation

ρ

δ + nabla ∙ (ρU) = 0 (31)

And the instantaneous equation for momentum is expressed as shown in equation

ρ

δ + nabla ∙ (ρU⨂U) = minusnablap + nabla ∙ τ + S (32)

Where τ is the stress tensor ( including both normal and shear components of the

stress)

These instantaneous equations are averaged for turbulent flows leading to additional

terms that need to be solved While the Navier-Stokes equations describe both

laminar and turbulent flows without addition terms realistic flows involve length

scales much smaller than the smallest finite volume mesh A Direct Numerical

Simulation of these flows would require significantly more computing power than

what is available now or in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

26

Therefore much research has been done to predict the effects of turbulence by

using turbulence models These models account for the effects of turbulence

without the use of a very fine mesh or direct numerical simulation

These turbulence models modify the transport equations by adding averaged and

fluctuating components The transport equations are changed to equations (33) and

(34)

ρ

δ + nabla ∙ (ρU) = 0 (33)

ρ

+ nabla ∙ ρ U ⨂U = minusnablap + nabla ∙ τ minus ρu⨂u + S (34)

The mass equation is not changed but the momentum equation contains extra terms

which are the Reynolds stresses ρ u⨂u and the Reynolds flux ρ u ϕ These

Reynolds stresses need to be modeled by additional equations to obtain closure

Obtaining closure implies that there are a sufficient number of equations to solve for

all the unknowns including the Reynolds stresses and Reynolds fluxes

Various turbulence models provide various ways to obtain closure In this

investigation Shear Stress Transport (SST) model and k-ε model were utilized The

advantage of SST model is that combines the advantages of other turbulence models

(the k-ε Wilcox k-ω) The 3 turbulence models will be discussed briefly

32 Turbulence models

Turbulence models are necessary because we cannot afford big enough computers

to directly capture every scale of motion There are three modelling frameworks

Direct Numerical Simulation (DNS) Reynolds-Averaged Navier-Stokes

Equations(RANS) Large Eddy Simulation(LES)

As RANS method is applied in most productionseveral RANS turbulence models

differ in their usage of wall functions the number of additional variables solved for

and what these variables represent

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

27

Two equation turbulence models

Two-equation turbulence models are widely used in CFD as they give a good

compromise between computational power needed and accuracy The term lsquotwo

equationrsquo refers to the fact that these models solve for the velocity and length scales

using separate transport equations The turbulence velocity scale is obtained by

solving the transport equation The turbulent length scale is estimated from two

properties of the turbulence field namely the turbulent kinetic energy and the

dissipation rate The dissipation rate of the kinetic energy is obtained from its

transport equation The most widely used are k-ε and k-ω two equation models In

the next section the k-ε Wilcox k-ω BSL k-ω and SST models will be briefly

discussed

321 k-ε turbulence model

The k-epsilon model solves for two variables k the turbulent kinetic energy and

epsilon the rate of dissipation of kinetic energy Wall functions are used in this

model so the flow in the buffer region is not simulated The k-epsilon model is very

popular for industrial applications due to its good convergence rate and relatively

low memory requirements It does not very accurately compute flow fields that

exhibit adverse pressure gradients strong curvature to the flow or jet flow It does

perform well for external flow problems around complex geometries

The k-ε model introduces k (m2s2) as the turbulence kinetic energy and ε (m2s3) as

the turbulence eddy dissipation The continuity equation remains the same

ρ

+ nabla ∙ (ρ U) = 0 (35)

The momentum equation changes as shown by equation

δρ

δ + nabla ∙ (ρU⨂U) = minusnablapprime + nabla ∙ μ

nablaU + (nablaU) + S (36)

Where SM is the sum of body forces μeff is the effective viscosity accounting for

turbulence and prsquo is the modified pressure The k-ε model uses the concept of eddy

viscosity giving the equation for effective viscosity as shown by equation

μ

= μ + μ (37)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

28

μt is the turbulence viscosity is linked to the turbulence kinetic energy and

dissipation by the equation

μ = Cμρ

ε (38)

Where Cμ is a constant

The values for k and ε come from the differential transport equations for the

turbulence kinetic energy and the turbulence dissipation rate

The turbulence kinetic energy equation is given as equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus ρε (39)

The turbulence dissipation rate is given by equation

(ρε)

+ nabla ∙ (ρUε) = nabla ∙ μ +

μ

σε nablaε +

ε

(Cε (P + Pε ) minus Cε ρε) (310)

Where Cε1 Cε2 σk σε are constants

Pk is the turbulence production due to viscous forces and is modeled by the

equation

P = μ nablaU ∙ nablaU + nablaU minus

nabla ∙ U 3μ

nabla ∙ U + ρk (311)

If a buoyancy term is added to the previous equation if the full buoyancy model is

used

The buoyancy term Pkb is modeled as

P = minusμ

ρσρg ∙ nablaρ (312)

322 Wilcox k-ω turbulence model

The k-omega model is similar to k-epsilon instead however it solves for omega mdash

the specific rate of dissipation of kinetic energy It also uses wall functions and

therefore has comparable memory requirements It has more difficulty converging

and is quite sensitive to the initial guess at the solution Hence the k-epsilon model

is often used first to find an initial condition for solving the k-omega model The k-

omega model is useful in many cases where the k-epsilon model is not accurate

such as internal flows flows that exhibit strong curvature separated flows and jets

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

29

This model has an advantage over the k-ε model where it does not involve complex

linear damping functions for near wall calculations at low Reynolds numbers

The k-ω model assumes that the turbulence viscosity is related to the turbulence

kinetic energy k and the turbulent frequency ω by the equation

μ = ρ

ω (313)

The transport equation for k is given by the equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus βprimeρ k ω (314)

The transport equation for ω is shown as equation

(ρω)

δ + nabla ∙ (ρ U ω) = nabla ∙ μ +

μ

μω

nablaω + αω

P + Pω minus β ρ ω (315)

The production rate of turbulence (Pk) is calculated as shown previously in the k-ε

section

The recommended values for the constants in the above equations are

βprime = 009

α = 59

β = 0075

σ = 2

σω = 2

The Reynolds stress tensor ρ u⨂u is calculated by

minusρ u⨂u = μ nablaU + (nablaU) minus

part ρk + μ

nabla ∙ U (316)

323 Shear stress transport model

The disadvantage of the Wilcox model is the strong sensitivity to free-stream

conditions Therefore a blending of the k-ω model near the surface and the k-ε in

the outer region was made by Menter [23] which resulted in the formulation of the

BSL k-ω turbulence model It consists of a transformation of the k-ε model to a k-ω

formulation and subsequently adding the resulting equations The Wilcox model is

multiplied by a blending function F1 and the transformed k-ε by another function 1-

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

30

F1 F1 is a function of wall distance (being the value of one near the surface and zero

outside the boundary layer) Outside the boundary and on the edge of the boundary

layer the standard k-ε model is used [24]

SST model is a combination of the k-epsilon in the free stream and the k-omega

models near the walls It does not use wall functions and tends to be most accurate

when solving the flow near the wall The SST model does not always converge to

the solution quickly so the k-epsilon or k-omega models are often solved first to

give good initial conditions

The k-ω based Shear Stress Transport (SST) model of Menter was applied for

turbulence treatment The transport equations for the SST model are expressed

below where the turbulent kinetic energy lsquokrsquo and turbulent frequency or dissipation

per unit turbulent kinetic energy are computed by using the following relations

For Turbulence Kinetic Energy

kkPukt

kTkk Ntilde+Ntilde+-=Ntilde+

para

para)[()( nsnwb (317)

where Pk is the production limiter

For Specific Dissipation Rate

ww

swnsnbwaww

ww NtildeNtilde-+Ntilde+Ntilde+-=Ntilde+para

para

1)1(2])[()( 21

22 kFSut

T (318)

The first blending function F1 is calculated from

iumlthorn

iumlyacute

uuml

iumlicirc

iumliacute

igrave

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

uacuteuacuteucirc

ugrave

ecircecirceuml

eacute

dividedivideoslash

oumlccedilccedilegrave

aelig=

4

22

21

4500maxmintanh

yCD

k

yy

kF

kw

wrs

w

n

wb (319)

y is the distance to the nearest wall and v is the kinematic viscosity

divideoslash

oumlccedilegrave

aeligNtildeNtilde= -10

2 101

2max ww

rsww kCDk (320)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

31

And Kinematic eddy viscosity

)max( 21

1

SF

kT

wa

an = (321)

where F2 is a blending function which restricts the limiter to the wall boundary and

S is the invariant measure of the strain rate

uacuteuacute

ucirc

ugrave

ecircecirc

euml

eacute

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

dividedivideoslash

oumlccedilccedilegrave

aelig=

2

22

5002maxtanh

w

n

wb yy

kF (322)

33 Brief introduction of simulation process

All aspects of CFD study including modelling meshing solving and post-

processing of the results are briefly introduced in this chapter

331 Pre-processing technique

The speed and accuracy of simulation results depending fundamentally on meshing

and the ability to prepare model geometry correctly

The steps in pre-processing are explained as follows

1) Domain Definitions

The domain definition depends mainly on the physics involved in the problem and

objective of the problem

a) Geometry creation or geometry import

b) Geometry cleanup

i) Removing the parts not necessary for simulation under consideration

ii) Closing the gaps in the geometry

iii) Removing small surfaces curves to make meshing process simple

2)Grid generation

To solve the governing equations the numerical methods like Finite Difference

Finite Volume or Finite Element Method to divide the domain into small parts and

solve the equations on one part at a time are applied

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 18: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

6

Table 21 Acceptable velocity ranges for inlet bell diameter(ANSIHI 98)

Flow Rate

Qls

Recommended Inlet Bell Design

Velocityms

Acceptable Velocity

Range ms

Q<315 V=17 06leVle27

315 leQ<1260 V=17 09leVle24

Q ge1260 V=17 12leVle21

Fig 21 Recommended inlet bell design diameter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

7

212 Recommended dimensions for a rectangular sump

As the selected bell diameter has been determined the proportions of the inlet

structure can be estimated from Fig 22 and Fig 23 Table 22 shows the

recommended dimensions for a rectangular sump

Fig 22 Recommended intake structure layout

Fig 23 Filler wall details for proper bay width

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

8

Table 22 Recommended dimensions for a rectangular sump

Dimension Variable

Description Recommended value

A Distance from the pump inlet bell

centerline to the entrance A=5D minimum

a Length of constricted bay section near

the pump inlet 25D minimum

B Distance from the back wall to the

pump inlet bell centerline 075D

C Distance between the inlet bell and

floor 03D~05D

D Inlet bell outside diameter refer to section 211

H Minimum liquid depth H=S+C

h Minimum height of constricted bay

section near the pump inlet bell hgtH or 25D

S Minimum pump inlet bell submergence S=D(10+23Fr)

W Pump inlet bay entrance width 2D minimum

w Constricted bay width near the pump

inlet bell w=2D

X Pump inlet bay length 5D minimum

Y Distance from pump inlet bell

centerline to the through-flow traveling screen

4D minimum

Z1 Distance from sump inlet bell centerline to diverging walls

5D minimum

Z2 Distance from inlet bell centerline to

sloping floor 5D minimum

α Angle of floor slope -10deg~ +10deg

β Angle of wall convergence 0deg~ +10deg

empty Angle of convergence from constricted

area to by walls 10deg maximum

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

9

FD=Vb(gD)05 (21)

where

FD Froude number (dimensionless)

Vb velocity at suction inlet=FlowArea based on D

D Outside diameter of bell or pipe inlet

g gravitational acceleration

There is the design sequence summary for rectangular intake structures

middot Consider the flow patterns and boundary geometry of the body of liquid

from which the pump station is to receive flow Compare with the approach

flow condition and determine if a hydraulic physical model study is

required

middot Determine the number and size of pumps required to satisfy the range of

operating conditions likely to be encountered

middot Identify pump inlet bell diameter Refer to the section 211

middot Determine the bell-floor clearance 05D is a good preliminary design value

middot Determine the required bell submergence using the equation 21

middot Determine the minimum allowable liquid depth in the intake structure from

the sum of the floor clearance and the required bell submergence

middot Check bottom elevation near the entrance to the structure and determine if it

is necessary to slope the floor upstream of the bay entrance If the resulting

depth at the entrance to the intake structure is shallow then check to ensure

that gravity-driven flow is not restricted by the entrance condition

middot Check the pump bay velocity for the maximum single-pump flow and

minimum liquid depth with the bay width set to 2D If bay velocity exceeds

05 ms then increase the bay width to reduce the maximum velocity below

05 ms

middot If it is necessary to increase the pump bay width to greater than 2D then

decrease bay width in the vicinity of the pumps according to Fig 23

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

10

middot Compare cross-flow velocity (at maximum system flow) to average pump

bay velocity If cross-flow value exceeds 50 of the bay velocity a

physical hydraulic model study is necessary

middot Determine the length of the structure and dividing walls giving

consideration to minimum allowable distances to a sloping floor screening

equipment and length of dividing walls if dual flow traveling screens or

drum screens are to be used a physical hydraulic model study is required

middot If the final selected pump bell diameter and inlet velocity is within the

range given in section 211 then the sump dimensions (developed based on

the inlet bell design diameter) need not be changed and will comply with

these standards

22 Model test

The approach flow pattern in a pump intake facility is difficult to predict by

traditional mathematics or empirical formulae For large structures or those that

differ significantly from proven designs a model study is the only means to ensure

success A model study is used to identify adverse hydraulic conditions and

derive remedial measures for approach flow patterns generated by structures

upstream of the pump impeller Current model studies are not intended to

investigate flow patterns induced by the pump itself or the flow patterns within the

pump The objective of a model study is to ensure that the pump intake structure

generates favorable flow conditions at the inlet to the pump Evaluation for the

requirement of model test if

a Sump or piping geometry that deviates from this design standard

b Non-uniform or non-symmetric approach flow to the pump sump

exist

c The pumps have flows greater than 2520 ls per pump or the total

station flow with all pumps running would be greater than 6310 ls

d Situations that need ten times cost than the model study

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

11

Adverse hydraulic conditions that can affect pump performance include free and

submerged vortices swirl approaching the pump impeller flow separation at the

pump bell and a non-uniform axial velocity distribution at the suction

The purpose of model test is to examine these adverse hydraulic conditions and to

investigate measure against them

221 The scale effects

One possible difficulty in scale modeling is that it is impossible to reduce all

pertinent forces by the same factor With Froude-scale models the inertial and

gravitational forces are reduced similarly but the viscous and surface tension forces

cannot be simultaneously reduced The extra influence of these forces is known as

a ldquoscale effectrdquo A number of investigations have indicated that the scale effects for

the model are avoided if the Reynolds number based on the inlet flow and

submergence or intake diameter exceeds approximately 3x 104 and the Weber

number exceeds 150 [19]

In most engineering applications involving closed-conduit and open-channel flow

the Reynolds number limits are far exceeded and the flows are fully turbulent

In a sump model Froude number (Fr) and Reynolds number (Re) are the most

important non-dimensional parameters Some reduction in Re should be

introduced to the sump model However flow patterns are generally very similar

at high Re

No specific geometric scale rate is recommended but reasonable large

geometric scale to minimize viscous and surface tension scale effects and reproduce

the flow pattern

Re= uDν (22)

We= u2ρDσ (23)

where

u average axial velocity (ms)

ν kinematic viscosity of the liquid (m2s)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

12

ρ liquid density (kgm3)

σ surface tension of liquidair interface

According to the ANSIHI 98 the viscous scale effects on vortex may be negligible

if the resulting dimensionless numbers must meet these values Re>6times104 and

We>240

To allow practical visual observations of flow patterns accurate measurements of

swirl and velocity distribution and sufficient dimensional control the model scale

shall yield a bay width of at least 300mm a minimum liquid depth of at least

150mm and a pump throat or suction diameter of at least 80mm

222 Principle of similarity

In the similitude analysis the geometric and flow similarity requires the model and

the flow to be identical as the real model and flow This is to ensure the result

obtained from the model study is well presented in order to predict full scale

behavior Model involving a free surface are operated using Froude similarity since

the flow process is controlled by gravity and inertial forces

For similarity of the flow patterns the Froude number shall be equal in model and

prototype

Fr =

=

(24)

and as the scale ratio λ was selected in section 221 the following equation is

obtained

V = V times (

) = V times λ or Q = Q times (

) = Q times λ (25)

where Fr is the Froude number λ is the scale ratio of model dimensions to prototype

The test of 15 times Froude scaled flow should be conducted to ascertain the

potential scale affect on free surface vortices

Models of closed conduit piping system leading to a pump suction are not operated

based on Froude similitude However flow patterns are generally very similar at

high Re A minimum value of 1times105 for the Re at the pump suction is

recommended to ensure the flow patterns are correctly scaled

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

13

223 Preliminary model test

The initial design shall be tested first to identify any hydraulic problems

1) Vortex Formation Mechanisms

There are various investigations that provide insight into the fundamental processes

leading to the development of vortices both in experiment and numerical simulation

Shin et al [20] demonstrated that two basic mechanisms lead to inlet vortex

formation The first mechanism involves the development of an inlet vortex due to

the amplification of ambient vorticity in the approach flow as vortex lines

are convicted into the inlet The second mechanism involves the development of a

trailing vortex in the vicinity of the intake as a result of the variation in

circulation along the inlet For this second case a vortex can develop in a flow that

is irrotational upstream and the vortex development therefore does not depend

on the presence of ambient vorticity Shin et al investigation on kinematic

parameters indicate that the strength of an inlet-vortex or trailing vortex system

increases with decreasing distance from the surface However for an inlet in

an upstream irrotational flow two counter rotating vortices can still trail from

the rear of the inlet

2) Classification of vortex type

As illustrated in Fig 24 vortices in the vicinity of pump intakes may be adjacent to

the channel bottom or a channel wall (ie submerged vortices) or they may appear

adjacent to the free surface (ie free surface vortex)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

14

(a) Schematic diagram of vortices in a sump

(b) Experimental results of different vortex

Fig 24 Typical vortex in a sump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

15

For the model visual observation dye was injected around the intake structure near

the surface and submergence as Fig 24 (b) shows And the model test results of

vortex was classified as Fig 25

(a) Free Surface Vortices

(b) Submerged Vortices

Fig 25 Vortex classifications

There are a number of researches focused on the numerical theory to detect the

inception of a visible vortex As the diameters of vortices are usually much smaller

compared with the computing grid size to compensate for the lack of vortex

resolution in numerical simulation a stretching vortex model is applied to the local

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

16

flow field around semi-analytically identified vortex positions [21][22] The brief

explanations are as follows

a) Free surface vortex

The criteria expressed by equation 26 and 27 is applied to determine the

visible inception of a free surface vortex

≫ 1 (26)

=

gt 1 (27)

where P is the pressure drop at the vortex core and Ph is the static head at the

vortex element and f is the drag force acting at a unit length of water surface

vortex dimple based on Stokes approximation and caused by the down flow

Further f is the force acting on the bubble in the direction from the high

pressure region to the low pressure region like buoyancy

b) Submerged vortex

Equation 28 is applied to determine the visible inception of a submerged vortex

infin =

infin

infin gt 1 (28)

where is the static pressure at the vortex core infin is the atmospheric pressure

and is the critical cavitation inception pressure and nearly equal to the

saturated vapor pressure

3) Swirl angle

Swirl angle predicts the intensity of flow rotation Fig 26 shows experiment

calculation method with swirl meter and attempts of the CFD calculation method

were conducted as Fig 27 A four blade zero-pitch swirl meter which is supported

by a low friction bearing will be installed at about four suction pipe diameter (d)

downstream from the pump suction to measure swirl angle of pump approaching

flow For the swirl meter the tip to tip blade diameter is 075d and the length in

flow direction is 06 and one of the four blade is painted yellow as a reference to

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

17

count revolutions As flow swirl is usually unsteady the observation time of swirl

meter reading should be a continuous period of time

Fig 26 Typical swirl meter

(29)

(210)

(211)

Where qV flow rotational speed at swirl meter (ms)

zV average axial velocity at swirl meter (ms)

n Number of revolution per minute (rpm)

d Diameter of pipe at the swirl meter (m)

q flow rate (m3s)

zV

Vqq 1tan -=

)(60

smnd

Vtimestimes

=p

q

)(4

2

smd

qVz

times=

p

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

18

For the swirl angle calculating method with CFD the key point is to obtain the

average tangential velocity so the swirl check circles were created as illustrated in

Fig 27 The swirl check circle is located at the section of 4d height with diameter

of 025d 05d and 075d respectively

Swirl check circle

Fig 27 Schematic view of swirl angle numerical calculation

23

1 7501 501 250

acute++

=aringaringaring

N

i

N

i

NNVNVNV

Vi

q (212)

where

qV average tangential velocity (ms)

V circumferential velocity of check point in check circle

respectively (ms)

N check point total numbers in single check circle

4) Velocity

Velocity measurement at bell throat by Pitot tube was implemented by ANSIHI

method Typical Pitot tube configuration is shown in Fig 28 Eight measuring

points are defined as Fig 29 Time averaged velocity at eight points was obtained

and these eight data were again averaged to give throat sectional spatial velocity

average

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

19

Fig 28 Example of Pitot tube Fig 29 Measuring points at bell throat

224 Test criteria

There is the criteria for satisfactory performance of the sump model test and

summarized in Table 23

1) Free surface and submerged vortices

Free surface vortices entering the pump must be less severe than vortices with

coherent core Dye core vortices may be acceptable only if they occur less than 10

of the time or only for infrequent pump operating conditions

Submerged vortex which starts from the sump walls induce vibration and noise

problems Dye core submerged vortices extending into the pump bell mouth is not

allowed

2) Swirl angles

Swirl angle is used to check the flow rotation in the suction pipe The calculation

method is referred to section 223

Short-term maximum and long-term average values must be less than 5 degrees

Maximum short-term swirl angles up to 7 degrees may be acceptable only if they

occur less than 10 of the time or only for infrequent pump operating conditions

3)Velocity

Each point time-averaged velocity shall not be larger than 10 of sectional average

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

20

Standard deviation of time-varying velocity at each point was obtained and its

value shall not be larger than 10 of the time-averaged velocity at each point

1) Time-averaged

Variation[] = 100acute-

a

a

V

VV aV Averaged at section

2) Time-varying

Standard deviation= )1(

)(1i

2

-

-aring=

n

xxn

i

Variation[] = (Standard deviation X ) x 100

X Averaged at point

Table 23 Permission criteria of ANSIHI

Vortex occurence Swirl angle

Free surface vortex Submerged vortex

Observation

time Above 10 minutes Above 10 minutes

Above 10

minutes

Permission

criteria

Dye-core

Vortex(F3)

below 110 total

time

Dye-core

Vortex(S2)

below 110 total

time

Below

5degConditional

intermittent

below 7deg

225 Remedial measures for problem intakes

Usually in the preliminary test the worst flow condition is simulated And based on

the results remedial measures for problem intakes are discussed in this part

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

21

1) Overall approach flow patterns

The characteristics of flow approaching an intake structure is one of the foremost

considerations for the designer It is difficult to predict the effects of a given set of

flow conditions upstream from an intake structure Fig 210 depicts several typical

approach flow patterns of pump sump with different operating conditions

Fig 210 Examples of approach flow patterns

Generally speaking for the intake problems especially for the multiple pumps

sump there are the main measures using dividing walls placed between the pumps

as a partitioned structure trash racks with elongated bars guide piers or a number

of structural columns at the bay entrance can provide some assistance in controlling

cross-flow baffles flow distributors and screen mesh etc are applied to improve

the flow distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

22

2) Problems around pump intake

a) Free surface vortices

There are a number of factors that have an effect on the formation of surface

vortices To achieve a higher degree of certainty that objectionable surface vortices

do not form modifications can be made to intake structure to allow operation at

practical depths of submergence and the minimum submergence is referred in Table

22 the use of suction umbrellas vertical curtain walls and horizontal gratings

b) Submerged vortices

In the vicinity of intake bell there are the most complicated flow patterns and flow

makes the most changes in direction while maintaining a constant acceleration into

the pump bell to prevent local flow separation turbulence and vortex formation

Fig 211 represents a sample of various devices to inhibit the different submerged

vortex

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

23

Fig 211 Methods to reduce submerged vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

24

c) Preswirl

The approach flow distributions determine whether flow preswirl exists and a

sufficiently laterally skewed approach flow cause rotating around the intake bell

The most effective way of reducing preswirl is to establish a relatively uniform

approach flow within each pump bay by using the baffling schemes controlling

cross-flow and expanding concentrated flow discussed in the above part

To prevent non-uniform velocity distribution swirls and vortices necessary

modifications are conducted such as fore bay length baffle wall distributors

curtain wall mesh screen guide vane floor splitter wall fillet etc As the above

measures applied modification test will be conducted to check all the flow

conditions and compare with the preliminary results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

25

Chapter 3 Methodology

CFD (Computational fluid dynamics) is the technique which can design cycles to

better performance and reduce costs and time In CFD we solve the governing

equations of given physics (may be differential form or integral form) using some

numerical techniques like Finite Difference Method (FDM) Finite Element Method

(FEM) or Finite Volume Method (FVM) Although CFD plays a great part in the

design of turbomachinery CFD is not a replacement for experimental or analytical

approach In this study the numerical analysis was performed by the commercial

code ANSYS CFX 130 All aspects of the CFD study including modelling

meshing solving and post-processing of the results

31 Numerical modeling

ANSYS CFX solves the unsteady Navier-Stokes equations in their conservation

form The instantaneous equation of mass (continuity) in the stationary frame is

expressed as equation

ρ

δ + nabla ∙ (ρU) = 0 (31)

And the instantaneous equation for momentum is expressed as shown in equation

ρ

δ + nabla ∙ (ρU⨂U) = minusnablap + nabla ∙ τ + S (32)

Where τ is the stress tensor ( including both normal and shear components of the

stress)

These instantaneous equations are averaged for turbulent flows leading to additional

terms that need to be solved While the Navier-Stokes equations describe both

laminar and turbulent flows without addition terms realistic flows involve length

scales much smaller than the smallest finite volume mesh A Direct Numerical

Simulation of these flows would require significantly more computing power than

what is available now or in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

26

Therefore much research has been done to predict the effects of turbulence by

using turbulence models These models account for the effects of turbulence

without the use of a very fine mesh or direct numerical simulation

These turbulence models modify the transport equations by adding averaged and

fluctuating components The transport equations are changed to equations (33) and

(34)

ρ

δ + nabla ∙ (ρU) = 0 (33)

ρ

+ nabla ∙ ρ U ⨂U = minusnablap + nabla ∙ τ minus ρu⨂u + S (34)

The mass equation is not changed but the momentum equation contains extra terms

which are the Reynolds stresses ρ u⨂u and the Reynolds flux ρ u ϕ These

Reynolds stresses need to be modeled by additional equations to obtain closure

Obtaining closure implies that there are a sufficient number of equations to solve for

all the unknowns including the Reynolds stresses and Reynolds fluxes

Various turbulence models provide various ways to obtain closure In this

investigation Shear Stress Transport (SST) model and k-ε model were utilized The

advantage of SST model is that combines the advantages of other turbulence models

(the k-ε Wilcox k-ω) The 3 turbulence models will be discussed briefly

32 Turbulence models

Turbulence models are necessary because we cannot afford big enough computers

to directly capture every scale of motion There are three modelling frameworks

Direct Numerical Simulation (DNS) Reynolds-Averaged Navier-Stokes

Equations(RANS) Large Eddy Simulation(LES)

As RANS method is applied in most productionseveral RANS turbulence models

differ in their usage of wall functions the number of additional variables solved for

and what these variables represent

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

27

Two equation turbulence models

Two-equation turbulence models are widely used in CFD as they give a good

compromise between computational power needed and accuracy The term lsquotwo

equationrsquo refers to the fact that these models solve for the velocity and length scales

using separate transport equations The turbulence velocity scale is obtained by

solving the transport equation The turbulent length scale is estimated from two

properties of the turbulence field namely the turbulent kinetic energy and the

dissipation rate The dissipation rate of the kinetic energy is obtained from its

transport equation The most widely used are k-ε and k-ω two equation models In

the next section the k-ε Wilcox k-ω BSL k-ω and SST models will be briefly

discussed

321 k-ε turbulence model

The k-epsilon model solves for two variables k the turbulent kinetic energy and

epsilon the rate of dissipation of kinetic energy Wall functions are used in this

model so the flow in the buffer region is not simulated The k-epsilon model is very

popular for industrial applications due to its good convergence rate and relatively

low memory requirements It does not very accurately compute flow fields that

exhibit adverse pressure gradients strong curvature to the flow or jet flow It does

perform well for external flow problems around complex geometries

The k-ε model introduces k (m2s2) as the turbulence kinetic energy and ε (m2s3) as

the turbulence eddy dissipation The continuity equation remains the same

ρ

+ nabla ∙ (ρ U) = 0 (35)

The momentum equation changes as shown by equation

δρ

δ + nabla ∙ (ρU⨂U) = minusnablapprime + nabla ∙ μ

nablaU + (nablaU) + S (36)

Where SM is the sum of body forces μeff is the effective viscosity accounting for

turbulence and prsquo is the modified pressure The k-ε model uses the concept of eddy

viscosity giving the equation for effective viscosity as shown by equation

μ

= μ + μ (37)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

28

μt is the turbulence viscosity is linked to the turbulence kinetic energy and

dissipation by the equation

μ = Cμρ

ε (38)

Where Cμ is a constant

The values for k and ε come from the differential transport equations for the

turbulence kinetic energy and the turbulence dissipation rate

The turbulence kinetic energy equation is given as equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus ρε (39)

The turbulence dissipation rate is given by equation

(ρε)

+ nabla ∙ (ρUε) = nabla ∙ μ +

μ

σε nablaε +

ε

(Cε (P + Pε ) minus Cε ρε) (310)

Where Cε1 Cε2 σk σε are constants

Pk is the turbulence production due to viscous forces and is modeled by the

equation

P = μ nablaU ∙ nablaU + nablaU minus

nabla ∙ U 3μ

nabla ∙ U + ρk (311)

If a buoyancy term is added to the previous equation if the full buoyancy model is

used

The buoyancy term Pkb is modeled as

P = minusμ

ρσρg ∙ nablaρ (312)

322 Wilcox k-ω turbulence model

The k-omega model is similar to k-epsilon instead however it solves for omega mdash

the specific rate of dissipation of kinetic energy It also uses wall functions and

therefore has comparable memory requirements It has more difficulty converging

and is quite sensitive to the initial guess at the solution Hence the k-epsilon model

is often used first to find an initial condition for solving the k-omega model The k-

omega model is useful in many cases where the k-epsilon model is not accurate

such as internal flows flows that exhibit strong curvature separated flows and jets

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

29

This model has an advantage over the k-ε model where it does not involve complex

linear damping functions for near wall calculations at low Reynolds numbers

The k-ω model assumes that the turbulence viscosity is related to the turbulence

kinetic energy k and the turbulent frequency ω by the equation

μ = ρ

ω (313)

The transport equation for k is given by the equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus βprimeρ k ω (314)

The transport equation for ω is shown as equation

(ρω)

δ + nabla ∙ (ρ U ω) = nabla ∙ μ +

μ

μω

nablaω + αω

P + Pω minus β ρ ω (315)

The production rate of turbulence (Pk) is calculated as shown previously in the k-ε

section

The recommended values for the constants in the above equations are

βprime = 009

α = 59

β = 0075

σ = 2

σω = 2

The Reynolds stress tensor ρ u⨂u is calculated by

minusρ u⨂u = μ nablaU + (nablaU) minus

part ρk + μ

nabla ∙ U (316)

323 Shear stress transport model

The disadvantage of the Wilcox model is the strong sensitivity to free-stream

conditions Therefore a blending of the k-ω model near the surface and the k-ε in

the outer region was made by Menter [23] which resulted in the formulation of the

BSL k-ω turbulence model It consists of a transformation of the k-ε model to a k-ω

formulation and subsequently adding the resulting equations The Wilcox model is

multiplied by a blending function F1 and the transformed k-ε by another function 1-

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

30

F1 F1 is a function of wall distance (being the value of one near the surface and zero

outside the boundary layer) Outside the boundary and on the edge of the boundary

layer the standard k-ε model is used [24]

SST model is a combination of the k-epsilon in the free stream and the k-omega

models near the walls It does not use wall functions and tends to be most accurate

when solving the flow near the wall The SST model does not always converge to

the solution quickly so the k-epsilon or k-omega models are often solved first to

give good initial conditions

The k-ω based Shear Stress Transport (SST) model of Menter was applied for

turbulence treatment The transport equations for the SST model are expressed

below where the turbulent kinetic energy lsquokrsquo and turbulent frequency or dissipation

per unit turbulent kinetic energy are computed by using the following relations

For Turbulence Kinetic Energy

kkPukt

kTkk Ntilde+Ntilde+-=Ntilde+

para

para)[()( nsnwb (317)

where Pk is the production limiter

For Specific Dissipation Rate

ww

swnsnbwaww

ww NtildeNtilde-+Ntilde+Ntilde+-=Ntilde+para

para

1)1(2])[()( 21

22 kFSut

T (318)

The first blending function F1 is calculated from

iumlthorn

iumlyacute

uuml

iumlicirc

iumliacute

igrave

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

uacuteuacuteucirc

ugrave

ecircecirceuml

eacute

dividedivideoslash

oumlccedilccedilegrave

aelig=

4

22

21

4500maxmintanh

yCD

k

yy

kF

kw

wrs

w

n

wb (319)

y is the distance to the nearest wall and v is the kinematic viscosity

divideoslash

oumlccedilegrave

aeligNtildeNtilde= -10

2 101

2max ww

rsww kCDk (320)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

31

And Kinematic eddy viscosity

)max( 21

1

SF

kT

wa

an = (321)

where F2 is a blending function which restricts the limiter to the wall boundary and

S is the invariant measure of the strain rate

uacuteuacute

ucirc

ugrave

ecircecirc

euml

eacute

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

dividedivideoslash

oumlccedilccedilegrave

aelig=

2

22

5002maxtanh

w

n

wb yy

kF (322)

33 Brief introduction of simulation process

All aspects of CFD study including modelling meshing solving and post-

processing of the results are briefly introduced in this chapter

331 Pre-processing technique

The speed and accuracy of simulation results depending fundamentally on meshing

and the ability to prepare model geometry correctly

The steps in pre-processing are explained as follows

1) Domain Definitions

The domain definition depends mainly on the physics involved in the problem and

objective of the problem

a) Geometry creation or geometry import

b) Geometry cleanup

i) Removing the parts not necessary for simulation under consideration

ii) Closing the gaps in the geometry

iii) Removing small surfaces curves to make meshing process simple

2)Grid generation

To solve the governing equations the numerical methods like Finite Difference

Finite Volume or Finite Element Method to divide the domain into small parts and

solve the equations on one part at a time are applied

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 19: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

7

212 Recommended dimensions for a rectangular sump

As the selected bell diameter has been determined the proportions of the inlet

structure can be estimated from Fig 22 and Fig 23 Table 22 shows the

recommended dimensions for a rectangular sump

Fig 22 Recommended intake structure layout

Fig 23 Filler wall details for proper bay width

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

8

Table 22 Recommended dimensions for a rectangular sump

Dimension Variable

Description Recommended value

A Distance from the pump inlet bell

centerline to the entrance A=5D minimum

a Length of constricted bay section near

the pump inlet 25D minimum

B Distance from the back wall to the

pump inlet bell centerline 075D

C Distance between the inlet bell and

floor 03D~05D

D Inlet bell outside diameter refer to section 211

H Minimum liquid depth H=S+C

h Minimum height of constricted bay

section near the pump inlet bell hgtH or 25D

S Minimum pump inlet bell submergence S=D(10+23Fr)

W Pump inlet bay entrance width 2D minimum

w Constricted bay width near the pump

inlet bell w=2D

X Pump inlet bay length 5D minimum

Y Distance from pump inlet bell

centerline to the through-flow traveling screen

4D minimum

Z1 Distance from sump inlet bell centerline to diverging walls

5D minimum

Z2 Distance from inlet bell centerline to

sloping floor 5D minimum

α Angle of floor slope -10deg~ +10deg

β Angle of wall convergence 0deg~ +10deg

empty Angle of convergence from constricted

area to by walls 10deg maximum

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

9

FD=Vb(gD)05 (21)

where

FD Froude number (dimensionless)

Vb velocity at suction inlet=FlowArea based on D

D Outside diameter of bell or pipe inlet

g gravitational acceleration

There is the design sequence summary for rectangular intake structures

middot Consider the flow patterns and boundary geometry of the body of liquid

from which the pump station is to receive flow Compare with the approach

flow condition and determine if a hydraulic physical model study is

required

middot Determine the number and size of pumps required to satisfy the range of

operating conditions likely to be encountered

middot Identify pump inlet bell diameter Refer to the section 211

middot Determine the bell-floor clearance 05D is a good preliminary design value

middot Determine the required bell submergence using the equation 21

middot Determine the minimum allowable liquid depth in the intake structure from

the sum of the floor clearance and the required bell submergence

middot Check bottom elevation near the entrance to the structure and determine if it

is necessary to slope the floor upstream of the bay entrance If the resulting

depth at the entrance to the intake structure is shallow then check to ensure

that gravity-driven flow is not restricted by the entrance condition

middot Check the pump bay velocity for the maximum single-pump flow and

minimum liquid depth with the bay width set to 2D If bay velocity exceeds

05 ms then increase the bay width to reduce the maximum velocity below

05 ms

middot If it is necessary to increase the pump bay width to greater than 2D then

decrease bay width in the vicinity of the pumps according to Fig 23

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

10

middot Compare cross-flow velocity (at maximum system flow) to average pump

bay velocity If cross-flow value exceeds 50 of the bay velocity a

physical hydraulic model study is necessary

middot Determine the length of the structure and dividing walls giving

consideration to minimum allowable distances to a sloping floor screening

equipment and length of dividing walls if dual flow traveling screens or

drum screens are to be used a physical hydraulic model study is required

middot If the final selected pump bell diameter and inlet velocity is within the

range given in section 211 then the sump dimensions (developed based on

the inlet bell design diameter) need not be changed and will comply with

these standards

22 Model test

The approach flow pattern in a pump intake facility is difficult to predict by

traditional mathematics or empirical formulae For large structures or those that

differ significantly from proven designs a model study is the only means to ensure

success A model study is used to identify adverse hydraulic conditions and

derive remedial measures for approach flow patterns generated by structures

upstream of the pump impeller Current model studies are not intended to

investigate flow patterns induced by the pump itself or the flow patterns within the

pump The objective of a model study is to ensure that the pump intake structure

generates favorable flow conditions at the inlet to the pump Evaluation for the

requirement of model test if

a Sump or piping geometry that deviates from this design standard

b Non-uniform or non-symmetric approach flow to the pump sump

exist

c The pumps have flows greater than 2520 ls per pump or the total

station flow with all pumps running would be greater than 6310 ls

d Situations that need ten times cost than the model study

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

11

Adverse hydraulic conditions that can affect pump performance include free and

submerged vortices swirl approaching the pump impeller flow separation at the

pump bell and a non-uniform axial velocity distribution at the suction

The purpose of model test is to examine these adverse hydraulic conditions and to

investigate measure against them

221 The scale effects

One possible difficulty in scale modeling is that it is impossible to reduce all

pertinent forces by the same factor With Froude-scale models the inertial and

gravitational forces are reduced similarly but the viscous and surface tension forces

cannot be simultaneously reduced The extra influence of these forces is known as

a ldquoscale effectrdquo A number of investigations have indicated that the scale effects for

the model are avoided if the Reynolds number based on the inlet flow and

submergence or intake diameter exceeds approximately 3x 104 and the Weber

number exceeds 150 [19]

In most engineering applications involving closed-conduit and open-channel flow

the Reynolds number limits are far exceeded and the flows are fully turbulent

In a sump model Froude number (Fr) and Reynolds number (Re) are the most

important non-dimensional parameters Some reduction in Re should be

introduced to the sump model However flow patterns are generally very similar

at high Re

No specific geometric scale rate is recommended but reasonable large

geometric scale to minimize viscous and surface tension scale effects and reproduce

the flow pattern

Re= uDν (22)

We= u2ρDσ (23)

where

u average axial velocity (ms)

ν kinematic viscosity of the liquid (m2s)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

12

ρ liquid density (kgm3)

σ surface tension of liquidair interface

According to the ANSIHI 98 the viscous scale effects on vortex may be negligible

if the resulting dimensionless numbers must meet these values Re>6times104 and

We>240

To allow practical visual observations of flow patterns accurate measurements of

swirl and velocity distribution and sufficient dimensional control the model scale

shall yield a bay width of at least 300mm a minimum liquid depth of at least

150mm and a pump throat or suction diameter of at least 80mm

222 Principle of similarity

In the similitude analysis the geometric and flow similarity requires the model and

the flow to be identical as the real model and flow This is to ensure the result

obtained from the model study is well presented in order to predict full scale

behavior Model involving a free surface are operated using Froude similarity since

the flow process is controlled by gravity and inertial forces

For similarity of the flow patterns the Froude number shall be equal in model and

prototype

Fr =

=

(24)

and as the scale ratio λ was selected in section 221 the following equation is

obtained

V = V times (

) = V times λ or Q = Q times (

) = Q times λ (25)

where Fr is the Froude number λ is the scale ratio of model dimensions to prototype

The test of 15 times Froude scaled flow should be conducted to ascertain the

potential scale affect on free surface vortices

Models of closed conduit piping system leading to a pump suction are not operated

based on Froude similitude However flow patterns are generally very similar at

high Re A minimum value of 1times105 for the Re at the pump suction is

recommended to ensure the flow patterns are correctly scaled

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

13

223 Preliminary model test

The initial design shall be tested first to identify any hydraulic problems

1) Vortex Formation Mechanisms

There are various investigations that provide insight into the fundamental processes

leading to the development of vortices both in experiment and numerical simulation

Shin et al [20] demonstrated that two basic mechanisms lead to inlet vortex

formation The first mechanism involves the development of an inlet vortex due to

the amplification of ambient vorticity in the approach flow as vortex lines

are convicted into the inlet The second mechanism involves the development of a

trailing vortex in the vicinity of the intake as a result of the variation in

circulation along the inlet For this second case a vortex can develop in a flow that

is irrotational upstream and the vortex development therefore does not depend

on the presence of ambient vorticity Shin et al investigation on kinematic

parameters indicate that the strength of an inlet-vortex or trailing vortex system

increases with decreasing distance from the surface However for an inlet in

an upstream irrotational flow two counter rotating vortices can still trail from

the rear of the inlet

2) Classification of vortex type

As illustrated in Fig 24 vortices in the vicinity of pump intakes may be adjacent to

the channel bottom or a channel wall (ie submerged vortices) or they may appear

adjacent to the free surface (ie free surface vortex)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

14

(a) Schematic diagram of vortices in a sump

(b) Experimental results of different vortex

Fig 24 Typical vortex in a sump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

15

For the model visual observation dye was injected around the intake structure near

the surface and submergence as Fig 24 (b) shows And the model test results of

vortex was classified as Fig 25

(a) Free Surface Vortices

(b) Submerged Vortices

Fig 25 Vortex classifications

There are a number of researches focused on the numerical theory to detect the

inception of a visible vortex As the diameters of vortices are usually much smaller

compared with the computing grid size to compensate for the lack of vortex

resolution in numerical simulation a stretching vortex model is applied to the local

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

16

flow field around semi-analytically identified vortex positions [21][22] The brief

explanations are as follows

a) Free surface vortex

The criteria expressed by equation 26 and 27 is applied to determine the

visible inception of a free surface vortex

≫ 1 (26)

=

gt 1 (27)

where P is the pressure drop at the vortex core and Ph is the static head at the

vortex element and f is the drag force acting at a unit length of water surface

vortex dimple based on Stokes approximation and caused by the down flow

Further f is the force acting on the bubble in the direction from the high

pressure region to the low pressure region like buoyancy

b) Submerged vortex

Equation 28 is applied to determine the visible inception of a submerged vortex

infin =

infin

infin gt 1 (28)

where is the static pressure at the vortex core infin is the atmospheric pressure

and is the critical cavitation inception pressure and nearly equal to the

saturated vapor pressure

3) Swirl angle

Swirl angle predicts the intensity of flow rotation Fig 26 shows experiment

calculation method with swirl meter and attempts of the CFD calculation method

were conducted as Fig 27 A four blade zero-pitch swirl meter which is supported

by a low friction bearing will be installed at about four suction pipe diameter (d)

downstream from the pump suction to measure swirl angle of pump approaching

flow For the swirl meter the tip to tip blade diameter is 075d and the length in

flow direction is 06 and one of the four blade is painted yellow as a reference to

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

17

count revolutions As flow swirl is usually unsteady the observation time of swirl

meter reading should be a continuous period of time

Fig 26 Typical swirl meter

(29)

(210)

(211)

Where qV flow rotational speed at swirl meter (ms)

zV average axial velocity at swirl meter (ms)

n Number of revolution per minute (rpm)

d Diameter of pipe at the swirl meter (m)

q flow rate (m3s)

zV

Vqq 1tan -=

)(60

smnd

Vtimestimes

=p

q

)(4

2

smd

qVz

times=

p

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

18

For the swirl angle calculating method with CFD the key point is to obtain the

average tangential velocity so the swirl check circles were created as illustrated in

Fig 27 The swirl check circle is located at the section of 4d height with diameter

of 025d 05d and 075d respectively

Swirl check circle

Fig 27 Schematic view of swirl angle numerical calculation

23

1 7501 501 250

acute++

=aringaringaring

N

i

N

i

NNVNVNV

Vi

q (212)

where

qV average tangential velocity (ms)

V circumferential velocity of check point in check circle

respectively (ms)

N check point total numbers in single check circle

4) Velocity

Velocity measurement at bell throat by Pitot tube was implemented by ANSIHI

method Typical Pitot tube configuration is shown in Fig 28 Eight measuring

points are defined as Fig 29 Time averaged velocity at eight points was obtained

and these eight data were again averaged to give throat sectional spatial velocity

average

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

19

Fig 28 Example of Pitot tube Fig 29 Measuring points at bell throat

224 Test criteria

There is the criteria for satisfactory performance of the sump model test and

summarized in Table 23

1) Free surface and submerged vortices

Free surface vortices entering the pump must be less severe than vortices with

coherent core Dye core vortices may be acceptable only if they occur less than 10

of the time or only for infrequent pump operating conditions

Submerged vortex which starts from the sump walls induce vibration and noise

problems Dye core submerged vortices extending into the pump bell mouth is not

allowed

2) Swirl angles

Swirl angle is used to check the flow rotation in the suction pipe The calculation

method is referred to section 223

Short-term maximum and long-term average values must be less than 5 degrees

Maximum short-term swirl angles up to 7 degrees may be acceptable only if they

occur less than 10 of the time or only for infrequent pump operating conditions

3)Velocity

Each point time-averaged velocity shall not be larger than 10 of sectional average

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

20

Standard deviation of time-varying velocity at each point was obtained and its

value shall not be larger than 10 of the time-averaged velocity at each point

1) Time-averaged

Variation[] = 100acute-

a

a

V

VV aV Averaged at section

2) Time-varying

Standard deviation= )1(

)(1i

2

-

-aring=

n

xxn

i

Variation[] = (Standard deviation X ) x 100

X Averaged at point

Table 23 Permission criteria of ANSIHI

Vortex occurence Swirl angle

Free surface vortex Submerged vortex

Observation

time Above 10 minutes Above 10 minutes

Above 10

minutes

Permission

criteria

Dye-core

Vortex(F3)

below 110 total

time

Dye-core

Vortex(S2)

below 110 total

time

Below

5degConditional

intermittent

below 7deg

225 Remedial measures for problem intakes

Usually in the preliminary test the worst flow condition is simulated And based on

the results remedial measures for problem intakes are discussed in this part

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

21

1) Overall approach flow patterns

The characteristics of flow approaching an intake structure is one of the foremost

considerations for the designer It is difficult to predict the effects of a given set of

flow conditions upstream from an intake structure Fig 210 depicts several typical

approach flow patterns of pump sump with different operating conditions

Fig 210 Examples of approach flow patterns

Generally speaking for the intake problems especially for the multiple pumps

sump there are the main measures using dividing walls placed between the pumps

as a partitioned structure trash racks with elongated bars guide piers or a number

of structural columns at the bay entrance can provide some assistance in controlling

cross-flow baffles flow distributors and screen mesh etc are applied to improve

the flow distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

22

2) Problems around pump intake

a) Free surface vortices

There are a number of factors that have an effect on the formation of surface

vortices To achieve a higher degree of certainty that objectionable surface vortices

do not form modifications can be made to intake structure to allow operation at

practical depths of submergence and the minimum submergence is referred in Table

22 the use of suction umbrellas vertical curtain walls and horizontal gratings

b) Submerged vortices

In the vicinity of intake bell there are the most complicated flow patterns and flow

makes the most changes in direction while maintaining a constant acceleration into

the pump bell to prevent local flow separation turbulence and vortex formation

Fig 211 represents a sample of various devices to inhibit the different submerged

vortex

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

23

Fig 211 Methods to reduce submerged vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

24

c) Preswirl

The approach flow distributions determine whether flow preswirl exists and a

sufficiently laterally skewed approach flow cause rotating around the intake bell

The most effective way of reducing preswirl is to establish a relatively uniform

approach flow within each pump bay by using the baffling schemes controlling

cross-flow and expanding concentrated flow discussed in the above part

To prevent non-uniform velocity distribution swirls and vortices necessary

modifications are conducted such as fore bay length baffle wall distributors

curtain wall mesh screen guide vane floor splitter wall fillet etc As the above

measures applied modification test will be conducted to check all the flow

conditions and compare with the preliminary results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

25

Chapter 3 Methodology

CFD (Computational fluid dynamics) is the technique which can design cycles to

better performance and reduce costs and time In CFD we solve the governing

equations of given physics (may be differential form or integral form) using some

numerical techniques like Finite Difference Method (FDM) Finite Element Method

(FEM) or Finite Volume Method (FVM) Although CFD plays a great part in the

design of turbomachinery CFD is not a replacement for experimental or analytical

approach In this study the numerical analysis was performed by the commercial

code ANSYS CFX 130 All aspects of the CFD study including modelling

meshing solving and post-processing of the results

31 Numerical modeling

ANSYS CFX solves the unsteady Navier-Stokes equations in their conservation

form The instantaneous equation of mass (continuity) in the stationary frame is

expressed as equation

ρ

δ + nabla ∙ (ρU) = 0 (31)

And the instantaneous equation for momentum is expressed as shown in equation

ρ

δ + nabla ∙ (ρU⨂U) = minusnablap + nabla ∙ τ + S (32)

Where τ is the stress tensor ( including both normal and shear components of the

stress)

These instantaneous equations are averaged for turbulent flows leading to additional

terms that need to be solved While the Navier-Stokes equations describe both

laminar and turbulent flows without addition terms realistic flows involve length

scales much smaller than the smallest finite volume mesh A Direct Numerical

Simulation of these flows would require significantly more computing power than

what is available now or in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

26

Therefore much research has been done to predict the effects of turbulence by

using turbulence models These models account for the effects of turbulence

without the use of a very fine mesh or direct numerical simulation

These turbulence models modify the transport equations by adding averaged and

fluctuating components The transport equations are changed to equations (33) and

(34)

ρ

δ + nabla ∙ (ρU) = 0 (33)

ρ

+ nabla ∙ ρ U ⨂U = minusnablap + nabla ∙ τ minus ρu⨂u + S (34)

The mass equation is not changed but the momentum equation contains extra terms

which are the Reynolds stresses ρ u⨂u and the Reynolds flux ρ u ϕ These

Reynolds stresses need to be modeled by additional equations to obtain closure

Obtaining closure implies that there are a sufficient number of equations to solve for

all the unknowns including the Reynolds stresses and Reynolds fluxes

Various turbulence models provide various ways to obtain closure In this

investigation Shear Stress Transport (SST) model and k-ε model were utilized The

advantage of SST model is that combines the advantages of other turbulence models

(the k-ε Wilcox k-ω) The 3 turbulence models will be discussed briefly

32 Turbulence models

Turbulence models are necessary because we cannot afford big enough computers

to directly capture every scale of motion There are three modelling frameworks

Direct Numerical Simulation (DNS) Reynolds-Averaged Navier-Stokes

Equations(RANS) Large Eddy Simulation(LES)

As RANS method is applied in most productionseveral RANS turbulence models

differ in their usage of wall functions the number of additional variables solved for

and what these variables represent

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

27

Two equation turbulence models

Two-equation turbulence models are widely used in CFD as they give a good

compromise between computational power needed and accuracy The term lsquotwo

equationrsquo refers to the fact that these models solve for the velocity and length scales

using separate transport equations The turbulence velocity scale is obtained by

solving the transport equation The turbulent length scale is estimated from two

properties of the turbulence field namely the turbulent kinetic energy and the

dissipation rate The dissipation rate of the kinetic energy is obtained from its

transport equation The most widely used are k-ε and k-ω two equation models In

the next section the k-ε Wilcox k-ω BSL k-ω and SST models will be briefly

discussed

321 k-ε turbulence model

The k-epsilon model solves for two variables k the turbulent kinetic energy and

epsilon the rate of dissipation of kinetic energy Wall functions are used in this

model so the flow in the buffer region is not simulated The k-epsilon model is very

popular for industrial applications due to its good convergence rate and relatively

low memory requirements It does not very accurately compute flow fields that

exhibit adverse pressure gradients strong curvature to the flow or jet flow It does

perform well for external flow problems around complex geometries

The k-ε model introduces k (m2s2) as the turbulence kinetic energy and ε (m2s3) as

the turbulence eddy dissipation The continuity equation remains the same

ρ

+ nabla ∙ (ρ U) = 0 (35)

The momentum equation changes as shown by equation

δρ

δ + nabla ∙ (ρU⨂U) = minusnablapprime + nabla ∙ μ

nablaU + (nablaU) + S (36)

Where SM is the sum of body forces μeff is the effective viscosity accounting for

turbulence and prsquo is the modified pressure The k-ε model uses the concept of eddy

viscosity giving the equation for effective viscosity as shown by equation

μ

= μ + μ (37)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

28

μt is the turbulence viscosity is linked to the turbulence kinetic energy and

dissipation by the equation

μ = Cμρ

ε (38)

Where Cμ is a constant

The values for k and ε come from the differential transport equations for the

turbulence kinetic energy and the turbulence dissipation rate

The turbulence kinetic energy equation is given as equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus ρε (39)

The turbulence dissipation rate is given by equation

(ρε)

+ nabla ∙ (ρUε) = nabla ∙ μ +

μ

σε nablaε +

ε

(Cε (P + Pε ) minus Cε ρε) (310)

Where Cε1 Cε2 σk σε are constants

Pk is the turbulence production due to viscous forces and is modeled by the

equation

P = μ nablaU ∙ nablaU + nablaU minus

nabla ∙ U 3μ

nabla ∙ U + ρk (311)

If a buoyancy term is added to the previous equation if the full buoyancy model is

used

The buoyancy term Pkb is modeled as

P = minusμ

ρσρg ∙ nablaρ (312)

322 Wilcox k-ω turbulence model

The k-omega model is similar to k-epsilon instead however it solves for omega mdash

the specific rate of dissipation of kinetic energy It also uses wall functions and

therefore has comparable memory requirements It has more difficulty converging

and is quite sensitive to the initial guess at the solution Hence the k-epsilon model

is often used first to find an initial condition for solving the k-omega model The k-

omega model is useful in many cases where the k-epsilon model is not accurate

such as internal flows flows that exhibit strong curvature separated flows and jets

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

29

This model has an advantage over the k-ε model where it does not involve complex

linear damping functions for near wall calculations at low Reynolds numbers

The k-ω model assumes that the turbulence viscosity is related to the turbulence

kinetic energy k and the turbulent frequency ω by the equation

μ = ρ

ω (313)

The transport equation for k is given by the equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus βprimeρ k ω (314)

The transport equation for ω is shown as equation

(ρω)

δ + nabla ∙ (ρ U ω) = nabla ∙ μ +

μ

μω

nablaω + αω

P + Pω minus β ρ ω (315)

The production rate of turbulence (Pk) is calculated as shown previously in the k-ε

section

The recommended values for the constants in the above equations are

βprime = 009

α = 59

β = 0075

σ = 2

σω = 2

The Reynolds stress tensor ρ u⨂u is calculated by

minusρ u⨂u = μ nablaU + (nablaU) minus

part ρk + μ

nabla ∙ U (316)

323 Shear stress transport model

The disadvantage of the Wilcox model is the strong sensitivity to free-stream

conditions Therefore a blending of the k-ω model near the surface and the k-ε in

the outer region was made by Menter [23] which resulted in the formulation of the

BSL k-ω turbulence model It consists of a transformation of the k-ε model to a k-ω

formulation and subsequently adding the resulting equations The Wilcox model is

multiplied by a blending function F1 and the transformed k-ε by another function 1-

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

30

F1 F1 is a function of wall distance (being the value of one near the surface and zero

outside the boundary layer) Outside the boundary and on the edge of the boundary

layer the standard k-ε model is used [24]

SST model is a combination of the k-epsilon in the free stream and the k-omega

models near the walls It does not use wall functions and tends to be most accurate

when solving the flow near the wall The SST model does not always converge to

the solution quickly so the k-epsilon or k-omega models are often solved first to

give good initial conditions

The k-ω based Shear Stress Transport (SST) model of Menter was applied for

turbulence treatment The transport equations for the SST model are expressed

below where the turbulent kinetic energy lsquokrsquo and turbulent frequency or dissipation

per unit turbulent kinetic energy are computed by using the following relations

For Turbulence Kinetic Energy

kkPukt

kTkk Ntilde+Ntilde+-=Ntilde+

para

para)[()( nsnwb (317)

where Pk is the production limiter

For Specific Dissipation Rate

ww

swnsnbwaww

ww NtildeNtilde-+Ntilde+Ntilde+-=Ntilde+para

para

1)1(2])[()( 21

22 kFSut

T (318)

The first blending function F1 is calculated from

iumlthorn

iumlyacute

uuml

iumlicirc

iumliacute

igrave

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

uacuteuacuteucirc

ugrave

ecircecirceuml

eacute

dividedivideoslash

oumlccedilccedilegrave

aelig=

4

22

21

4500maxmintanh

yCD

k

yy

kF

kw

wrs

w

n

wb (319)

y is the distance to the nearest wall and v is the kinematic viscosity

divideoslash

oumlccedilegrave

aeligNtildeNtilde= -10

2 101

2max ww

rsww kCDk (320)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

31

And Kinematic eddy viscosity

)max( 21

1

SF

kT

wa

an = (321)

where F2 is a blending function which restricts the limiter to the wall boundary and

S is the invariant measure of the strain rate

uacuteuacute

ucirc

ugrave

ecircecirc

euml

eacute

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

dividedivideoslash

oumlccedilccedilegrave

aelig=

2

22

5002maxtanh

w

n

wb yy

kF (322)

33 Brief introduction of simulation process

All aspects of CFD study including modelling meshing solving and post-

processing of the results are briefly introduced in this chapter

331 Pre-processing technique

The speed and accuracy of simulation results depending fundamentally on meshing

and the ability to prepare model geometry correctly

The steps in pre-processing are explained as follows

1) Domain Definitions

The domain definition depends mainly on the physics involved in the problem and

objective of the problem

a) Geometry creation or geometry import

b) Geometry cleanup

i) Removing the parts not necessary for simulation under consideration

ii) Closing the gaps in the geometry

iii) Removing small surfaces curves to make meshing process simple

2)Grid generation

To solve the governing equations the numerical methods like Finite Difference

Finite Volume or Finite Element Method to divide the domain into small parts and

solve the equations on one part at a time are applied

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 20: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

8

Table 22 Recommended dimensions for a rectangular sump

Dimension Variable

Description Recommended value

A Distance from the pump inlet bell

centerline to the entrance A=5D minimum

a Length of constricted bay section near

the pump inlet 25D minimum

B Distance from the back wall to the

pump inlet bell centerline 075D

C Distance between the inlet bell and

floor 03D~05D

D Inlet bell outside diameter refer to section 211

H Minimum liquid depth H=S+C

h Minimum height of constricted bay

section near the pump inlet bell hgtH or 25D

S Minimum pump inlet bell submergence S=D(10+23Fr)

W Pump inlet bay entrance width 2D minimum

w Constricted bay width near the pump

inlet bell w=2D

X Pump inlet bay length 5D minimum

Y Distance from pump inlet bell

centerline to the through-flow traveling screen

4D minimum

Z1 Distance from sump inlet bell centerline to diverging walls

5D minimum

Z2 Distance from inlet bell centerline to

sloping floor 5D minimum

α Angle of floor slope -10deg~ +10deg

β Angle of wall convergence 0deg~ +10deg

empty Angle of convergence from constricted

area to by walls 10deg maximum

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

9

FD=Vb(gD)05 (21)

where

FD Froude number (dimensionless)

Vb velocity at suction inlet=FlowArea based on D

D Outside diameter of bell or pipe inlet

g gravitational acceleration

There is the design sequence summary for rectangular intake structures

middot Consider the flow patterns and boundary geometry of the body of liquid

from which the pump station is to receive flow Compare with the approach

flow condition and determine if a hydraulic physical model study is

required

middot Determine the number and size of pumps required to satisfy the range of

operating conditions likely to be encountered

middot Identify pump inlet bell diameter Refer to the section 211

middot Determine the bell-floor clearance 05D is a good preliminary design value

middot Determine the required bell submergence using the equation 21

middot Determine the minimum allowable liquid depth in the intake structure from

the sum of the floor clearance and the required bell submergence

middot Check bottom elevation near the entrance to the structure and determine if it

is necessary to slope the floor upstream of the bay entrance If the resulting

depth at the entrance to the intake structure is shallow then check to ensure

that gravity-driven flow is not restricted by the entrance condition

middot Check the pump bay velocity for the maximum single-pump flow and

minimum liquid depth with the bay width set to 2D If bay velocity exceeds

05 ms then increase the bay width to reduce the maximum velocity below

05 ms

middot If it is necessary to increase the pump bay width to greater than 2D then

decrease bay width in the vicinity of the pumps according to Fig 23

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

10

middot Compare cross-flow velocity (at maximum system flow) to average pump

bay velocity If cross-flow value exceeds 50 of the bay velocity a

physical hydraulic model study is necessary

middot Determine the length of the structure and dividing walls giving

consideration to minimum allowable distances to a sloping floor screening

equipment and length of dividing walls if dual flow traveling screens or

drum screens are to be used a physical hydraulic model study is required

middot If the final selected pump bell diameter and inlet velocity is within the

range given in section 211 then the sump dimensions (developed based on

the inlet bell design diameter) need not be changed and will comply with

these standards

22 Model test

The approach flow pattern in a pump intake facility is difficult to predict by

traditional mathematics or empirical formulae For large structures or those that

differ significantly from proven designs a model study is the only means to ensure

success A model study is used to identify adverse hydraulic conditions and

derive remedial measures for approach flow patterns generated by structures

upstream of the pump impeller Current model studies are not intended to

investigate flow patterns induced by the pump itself or the flow patterns within the

pump The objective of a model study is to ensure that the pump intake structure

generates favorable flow conditions at the inlet to the pump Evaluation for the

requirement of model test if

a Sump or piping geometry that deviates from this design standard

b Non-uniform or non-symmetric approach flow to the pump sump

exist

c The pumps have flows greater than 2520 ls per pump or the total

station flow with all pumps running would be greater than 6310 ls

d Situations that need ten times cost than the model study

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

11

Adverse hydraulic conditions that can affect pump performance include free and

submerged vortices swirl approaching the pump impeller flow separation at the

pump bell and a non-uniform axial velocity distribution at the suction

The purpose of model test is to examine these adverse hydraulic conditions and to

investigate measure against them

221 The scale effects

One possible difficulty in scale modeling is that it is impossible to reduce all

pertinent forces by the same factor With Froude-scale models the inertial and

gravitational forces are reduced similarly but the viscous and surface tension forces

cannot be simultaneously reduced The extra influence of these forces is known as

a ldquoscale effectrdquo A number of investigations have indicated that the scale effects for

the model are avoided if the Reynolds number based on the inlet flow and

submergence or intake diameter exceeds approximately 3x 104 and the Weber

number exceeds 150 [19]

In most engineering applications involving closed-conduit and open-channel flow

the Reynolds number limits are far exceeded and the flows are fully turbulent

In a sump model Froude number (Fr) and Reynolds number (Re) are the most

important non-dimensional parameters Some reduction in Re should be

introduced to the sump model However flow patterns are generally very similar

at high Re

No specific geometric scale rate is recommended but reasonable large

geometric scale to minimize viscous and surface tension scale effects and reproduce

the flow pattern

Re= uDν (22)

We= u2ρDσ (23)

where

u average axial velocity (ms)

ν kinematic viscosity of the liquid (m2s)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

12

ρ liquid density (kgm3)

σ surface tension of liquidair interface

According to the ANSIHI 98 the viscous scale effects on vortex may be negligible

if the resulting dimensionless numbers must meet these values Re>6times104 and

We>240

To allow practical visual observations of flow patterns accurate measurements of

swirl and velocity distribution and sufficient dimensional control the model scale

shall yield a bay width of at least 300mm a minimum liquid depth of at least

150mm and a pump throat or suction diameter of at least 80mm

222 Principle of similarity

In the similitude analysis the geometric and flow similarity requires the model and

the flow to be identical as the real model and flow This is to ensure the result

obtained from the model study is well presented in order to predict full scale

behavior Model involving a free surface are operated using Froude similarity since

the flow process is controlled by gravity and inertial forces

For similarity of the flow patterns the Froude number shall be equal in model and

prototype

Fr =

=

(24)

and as the scale ratio λ was selected in section 221 the following equation is

obtained

V = V times (

) = V times λ or Q = Q times (

) = Q times λ (25)

where Fr is the Froude number λ is the scale ratio of model dimensions to prototype

The test of 15 times Froude scaled flow should be conducted to ascertain the

potential scale affect on free surface vortices

Models of closed conduit piping system leading to a pump suction are not operated

based on Froude similitude However flow patterns are generally very similar at

high Re A minimum value of 1times105 for the Re at the pump suction is

recommended to ensure the flow patterns are correctly scaled

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

13

223 Preliminary model test

The initial design shall be tested first to identify any hydraulic problems

1) Vortex Formation Mechanisms

There are various investigations that provide insight into the fundamental processes

leading to the development of vortices both in experiment and numerical simulation

Shin et al [20] demonstrated that two basic mechanisms lead to inlet vortex

formation The first mechanism involves the development of an inlet vortex due to

the amplification of ambient vorticity in the approach flow as vortex lines

are convicted into the inlet The second mechanism involves the development of a

trailing vortex in the vicinity of the intake as a result of the variation in

circulation along the inlet For this second case a vortex can develop in a flow that

is irrotational upstream and the vortex development therefore does not depend

on the presence of ambient vorticity Shin et al investigation on kinematic

parameters indicate that the strength of an inlet-vortex or trailing vortex system

increases with decreasing distance from the surface However for an inlet in

an upstream irrotational flow two counter rotating vortices can still trail from

the rear of the inlet

2) Classification of vortex type

As illustrated in Fig 24 vortices in the vicinity of pump intakes may be adjacent to

the channel bottom or a channel wall (ie submerged vortices) or they may appear

adjacent to the free surface (ie free surface vortex)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

14

(a) Schematic diagram of vortices in a sump

(b) Experimental results of different vortex

Fig 24 Typical vortex in a sump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

15

For the model visual observation dye was injected around the intake structure near

the surface and submergence as Fig 24 (b) shows And the model test results of

vortex was classified as Fig 25

(a) Free Surface Vortices

(b) Submerged Vortices

Fig 25 Vortex classifications

There are a number of researches focused on the numerical theory to detect the

inception of a visible vortex As the diameters of vortices are usually much smaller

compared with the computing grid size to compensate for the lack of vortex

resolution in numerical simulation a stretching vortex model is applied to the local

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

16

flow field around semi-analytically identified vortex positions [21][22] The brief

explanations are as follows

a) Free surface vortex

The criteria expressed by equation 26 and 27 is applied to determine the

visible inception of a free surface vortex

≫ 1 (26)

=

gt 1 (27)

where P is the pressure drop at the vortex core and Ph is the static head at the

vortex element and f is the drag force acting at a unit length of water surface

vortex dimple based on Stokes approximation and caused by the down flow

Further f is the force acting on the bubble in the direction from the high

pressure region to the low pressure region like buoyancy

b) Submerged vortex

Equation 28 is applied to determine the visible inception of a submerged vortex

infin =

infin

infin gt 1 (28)

where is the static pressure at the vortex core infin is the atmospheric pressure

and is the critical cavitation inception pressure and nearly equal to the

saturated vapor pressure

3) Swirl angle

Swirl angle predicts the intensity of flow rotation Fig 26 shows experiment

calculation method with swirl meter and attempts of the CFD calculation method

were conducted as Fig 27 A four blade zero-pitch swirl meter which is supported

by a low friction bearing will be installed at about four suction pipe diameter (d)

downstream from the pump suction to measure swirl angle of pump approaching

flow For the swirl meter the tip to tip blade diameter is 075d and the length in

flow direction is 06 and one of the four blade is painted yellow as a reference to

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

17

count revolutions As flow swirl is usually unsteady the observation time of swirl

meter reading should be a continuous period of time

Fig 26 Typical swirl meter

(29)

(210)

(211)

Where qV flow rotational speed at swirl meter (ms)

zV average axial velocity at swirl meter (ms)

n Number of revolution per minute (rpm)

d Diameter of pipe at the swirl meter (m)

q flow rate (m3s)

zV

Vqq 1tan -=

)(60

smnd

Vtimestimes

=p

q

)(4

2

smd

qVz

times=

p

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

18

For the swirl angle calculating method with CFD the key point is to obtain the

average tangential velocity so the swirl check circles were created as illustrated in

Fig 27 The swirl check circle is located at the section of 4d height with diameter

of 025d 05d and 075d respectively

Swirl check circle

Fig 27 Schematic view of swirl angle numerical calculation

23

1 7501 501 250

acute++

=aringaringaring

N

i

N

i

NNVNVNV

Vi

q (212)

where

qV average tangential velocity (ms)

V circumferential velocity of check point in check circle

respectively (ms)

N check point total numbers in single check circle

4) Velocity

Velocity measurement at bell throat by Pitot tube was implemented by ANSIHI

method Typical Pitot tube configuration is shown in Fig 28 Eight measuring

points are defined as Fig 29 Time averaged velocity at eight points was obtained

and these eight data were again averaged to give throat sectional spatial velocity

average

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

19

Fig 28 Example of Pitot tube Fig 29 Measuring points at bell throat

224 Test criteria

There is the criteria for satisfactory performance of the sump model test and

summarized in Table 23

1) Free surface and submerged vortices

Free surface vortices entering the pump must be less severe than vortices with

coherent core Dye core vortices may be acceptable only if they occur less than 10

of the time or only for infrequent pump operating conditions

Submerged vortex which starts from the sump walls induce vibration and noise

problems Dye core submerged vortices extending into the pump bell mouth is not

allowed

2) Swirl angles

Swirl angle is used to check the flow rotation in the suction pipe The calculation

method is referred to section 223

Short-term maximum and long-term average values must be less than 5 degrees

Maximum short-term swirl angles up to 7 degrees may be acceptable only if they

occur less than 10 of the time or only for infrequent pump operating conditions

3)Velocity

Each point time-averaged velocity shall not be larger than 10 of sectional average

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

20

Standard deviation of time-varying velocity at each point was obtained and its

value shall not be larger than 10 of the time-averaged velocity at each point

1) Time-averaged

Variation[] = 100acute-

a

a

V

VV aV Averaged at section

2) Time-varying

Standard deviation= )1(

)(1i

2

-

-aring=

n

xxn

i

Variation[] = (Standard deviation X ) x 100

X Averaged at point

Table 23 Permission criteria of ANSIHI

Vortex occurence Swirl angle

Free surface vortex Submerged vortex

Observation

time Above 10 minutes Above 10 minutes

Above 10

minutes

Permission

criteria

Dye-core

Vortex(F3)

below 110 total

time

Dye-core

Vortex(S2)

below 110 total

time

Below

5degConditional

intermittent

below 7deg

225 Remedial measures for problem intakes

Usually in the preliminary test the worst flow condition is simulated And based on

the results remedial measures for problem intakes are discussed in this part

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

21

1) Overall approach flow patterns

The characteristics of flow approaching an intake structure is one of the foremost

considerations for the designer It is difficult to predict the effects of a given set of

flow conditions upstream from an intake structure Fig 210 depicts several typical

approach flow patterns of pump sump with different operating conditions

Fig 210 Examples of approach flow patterns

Generally speaking for the intake problems especially for the multiple pumps

sump there are the main measures using dividing walls placed between the pumps

as a partitioned structure trash racks with elongated bars guide piers or a number

of structural columns at the bay entrance can provide some assistance in controlling

cross-flow baffles flow distributors and screen mesh etc are applied to improve

the flow distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

22

2) Problems around pump intake

a) Free surface vortices

There are a number of factors that have an effect on the formation of surface

vortices To achieve a higher degree of certainty that objectionable surface vortices

do not form modifications can be made to intake structure to allow operation at

practical depths of submergence and the minimum submergence is referred in Table

22 the use of suction umbrellas vertical curtain walls and horizontal gratings

b) Submerged vortices

In the vicinity of intake bell there are the most complicated flow patterns and flow

makes the most changes in direction while maintaining a constant acceleration into

the pump bell to prevent local flow separation turbulence and vortex formation

Fig 211 represents a sample of various devices to inhibit the different submerged

vortex

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

23

Fig 211 Methods to reduce submerged vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

24

c) Preswirl

The approach flow distributions determine whether flow preswirl exists and a

sufficiently laterally skewed approach flow cause rotating around the intake bell

The most effective way of reducing preswirl is to establish a relatively uniform

approach flow within each pump bay by using the baffling schemes controlling

cross-flow and expanding concentrated flow discussed in the above part

To prevent non-uniform velocity distribution swirls and vortices necessary

modifications are conducted such as fore bay length baffle wall distributors

curtain wall mesh screen guide vane floor splitter wall fillet etc As the above

measures applied modification test will be conducted to check all the flow

conditions and compare with the preliminary results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

25

Chapter 3 Methodology

CFD (Computational fluid dynamics) is the technique which can design cycles to

better performance and reduce costs and time In CFD we solve the governing

equations of given physics (may be differential form or integral form) using some

numerical techniques like Finite Difference Method (FDM) Finite Element Method

(FEM) or Finite Volume Method (FVM) Although CFD plays a great part in the

design of turbomachinery CFD is not a replacement for experimental or analytical

approach In this study the numerical analysis was performed by the commercial

code ANSYS CFX 130 All aspects of the CFD study including modelling

meshing solving and post-processing of the results

31 Numerical modeling

ANSYS CFX solves the unsteady Navier-Stokes equations in their conservation

form The instantaneous equation of mass (continuity) in the stationary frame is

expressed as equation

ρ

δ + nabla ∙ (ρU) = 0 (31)

And the instantaneous equation for momentum is expressed as shown in equation

ρ

δ + nabla ∙ (ρU⨂U) = minusnablap + nabla ∙ τ + S (32)

Where τ is the stress tensor ( including both normal and shear components of the

stress)

These instantaneous equations are averaged for turbulent flows leading to additional

terms that need to be solved While the Navier-Stokes equations describe both

laminar and turbulent flows without addition terms realistic flows involve length

scales much smaller than the smallest finite volume mesh A Direct Numerical

Simulation of these flows would require significantly more computing power than

what is available now or in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

26

Therefore much research has been done to predict the effects of turbulence by

using turbulence models These models account for the effects of turbulence

without the use of a very fine mesh or direct numerical simulation

These turbulence models modify the transport equations by adding averaged and

fluctuating components The transport equations are changed to equations (33) and

(34)

ρ

δ + nabla ∙ (ρU) = 0 (33)

ρ

+ nabla ∙ ρ U ⨂U = minusnablap + nabla ∙ τ minus ρu⨂u + S (34)

The mass equation is not changed but the momentum equation contains extra terms

which are the Reynolds stresses ρ u⨂u and the Reynolds flux ρ u ϕ These

Reynolds stresses need to be modeled by additional equations to obtain closure

Obtaining closure implies that there are a sufficient number of equations to solve for

all the unknowns including the Reynolds stresses and Reynolds fluxes

Various turbulence models provide various ways to obtain closure In this

investigation Shear Stress Transport (SST) model and k-ε model were utilized The

advantage of SST model is that combines the advantages of other turbulence models

(the k-ε Wilcox k-ω) The 3 turbulence models will be discussed briefly

32 Turbulence models

Turbulence models are necessary because we cannot afford big enough computers

to directly capture every scale of motion There are three modelling frameworks

Direct Numerical Simulation (DNS) Reynolds-Averaged Navier-Stokes

Equations(RANS) Large Eddy Simulation(LES)

As RANS method is applied in most productionseveral RANS turbulence models

differ in their usage of wall functions the number of additional variables solved for

and what these variables represent

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

27

Two equation turbulence models

Two-equation turbulence models are widely used in CFD as they give a good

compromise between computational power needed and accuracy The term lsquotwo

equationrsquo refers to the fact that these models solve for the velocity and length scales

using separate transport equations The turbulence velocity scale is obtained by

solving the transport equation The turbulent length scale is estimated from two

properties of the turbulence field namely the turbulent kinetic energy and the

dissipation rate The dissipation rate of the kinetic energy is obtained from its

transport equation The most widely used are k-ε and k-ω two equation models In

the next section the k-ε Wilcox k-ω BSL k-ω and SST models will be briefly

discussed

321 k-ε turbulence model

The k-epsilon model solves for two variables k the turbulent kinetic energy and

epsilon the rate of dissipation of kinetic energy Wall functions are used in this

model so the flow in the buffer region is not simulated The k-epsilon model is very

popular for industrial applications due to its good convergence rate and relatively

low memory requirements It does not very accurately compute flow fields that

exhibit adverse pressure gradients strong curvature to the flow or jet flow It does

perform well for external flow problems around complex geometries

The k-ε model introduces k (m2s2) as the turbulence kinetic energy and ε (m2s3) as

the turbulence eddy dissipation The continuity equation remains the same

ρ

+ nabla ∙ (ρ U) = 0 (35)

The momentum equation changes as shown by equation

δρ

δ + nabla ∙ (ρU⨂U) = minusnablapprime + nabla ∙ μ

nablaU + (nablaU) + S (36)

Where SM is the sum of body forces μeff is the effective viscosity accounting for

turbulence and prsquo is the modified pressure The k-ε model uses the concept of eddy

viscosity giving the equation for effective viscosity as shown by equation

μ

= μ + μ (37)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

28

μt is the turbulence viscosity is linked to the turbulence kinetic energy and

dissipation by the equation

μ = Cμρ

ε (38)

Where Cμ is a constant

The values for k and ε come from the differential transport equations for the

turbulence kinetic energy and the turbulence dissipation rate

The turbulence kinetic energy equation is given as equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus ρε (39)

The turbulence dissipation rate is given by equation

(ρε)

+ nabla ∙ (ρUε) = nabla ∙ μ +

μ

σε nablaε +

ε

(Cε (P + Pε ) minus Cε ρε) (310)

Where Cε1 Cε2 σk σε are constants

Pk is the turbulence production due to viscous forces and is modeled by the

equation

P = μ nablaU ∙ nablaU + nablaU minus

nabla ∙ U 3μ

nabla ∙ U + ρk (311)

If a buoyancy term is added to the previous equation if the full buoyancy model is

used

The buoyancy term Pkb is modeled as

P = minusμ

ρσρg ∙ nablaρ (312)

322 Wilcox k-ω turbulence model

The k-omega model is similar to k-epsilon instead however it solves for omega mdash

the specific rate of dissipation of kinetic energy It also uses wall functions and

therefore has comparable memory requirements It has more difficulty converging

and is quite sensitive to the initial guess at the solution Hence the k-epsilon model

is often used first to find an initial condition for solving the k-omega model The k-

omega model is useful in many cases where the k-epsilon model is not accurate

such as internal flows flows that exhibit strong curvature separated flows and jets

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

29

This model has an advantage over the k-ε model where it does not involve complex

linear damping functions for near wall calculations at low Reynolds numbers

The k-ω model assumes that the turbulence viscosity is related to the turbulence

kinetic energy k and the turbulent frequency ω by the equation

μ = ρ

ω (313)

The transport equation for k is given by the equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus βprimeρ k ω (314)

The transport equation for ω is shown as equation

(ρω)

δ + nabla ∙ (ρ U ω) = nabla ∙ μ +

μ

μω

nablaω + αω

P + Pω minus β ρ ω (315)

The production rate of turbulence (Pk) is calculated as shown previously in the k-ε

section

The recommended values for the constants in the above equations are

βprime = 009

α = 59

β = 0075

σ = 2

σω = 2

The Reynolds stress tensor ρ u⨂u is calculated by

minusρ u⨂u = μ nablaU + (nablaU) minus

part ρk + μ

nabla ∙ U (316)

323 Shear stress transport model

The disadvantage of the Wilcox model is the strong sensitivity to free-stream

conditions Therefore a blending of the k-ω model near the surface and the k-ε in

the outer region was made by Menter [23] which resulted in the formulation of the

BSL k-ω turbulence model It consists of a transformation of the k-ε model to a k-ω

formulation and subsequently adding the resulting equations The Wilcox model is

multiplied by a blending function F1 and the transformed k-ε by another function 1-

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

30

F1 F1 is a function of wall distance (being the value of one near the surface and zero

outside the boundary layer) Outside the boundary and on the edge of the boundary

layer the standard k-ε model is used [24]

SST model is a combination of the k-epsilon in the free stream and the k-omega

models near the walls It does not use wall functions and tends to be most accurate

when solving the flow near the wall The SST model does not always converge to

the solution quickly so the k-epsilon or k-omega models are often solved first to

give good initial conditions

The k-ω based Shear Stress Transport (SST) model of Menter was applied for

turbulence treatment The transport equations for the SST model are expressed

below where the turbulent kinetic energy lsquokrsquo and turbulent frequency or dissipation

per unit turbulent kinetic energy are computed by using the following relations

For Turbulence Kinetic Energy

kkPukt

kTkk Ntilde+Ntilde+-=Ntilde+

para

para)[()( nsnwb (317)

where Pk is the production limiter

For Specific Dissipation Rate

ww

swnsnbwaww

ww NtildeNtilde-+Ntilde+Ntilde+-=Ntilde+para

para

1)1(2])[()( 21

22 kFSut

T (318)

The first blending function F1 is calculated from

iumlthorn

iumlyacute

uuml

iumlicirc

iumliacute

igrave

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

uacuteuacuteucirc

ugrave

ecircecirceuml

eacute

dividedivideoslash

oumlccedilccedilegrave

aelig=

4

22

21

4500maxmintanh

yCD

k

yy

kF

kw

wrs

w

n

wb (319)

y is the distance to the nearest wall and v is the kinematic viscosity

divideoslash

oumlccedilegrave

aeligNtildeNtilde= -10

2 101

2max ww

rsww kCDk (320)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

31

And Kinematic eddy viscosity

)max( 21

1

SF

kT

wa

an = (321)

where F2 is a blending function which restricts the limiter to the wall boundary and

S is the invariant measure of the strain rate

uacuteuacute

ucirc

ugrave

ecircecirc

euml

eacute

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

dividedivideoslash

oumlccedilccedilegrave

aelig=

2

22

5002maxtanh

w

n

wb yy

kF (322)

33 Brief introduction of simulation process

All aspects of CFD study including modelling meshing solving and post-

processing of the results are briefly introduced in this chapter

331 Pre-processing technique

The speed and accuracy of simulation results depending fundamentally on meshing

and the ability to prepare model geometry correctly

The steps in pre-processing are explained as follows

1) Domain Definitions

The domain definition depends mainly on the physics involved in the problem and

objective of the problem

a) Geometry creation or geometry import

b) Geometry cleanup

i) Removing the parts not necessary for simulation under consideration

ii) Closing the gaps in the geometry

iii) Removing small surfaces curves to make meshing process simple

2)Grid generation

To solve the governing equations the numerical methods like Finite Difference

Finite Volume or Finite Element Method to divide the domain into small parts and

solve the equations on one part at a time are applied

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 21: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

9

FD=Vb(gD)05 (21)

where

FD Froude number (dimensionless)

Vb velocity at suction inlet=FlowArea based on D

D Outside diameter of bell or pipe inlet

g gravitational acceleration

There is the design sequence summary for rectangular intake structures

middot Consider the flow patterns and boundary geometry of the body of liquid

from which the pump station is to receive flow Compare with the approach

flow condition and determine if a hydraulic physical model study is

required

middot Determine the number and size of pumps required to satisfy the range of

operating conditions likely to be encountered

middot Identify pump inlet bell diameter Refer to the section 211

middot Determine the bell-floor clearance 05D is a good preliminary design value

middot Determine the required bell submergence using the equation 21

middot Determine the minimum allowable liquid depth in the intake structure from

the sum of the floor clearance and the required bell submergence

middot Check bottom elevation near the entrance to the structure and determine if it

is necessary to slope the floor upstream of the bay entrance If the resulting

depth at the entrance to the intake structure is shallow then check to ensure

that gravity-driven flow is not restricted by the entrance condition

middot Check the pump bay velocity for the maximum single-pump flow and

minimum liquid depth with the bay width set to 2D If bay velocity exceeds

05 ms then increase the bay width to reduce the maximum velocity below

05 ms

middot If it is necessary to increase the pump bay width to greater than 2D then

decrease bay width in the vicinity of the pumps according to Fig 23

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

10

middot Compare cross-flow velocity (at maximum system flow) to average pump

bay velocity If cross-flow value exceeds 50 of the bay velocity a

physical hydraulic model study is necessary

middot Determine the length of the structure and dividing walls giving

consideration to minimum allowable distances to a sloping floor screening

equipment and length of dividing walls if dual flow traveling screens or

drum screens are to be used a physical hydraulic model study is required

middot If the final selected pump bell diameter and inlet velocity is within the

range given in section 211 then the sump dimensions (developed based on

the inlet bell design diameter) need not be changed and will comply with

these standards

22 Model test

The approach flow pattern in a pump intake facility is difficult to predict by

traditional mathematics or empirical formulae For large structures or those that

differ significantly from proven designs a model study is the only means to ensure

success A model study is used to identify adverse hydraulic conditions and

derive remedial measures for approach flow patterns generated by structures

upstream of the pump impeller Current model studies are not intended to

investigate flow patterns induced by the pump itself or the flow patterns within the

pump The objective of a model study is to ensure that the pump intake structure

generates favorable flow conditions at the inlet to the pump Evaluation for the

requirement of model test if

a Sump or piping geometry that deviates from this design standard

b Non-uniform or non-symmetric approach flow to the pump sump

exist

c The pumps have flows greater than 2520 ls per pump or the total

station flow with all pumps running would be greater than 6310 ls

d Situations that need ten times cost than the model study

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

11

Adverse hydraulic conditions that can affect pump performance include free and

submerged vortices swirl approaching the pump impeller flow separation at the

pump bell and a non-uniform axial velocity distribution at the suction

The purpose of model test is to examine these adverse hydraulic conditions and to

investigate measure against them

221 The scale effects

One possible difficulty in scale modeling is that it is impossible to reduce all

pertinent forces by the same factor With Froude-scale models the inertial and

gravitational forces are reduced similarly but the viscous and surface tension forces

cannot be simultaneously reduced The extra influence of these forces is known as

a ldquoscale effectrdquo A number of investigations have indicated that the scale effects for

the model are avoided if the Reynolds number based on the inlet flow and

submergence or intake diameter exceeds approximately 3x 104 and the Weber

number exceeds 150 [19]

In most engineering applications involving closed-conduit and open-channel flow

the Reynolds number limits are far exceeded and the flows are fully turbulent

In a sump model Froude number (Fr) and Reynolds number (Re) are the most

important non-dimensional parameters Some reduction in Re should be

introduced to the sump model However flow patterns are generally very similar

at high Re

No specific geometric scale rate is recommended but reasonable large

geometric scale to minimize viscous and surface tension scale effects and reproduce

the flow pattern

Re= uDν (22)

We= u2ρDσ (23)

where

u average axial velocity (ms)

ν kinematic viscosity of the liquid (m2s)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

12

ρ liquid density (kgm3)

σ surface tension of liquidair interface

According to the ANSIHI 98 the viscous scale effects on vortex may be negligible

if the resulting dimensionless numbers must meet these values Re>6times104 and

We>240

To allow practical visual observations of flow patterns accurate measurements of

swirl and velocity distribution and sufficient dimensional control the model scale

shall yield a bay width of at least 300mm a minimum liquid depth of at least

150mm and a pump throat or suction diameter of at least 80mm

222 Principle of similarity

In the similitude analysis the geometric and flow similarity requires the model and

the flow to be identical as the real model and flow This is to ensure the result

obtained from the model study is well presented in order to predict full scale

behavior Model involving a free surface are operated using Froude similarity since

the flow process is controlled by gravity and inertial forces

For similarity of the flow patterns the Froude number shall be equal in model and

prototype

Fr =

=

(24)

and as the scale ratio λ was selected in section 221 the following equation is

obtained

V = V times (

) = V times λ or Q = Q times (

) = Q times λ (25)

where Fr is the Froude number λ is the scale ratio of model dimensions to prototype

The test of 15 times Froude scaled flow should be conducted to ascertain the

potential scale affect on free surface vortices

Models of closed conduit piping system leading to a pump suction are not operated

based on Froude similitude However flow patterns are generally very similar at

high Re A minimum value of 1times105 for the Re at the pump suction is

recommended to ensure the flow patterns are correctly scaled

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

13

223 Preliminary model test

The initial design shall be tested first to identify any hydraulic problems

1) Vortex Formation Mechanisms

There are various investigations that provide insight into the fundamental processes

leading to the development of vortices both in experiment and numerical simulation

Shin et al [20] demonstrated that two basic mechanisms lead to inlet vortex

formation The first mechanism involves the development of an inlet vortex due to

the amplification of ambient vorticity in the approach flow as vortex lines

are convicted into the inlet The second mechanism involves the development of a

trailing vortex in the vicinity of the intake as a result of the variation in

circulation along the inlet For this second case a vortex can develop in a flow that

is irrotational upstream and the vortex development therefore does not depend

on the presence of ambient vorticity Shin et al investigation on kinematic

parameters indicate that the strength of an inlet-vortex or trailing vortex system

increases with decreasing distance from the surface However for an inlet in

an upstream irrotational flow two counter rotating vortices can still trail from

the rear of the inlet

2) Classification of vortex type

As illustrated in Fig 24 vortices in the vicinity of pump intakes may be adjacent to

the channel bottom or a channel wall (ie submerged vortices) or they may appear

adjacent to the free surface (ie free surface vortex)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

14

(a) Schematic diagram of vortices in a sump

(b) Experimental results of different vortex

Fig 24 Typical vortex in a sump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

15

For the model visual observation dye was injected around the intake structure near

the surface and submergence as Fig 24 (b) shows And the model test results of

vortex was classified as Fig 25

(a) Free Surface Vortices

(b) Submerged Vortices

Fig 25 Vortex classifications

There are a number of researches focused on the numerical theory to detect the

inception of a visible vortex As the diameters of vortices are usually much smaller

compared with the computing grid size to compensate for the lack of vortex

resolution in numerical simulation a stretching vortex model is applied to the local

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

16

flow field around semi-analytically identified vortex positions [21][22] The brief

explanations are as follows

a) Free surface vortex

The criteria expressed by equation 26 and 27 is applied to determine the

visible inception of a free surface vortex

≫ 1 (26)

=

gt 1 (27)

where P is the pressure drop at the vortex core and Ph is the static head at the

vortex element and f is the drag force acting at a unit length of water surface

vortex dimple based on Stokes approximation and caused by the down flow

Further f is the force acting on the bubble in the direction from the high

pressure region to the low pressure region like buoyancy

b) Submerged vortex

Equation 28 is applied to determine the visible inception of a submerged vortex

infin =

infin

infin gt 1 (28)

where is the static pressure at the vortex core infin is the atmospheric pressure

and is the critical cavitation inception pressure and nearly equal to the

saturated vapor pressure

3) Swirl angle

Swirl angle predicts the intensity of flow rotation Fig 26 shows experiment

calculation method with swirl meter and attempts of the CFD calculation method

were conducted as Fig 27 A four blade zero-pitch swirl meter which is supported

by a low friction bearing will be installed at about four suction pipe diameter (d)

downstream from the pump suction to measure swirl angle of pump approaching

flow For the swirl meter the tip to tip blade diameter is 075d and the length in

flow direction is 06 and one of the four blade is painted yellow as a reference to

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

17

count revolutions As flow swirl is usually unsteady the observation time of swirl

meter reading should be a continuous period of time

Fig 26 Typical swirl meter

(29)

(210)

(211)

Where qV flow rotational speed at swirl meter (ms)

zV average axial velocity at swirl meter (ms)

n Number of revolution per minute (rpm)

d Diameter of pipe at the swirl meter (m)

q flow rate (m3s)

zV

Vqq 1tan -=

)(60

smnd

Vtimestimes

=p

q

)(4

2

smd

qVz

times=

p

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

18

For the swirl angle calculating method with CFD the key point is to obtain the

average tangential velocity so the swirl check circles were created as illustrated in

Fig 27 The swirl check circle is located at the section of 4d height with diameter

of 025d 05d and 075d respectively

Swirl check circle

Fig 27 Schematic view of swirl angle numerical calculation

23

1 7501 501 250

acute++

=aringaringaring

N

i

N

i

NNVNVNV

Vi

q (212)

where

qV average tangential velocity (ms)

V circumferential velocity of check point in check circle

respectively (ms)

N check point total numbers in single check circle

4) Velocity

Velocity measurement at bell throat by Pitot tube was implemented by ANSIHI

method Typical Pitot tube configuration is shown in Fig 28 Eight measuring

points are defined as Fig 29 Time averaged velocity at eight points was obtained

and these eight data were again averaged to give throat sectional spatial velocity

average

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

19

Fig 28 Example of Pitot tube Fig 29 Measuring points at bell throat

224 Test criteria

There is the criteria for satisfactory performance of the sump model test and

summarized in Table 23

1) Free surface and submerged vortices

Free surface vortices entering the pump must be less severe than vortices with

coherent core Dye core vortices may be acceptable only if they occur less than 10

of the time or only for infrequent pump operating conditions

Submerged vortex which starts from the sump walls induce vibration and noise

problems Dye core submerged vortices extending into the pump bell mouth is not

allowed

2) Swirl angles

Swirl angle is used to check the flow rotation in the suction pipe The calculation

method is referred to section 223

Short-term maximum and long-term average values must be less than 5 degrees

Maximum short-term swirl angles up to 7 degrees may be acceptable only if they

occur less than 10 of the time or only for infrequent pump operating conditions

3)Velocity

Each point time-averaged velocity shall not be larger than 10 of sectional average

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

20

Standard deviation of time-varying velocity at each point was obtained and its

value shall not be larger than 10 of the time-averaged velocity at each point

1) Time-averaged

Variation[] = 100acute-

a

a

V

VV aV Averaged at section

2) Time-varying

Standard deviation= )1(

)(1i

2

-

-aring=

n

xxn

i

Variation[] = (Standard deviation X ) x 100

X Averaged at point

Table 23 Permission criteria of ANSIHI

Vortex occurence Swirl angle

Free surface vortex Submerged vortex

Observation

time Above 10 minutes Above 10 minutes

Above 10

minutes

Permission

criteria

Dye-core

Vortex(F3)

below 110 total

time

Dye-core

Vortex(S2)

below 110 total

time

Below

5degConditional

intermittent

below 7deg

225 Remedial measures for problem intakes

Usually in the preliminary test the worst flow condition is simulated And based on

the results remedial measures for problem intakes are discussed in this part

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

21

1) Overall approach flow patterns

The characteristics of flow approaching an intake structure is one of the foremost

considerations for the designer It is difficult to predict the effects of a given set of

flow conditions upstream from an intake structure Fig 210 depicts several typical

approach flow patterns of pump sump with different operating conditions

Fig 210 Examples of approach flow patterns

Generally speaking for the intake problems especially for the multiple pumps

sump there are the main measures using dividing walls placed between the pumps

as a partitioned structure trash racks with elongated bars guide piers or a number

of structural columns at the bay entrance can provide some assistance in controlling

cross-flow baffles flow distributors and screen mesh etc are applied to improve

the flow distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

22

2) Problems around pump intake

a) Free surface vortices

There are a number of factors that have an effect on the formation of surface

vortices To achieve a higher degree of certainty that objectionable surface vortices

do not form modifications can be made to intake structure to allow operation at

practical depths of submergence and the minimum submergence is referred in Table

22 the use of suction umbrellas vertical curtain walls and horizontal gratings

b) Submerged vortices

In the vicinity of intake bell there are the most complicated flow patterns and flow

makes the most changes in direction while maintaining a constant acceleration into

the pump bell to prevent local flow separation turbulence and vortex formation

Fig 211 represents a sample of various devices to inhibit the different submerged

vortex

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

23

Fig 211 Methods to reduce submerged vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

24

c) Preswirl

The approach flow distributions determine whether flow preswirl exists and a

sufficiently laterally skewed approach flow cause rotating around the intake bell

The most effective way of reducing preswirl is to establish a relatively uniform

approach flow within each pump bay by using the baffling schemes controlling

cross-flow and expanding concentrated flow discussed in the above part

To prevent non-uniform velocity distribution swirls and vortices necessary

modifications are conducted such as fore bay length baffle wall distributors

curtain wall mesh screen guide vane floor splitter wall fillet etc As the above

measures applied modification test will be conducted to check all the flow

conditions and compare with the preliminary results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

25

Chapter 3 Methodology

CFD (Computational fluid dynamics) is the technique which can design cycles to

better performance and reduce costs and time In CFD we solve the governing

equations of given physics (may be differential form or integral form) using some

numerical techniques like Finite Difference Method (FDM) Finite Element Method

(FEM) or Finite Volume Method (FVM) Although CFD plays a great part in the

design of turbomachinery CFD is not a replacement for experimental or analytical

approach In this study the numerical analysis was performed by the commercial

code ANSYS CFX 130 All aspects of the CFD study including modelling

meshing solving and post-processing of the results

31 Numerical modeling

ANSYS CFX solves the unsteady Navier-Stokes equations in their conservation

form The instantaneous equation of mass (continuity) in the stationary frame is

expressed as equation

ρ

δ + nabla ∙ (ρU) = 0 (31)

And the instantaneous equation for momentum is expressed as shown in equation

ρ

δ + nabla ∙ (ρU⨂U) = minusnablap + nabla ∙ τ + S (32)

Where τ is the stress tensor ( including both normal and shear components of the

stress)

These instantaneous equations are averaged for turbulent flows leading to additional

terms that need to be solved While the Navier-Stokes equations describe both

laminar and turbulent flows without addition terms realistic flows involve length

scales much smaller than the smallest finite volume mesh A Direct Numerical

Simulation of these flows would require significantly more computing power than

what is available now or in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

26

Therefore much research has been done to predict the effects of turbulence by

using turbulence models These models account for the effects of turbulence

without the use of a very fine mesh or direct numerical simulation

These turbulence models modify the transport equations by adding averaged and

fluctuating components The transport equations are changed to equations (33) and

(34)

ρ

δ + nabla ∙ (ρU) = 0 (33)

ρ

+ nabla ∙ ρ U ⨂U = minusnablap + nabla ∙ τ minus ρu⨂u + S (34)

The mass equation is not changed but the momentum equation contains extra terms

which are the Reynolds stresses ρ u⨂u and the Reynolds flux ρ u ϕ These

Reynolds stresses need to be modeled by additional equations to obtain closure

Obtaining closure implies that there are a sufficient number of equations to solve for

all the unknowns including the Reynolds stresses and Reynolds fluxes

Various turbulence models provide various ways to obtain closure In this

investigation Shear Stress Transport (SST) model and k-ε model were utilized The

advantage of SST model is that combines the advantages of other turbulence models

(the k-ε Wilcox k-ω) The 3 turbulence models will be discussed briefly

32 Turbulence models

Turbulence models are necessary because we cannot afford big enough computers

to directly capture every scale of motion There are three modelling frameworks

Direct Numerical Simulation (DNS) Reynolds-Averaged Navier-Stokes

Equations(RANS) Large Eddy Simulation(LES)

As RANS method is applied in most productionseveral RANS turbulence models

differ in their usage of wall functions the number of additional variables solved for

and what these variables represent

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

27

Two equation turbulence models

Two-equation turbulence models are widely used in CFD as they give a good

compromise between computational power needed and accuracy The term lsquotwo

equationrsquo refers to the fact that these models solve for the velocity and length scales

using separate transport equations The turbulence velocity scale is obtained by

solving the transport equation The turbulent length scale is estimated from two

properties of the turbulence field namely the turbulent kinetic energy and the

dissipation rate The dissipation rate of the kinetic energy is obtained from its

transport equation The most widely used are k-ε and k-ω two equation models In

the next section the k-ε Wilcox k-ω BSL k-ω and SST models will be briefly

discussed

321 k-ε turbulence model

The k-epsilon model solves for two variables k the turbulent kinetic energy and

epsilon the rate of dissipation of kinetic energy Wall functions are used in this

model so the flow in the buffer region is not simulated The k-epsilon model is very

popular for industrial applications due to its good convergence rate and relatively

low memory requirements It does not very accurately compute flow fields that

exhibit adverse pressure gradients strong curvature to the flow or jet flow It does

perform well for external flow problems around complex geometries

The k-ε model introduces k (m2s2) as the turbulence kinetic energy and ε (m2s3) as

the turbulence eddy dissipation The continuity equation remains the same

ρ

+ nabla ∙ (ρ U) = 0 (35)

The momentum equation changes as shown by equation

δρ

δ + nabla ∙ (ρU⨂U) = minusnablapprime + nabla ∙ μ

nablaU + (nablaU) + S (36)

Where SM is the sum of body forces μeff is the effective viscosity accounting for

turbulence and prsquo is the modified pressure The k-ε model uses the concept of eddy

viscosity giving the equation for effective viscosity as shown by equation

μ

= μ + μ (37)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

28

μt is the turbulence viscosity is linked to the turbulence kinetic energy and

dissipation by the equation

μ = Cμρ

ε (38)

Where Cμ is a constant

The values for k and ε come from the differential transport equations for the

turbulence kinetic energy and the turbulence dissipation rate

The turbulence kinetic energy equation is given as equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus ρε (39)

The turbulence dissipation rate is given by equation

(ρε)

+ nabla ∙ (ρUε) = nabla ∙ μ +

μ

σε nablaε +

ε

(Cε (P + Pε ) minus Cε ρε) (310)

Where Cε1 Cε2 σk σε are constants

Pk is the turbulence production due to viscous forces and is modeled by the

equation

P = μ nablaU ∙ nablaU + nablaU minus

nabla ∙ U 3μ

nabla ∙ U + ρk (311)

If a buoyancy term is added to the previous equation if the full buoyancy model is

used

The buoyancy term Pkb is modeled as

P = minusμ

ρσρg ∙ nablaρ (312)

322 Wilcox k-ω turbulence model

The k-omega model is similar to k-epsilon instead however it solves for omega mdash

the specific rate of dissipation of kinetic energy It also uses wall functions and

therefore has comparable memory requirements It has more difficulty converging

and is quite sensitive to the initial guess at the solution Hence the k-epsilon model

is often used first to find an initial condition for solving the k-omega model The k-

omega model is useful in many cases where the k-epsilon model is not accurate

such as internal flows flows that exhibit strong curvature separated flows and jets

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

29

This model has an advantage over the k-ε model where it does not involve complex

linear damping functions for near wall calculations at low Reynolds numbers

The k-ω model assumes that the turbulence viscosity is related to the turbulence

kinetic energy k and the turbulent frequency ω by the equation

μ = ρ

ω (313)

The transport equation for k is given by the equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus βprimeρ k ω (314)

The transport equation for ω is shown as equation

(ρω)

δ + nabla ∙ (ρ U ω) = nabla ∙ μ +

μ

μω

nablaω + αω

P + Pω minus β ρ ω (315)

The production rate of turbulence (Pk) is calculated as shown previously in the k-ε

section

The recommended values for the constants in the above equations are

βprime = 009

α = 59

β = 0075

σ = 2

σω = 2

The Reynolds stress tensor ρ u⨂u is calculated by

minusρ u⨂u = μ nablaU + (nablaU) minus

part ρk + μ

nabla ∙ U (316)

323 Shear stress transport model

The disadvantage of the Wilcox model is the strong sensitivity to free-stream

conditions Therefore a blending of the k-ω model near the surface and the k-ε in

the outer region was made by Menter [23] which resulted in the formulation of the

BSL k-ω turbulence model It consists of a transformation of the k-ε model to a k-ω

formulation and subsequently adding the resulting equations The Wilcox model is

multiplied by a blending function F1 and the transformed k-ε by another function 1-

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

30

F1 F1 is a function of wall distance (being the value of one near the surface and zero

outside the boundary layer) Outside the boundary and on the edge of the boundary

layer the standard k-ε model is used [24]

SST model is a combination of the k-epsilon in the free stream and the k-omega

models near the walls It does not use wall functions and tends to be most accurate

when solving the flow near the wall The SST model does not always converge to

the solution quickly so the k-epsilon or k-omega models are often solved first to

give good initial conditions

The k-ω based Shear Stress Transport (SST) model of Menter was applied for

turbulence treatment The transport equations for the SST model are expressed

below where the turbulent kinetic energy lsquokrsquo and turbulent frequency or dissipation

per unit turbulent kinetic energy are computed by using the following relations

For Turbulence Kinetic Energy

kkPukt

kTkk Ntilde+Ntilde+-=Ntilde+

para

para)[()( nsnwb (317)

where Pk is the production limiter

For Specific Dissipation Rate

ww

swnsnbwaww

ww NtildeNtilde-+Ntilde+Ntilde+-=Ntilde+para

para

1)1(2])[()( 21

22 kFSut

T (318)

The first blending function F1 is calculated from

iumlthorn

iumlyacute

uuml

iumlicirc

iumliacute

igrave

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

uacuteuacuteucirc

ugrave

ecircecirceuml

eacute

dividedivideoslash

oumlccedilccedilegrave

aelig=

4

22

21

4500maxmintanh

yCD

k

yy

kF

kw

wrs

w

n

wb (319)

y is the distance to the nearest wall and v is the kinematic viscosity

divideoslash

oumlccedilegrave

aeligNtildeNtilde= -10

2 101

2max ww

rsww kCDk (320)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

31

And Kinematic eddy viscosity

)max( 21

1

SF

kT

wa

an = (321)

where F2 is a blending function which restricts the limiter to the wall boundary and

S is the invariant measure of the strain rate

uacuteuacute

ucirc

ugrave

ecircecirc

euml

eacute

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

dividedivideoslash

oumlccedilccedilegrave

aelig=

2

22

5002maxtanh

w

n

wb yy

kF (322)

33 Brief introduction of simulation process

All aspects of CFD study including modelling meshing solving and post-

processing of the results are briefly introduced in this chapter

331 Pre-processing technique

The speed and accuracy of simulation results depending fundamentally on meshing

and the ability to prepare model geometry correctly

The steps in pre-processing are explained as follows

1) Domain Definitions

The domain definition depends mainly on the physics involved in the problem and

objective of the problem

a) Geometry creation or geometry import

b) Geometry cleanup

i) Removing the parts not necessary for simulation under consideration

ii) Closing the gaps in the geometry

iii) Removing small surfaces curves to make meshing process simple

2)Grid generation

To solve the governing equations the numerical methods like Finite Difference

Finite Volume or Finite Element Method to divide the domain into small parts and

solve the equations on one part at a time are applied

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 22: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

10

middot Compare cross-flow velocity (at maximum system flow) to average pump

bay velocity If cross-flow value exceeds 50 of the bay velocity a

physical hydraulic model study is necessary

middot Determine the length of the structure and dividing walls giving

consideration to minimum allowable distances to a sloping floor screening

equipment and length of dividing walls if dual flow traveling screens or

drum screens are to be used a physical hydraulic model study is required

middot If the final selected pump bell diameter and inlet velocity is within the

range given in section 211 then the sump dimensions (developed based on

the inlet bell design diameter) need not be changed and will comply with

these standards

22 Model test

The approach flow pattern in a pump intake facility is difficult to predict by

traditional mathematics or empirical formulae For large structures or those that

differ significantly from proven designs a model study is the only means to ensure

success A model study is used to identify adverse hydraulic conditions and

derive remedial measures for approach flow patterns generated by structures

upstream of the pump impeller Current model studies are not intended to

investigate flow patterns induced by the pump itself or the flow patterns within the

pump The objective of a model study is to ensure that the pump intake structure

generates favorable flow conditions at the inlet to the pump Evaluation for the

requirement of model test if

a Sump or piping geometry that deviates from this design standard

b Non-uniform or non-symmetric approach flow to the pump sump

exist

c The pumps have flows greater than 2520 ls per pump or the total

station flow with all pumps running would be greater than 6310 ls

d Situations that need ten times cost than the model study

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

11

Adverse hydraulic conditions that can affect pump performance include free and

submerged vortices swirl approaching the pump impeller flow separation at the

pump bell and a non-uniform axial velocity distribution at the suction

The purpose of model test is to examine these adverse hydraulic conditions and to

investigate measure against them

221 The scale effects

One possible difficulty in scale modeling is that it is impossible to reduce all

pertinent forces by the same factor With Froude-scale models the inertial and

gravitational forces are reduced similarly but the viscous and surface tension forces

cannot be simultaneously reduced The extra influence of these forces is known as

a ldquoscale effectrdquo A number of investigations have indicated that the scale effects for

the model are avoided if the Reynolds number based on the inlet flow and

submergence or intake diameter exceeds approximately 3x 104 and the Weber

number exceeds 150 [19]

In most engineering applications involving closed-conduit and open-channel flow

the Reynolds number limits are far exceeded and the flows are fully turbulent

In a sump model Froude number (Fr) and Reynolds number (Re) are the most

important non-dimensional parameters Some reduction in Re should be

introduced to the sump model However flow patterns are generally very similar

at high Re

No specific geometric scale rate is recommended but reasonable large

geometric scale to minimize viscous and surface tension scale effects and reproduce

the flow pattern

Re= uDν (22)

We= u2ρDσ (23)

where

u average axial velocity (ms)

ν kinematic viscosity of the liquid (m2s)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

12

ρ liquid density (kgm3)

σ surface tension of liquidair interface

According to the ANSIHI 98 the viscous scale effects on vortex may be negligible

if the resulting dimensionless numbers must meet these values Re>6times104 and

We>240

To allow practical visual observations of flow patterns accurate measurements of

swirl and velocity distribution and sufficient dimensional control the model scale

shall yield a bay width of at least 300mm a minimum liquid depth of at least

150mm and a pump throat or suction diameter of at least 80mm

222 Principle of similarity

In the similitude analysis the geometric and flow similarity requires the model and

the flow to be identical as the real model and flow This is to ensure the result

obtained from the model study is well presented in order to predict full scale

behavior Model involving a free surface are operated using Froude similarity since

the flow process is controlled by gravity and inertial forces

For similarity of the flow patterns the Froude number shall be equal in model and

prototype

Fr =

=

(24)

and as the scale ratio λ was selected in section 221 the following equation is

obtained

V = V times (

) = V times λ or Q = Q times (

) = Q times λ (25)

where Fr is the Froude number λ is the scale ratio of model dimensions to prototype

The test of 15 times Froude scaled flow should be conducted to ascertain the

potential scale affect on free surface vortices

Models of closed conduit piping system leading to a pump suction are not operated

based on Froude similitude However flow patterns are generally very similar at

high Re A minimum value of 1times105 for the Re at the pump suction is

recommended to ensure the flow patterns are correctly scaled

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

13

223 Preliminary model test

The initial design shall be tested first to identify any hydraulic problems

1) Vortex Formation Mechanisms

There are various investigations that provide insight into the fundamental processes

leading to the development of vortices both in experiment and numerical simulation

Shin et al [20] demonstrated that two basic mechanisms lead to inlet vortex

formation The first mechanism involves the development of an inlet vortex due to

the amplification of ambient vorticity in the approach flow as vortex lines

are convicted into the inlet The second mechanism involves the development of a

trailing vortex in the vicinity of the intake as a result of the variation in

circulation along the inlet For this second case a vortex can develop in a flow that

is irrotational upstream and the vortex development therefore does not depend

on the presence of ambient vorticity Shin et al investigation on kinematic

parameters indicate that the strength of an inlet-vortex or trailing vortex system

increases with decreasing distance from the surface However for an inlet in

an upstream irrotational flow two counter rotating vortices can still trail from

the rear of the inlet

2) Classification of vortex type

As illustrated in Fig 24 vortices in the vicinity of pump intakes may be adjacent to

the channel bottom or a channel wall (ie submerged vortices) or they may appear

adjacent to the free surface (ie free surface vortex)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

14

(a) Schematic diagram of vortices in a sump

(b) Experimental results of different vortex

Fig 24 Typical vortex in a sump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

15

For the model visual observation dye was injected around the intake structure near

the surface and submergence as Fig 24 (b) shows And the model test results of

vortex was classified as Fig 25

(a) Free Surface Vortices

(b) Submerged Vortices

Fig 25 Vortex classifications

There are a number of researches focused on the numerical theory to detect the

inception of a visible vortex As the diameters of vortices are usually much smaller

compared with the computing grid size to compensate for the lack of vortex

resolution in numerical simulation a stretching vortex model is applied to the local

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

16

flow field around semi-analytically identified vortex positions [21][22] The brief

explanations are as follows

a) Free surface vortex

The criteria expressed by equation 26 and 27 is applied to determine the

visible inception of a free surface vortex

≫ 1 (26)

=

gt 1 (27)

where P is the pressure drop at the vortex core and Ph is the static head at the

vortex element and f is the drag force acting at a unit length of water surface

vortex dimple based on Stokes approximation and caused by the down flow

Further f is the force acting on the bubble in the direction from the high

pressure region to the low pressure region like buoyancy

b) Submerged vortex

Equation 28 is applied to determine the visible inception of a submerged vortex

infin =

infin

infin gt 1 (28)

where is the static pressure at the vortex core infin is the atmospheric pressure

and is the critical cavitation inception pressure and nearly equal to the

saturated vapor pressure

3) Swirl angle

Swirl angle predicts the intensity of flow rotation Fig 26 shows experiment

calculation method with swirl meter and attempts of the CFD calculation method

were conducted as Fig 27 A four blade zero-pitch swirl meter which is supported

by a low friction bearing will be installed at about four suction pipe diameter (d)

downstream from the pump suction to measure swirl angle of pump approaching

flow For the swirl meter the tip to tip blade diameter is 075d and the length in

flow direction is 06 and one of the four blade is painted yellow as a reference to

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

17

count revolutions As flow swirl is usually unsteady the observation time of swirl

meter reading should be a continuous period of time

Fig 26 Typical swirl meter

(29)

(210)

(211)

Where qV flow rotational speed at swirl meter (ms)

zV average axial velocity at swirl meter (ms)

n Number of revolution per minute (rpm)

d Diameter of pipe at the swirl meter (m)

q flow rate (m3s)

zV

Vqq 1tan -=

)(60

smnd

Vtimestimes

=p

q

)(4

2

smd

qVz

times=

p

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

18

For the swirl angle calculating method with CFD the key point is to obtain the

average tangential velocity so the swirl check circles were created as illustrated in

Fig 27 The swirl check circle is located at the section of 4d height with diameter

of 025d 05d and 075d respectively

Swirl check circle

Fig 27 Schematic view of swirl angle numerical calculation

23

1 7501 501 250

acute++

=aringaringaring

N

i

N

i

NNVNVNV

Vi

q (212)

where

qV average tangential velocity (ms)

V circumferential velocity of check point in check circle

respectively (ms)

N check point total numbers in single check circle

4) Velocity

Velocity measurement at bell throat by Pitot tube was implemented by ANSIHI

method Typical Pitot tube configuration is shown in Fig 28 Eight measuring

points are defined as Fig 29 Time averaged velocity at eight points was obtained

and these eight data were again averaged to give throat sectional spatial velocity

average

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

19

Fig 28 Example of Pitot tube Fig 29 Measuring points at bell throat

224 Test criteria

There is the criteria for satisfactory performance of the sump model test and

summarized in Table 23

1) Free surface and submerged vortices

Free surface vortices entering the pump must be less severe than vortices with

coherent core Dye core vortices may be acceptable only if they occur less than 10

of the time or only for infrequent pump operating conditions

Submerged vortex which starts from the sump walls induce vibration and noise

problems Dye core submerged vortices extending into the pump bell mouth is not

allowed

2) Swirl angles

Swirl angle is used to check the flow rotation in the suction pipe The calculation

method is referred to section 223

Short-term maximum and long-term average values must be less than 5 degrees

Maximum short-term swirl angles up to 7 degrees may be acceptable only if they

occur less than 10 of the time or only for infrequent pump operating conditions

3)Velocity

Each point time-averaged velocity shall not be larger than 10 of sectional average

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

20

Standard deviation of time-varying velocity at each point was obtained and its

value shall not be larger than 10 of the time-averaged velocity at each point

1) Time-averaged

Variation[] = 100acute-

a

a

V

VV aV Averaged at section

2) Time-varying

Standard deviation= )1(

)(1i

2

-

-aring=

n

xxn

i

Variation[] = (Standard deviation X ) x 100

X Averaged at point

Table 23 Permission criteria of ANSIHI

Vortex occurence Swirl angle

Free surface vortex Submerged vortex

Observation

time Above 10 minutes Above 10 minutes

Above 10

minutes

Permission

criteria

Dye-core

Vortex(F3)

below 110 total

time

Dye-core

Vortex(S2)

below 110 total

time

Below

5degConditional

intermittent

below 7deg

225 Remedial measures for problem intakes

Usually in the preliminary test the worst flow condition is simulated And based on

the results remedial measures for problem intakes are discussed in this part

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

21

1) Overall approach flow patterns

The characteristics of flow approaching an intake structure is one of the foremost

considerations for the designer It is difficult to predict the effects of a given set of

flow conditions upstream from an intake structure Fig 210 depicts several typical

approach flow patterns of pump sump with different operating conditions

Fig 210 Examples of approach flow patterns

Generally speaking for the intake problems especially for the multiple pumps

sump there are the main measures using dividing walls placed between the pumps

as a partitioned structure trash racks with elongated bars guide piers or a number

of structural columns at the bay entrance can provide some assistance in controlling

cross-flow baffles flow distributors and screen mesh etc are applied to improve

the flow distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

22

2) Problems around pump intake

a) Free surface vortices

There are a number of factors that have an effect on the formation of surface

vortices To achieve a higher degree of certainty that objectionable surface vortices

do not form modifications can be made to intake structure to allow operation at

practical depths of submergence and the minimum submergence is referred in Table

22 the use of suction umbrellas vertical curtain walls and horizontal gratings

b) Submerged vortices

In the vicinity of intake bell there are the most complicated flow patterns and flow

makes the most changes in direction while maintaining a constant acceleration into

the pump bell to prevent local flow separation turbulence and vortex formation

Fig 211 represents a sample of various devices to inhibit the different submerged

vortex

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

23

Fig 211 Methods to reduce submerged vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

24

c) Preswirl

The approach flow distributions determine whether flow preswirl exists and a

sufficiently laterally skewed approach flow cause rotating around the intake bell

The most effective way of reducing preswirl is to establish a relatively uniform

approach flow within each pump bay by using the baffling schemes controlling

cross-flow and expanding concentrated flow discussed in the above part

To prevent non-uniform velocity distribution swirls and vortices necessary

modifications are conducted such as fore bay length baffle wall distributors

curtain wall mesh screen guide vane floor splitter wall fillet etc As the above

measures applied modification test will be conducted to check all the flow

conditions and compare with the preliminary results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

25

Chapter 3 Methodology

CFD (Computational fluid dynamics) is the technique which can design cycles to

better performance and reduce costs and time In CFD we solve the governing

equations of given physics (may be differential form or integral form) using some

numerical techniques like Finite Difference Method (FDM) Finite Element Method

(FEM) or Finite Volume Method (FVM) Although CFD plays a great part in the

design of turbomachinery CFD is not a replacement for experimental or analytical

approach In this study the numerical analysis was performed by the commercial

code ANSYS CFX 130 All aspects of the CFD study including modelling

meshing solving and post-processing of the results

31 Numerical modeling

ANSYS CFX solves the unsteady Navier-Stokes equations in their conservation

form The instantaneous equation of mass (continuity) in the stationary frame is

expressed as equation

ρ

δ + nabla ∙ (ρU) = 0 (31)

And the instantaneous equation for momentum is expressed as shown in equation

ρ

δ + nabla ∙ (ρU⨂U) = minusnablap + nabla ∙ τ + S (32)

Where τ is the stress tensor ( including both normal and shear components of the

stress)

These instantaneous equations are averaged for turbulent flows leading to additional

terms that need to be solved While the Navier-Stokes equations describe both

laminar and turbulent flows without addition terms realistic flows involve length

scales much smaller than the smallest finite volume mesh A Direct Numerical

Simulation of these flows would require significantly more computing power than

what is available now or in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

26

Therefore much research has been done to predict the effects of turbulence by

using turbulence models These models account for the effects of turbulence

without the use of a very fine mesh or direct numerical simulation

These turbulence models modify the transport equations by adding averaged and

fluctuating components The transport equations are changed to equations (33) and

(34)

ρ

δ + nabla ∙ (ρU) = 0 (33)

ρ

+ nabla ∙ ρ U ⨂U = minusnablap + nabla ∙ τ minus ρu⨂u + S (34)

The mass equation is not changed but the momentum equation contains extra terms

which are the Reynolds stresses ρ u⨂u and the Reynolds flux ρ u ϕ These

Reynolds stresses need to be modeled by additional equations to obtain closure

Obtaining closure implies that there are a sufficient number of equations to solve for

all the unknowns including the Reynolds stresses and Reynolds fluxes

Various turbulence models provide various ways to obtain closure In this

investigation Shear Stress Transport (SST) model and k-ε model were utilized The

advantage of SST model is that combines the advantages of other turbulence models

(the k-ε Wilcox k-ω) The 3 turbulence models will be discussed briefly

32 Turbulence models

Turbulence models are necessary because we cannot afford big enough computers

to directly capture every scale of motion There are three modelling frameworks

Direct Numerical Simulation (DNS) Reynolds-Averaged Navier-Stokes

Equations(RANS) Large Eddy Simulation(LES)

As RANS method is applied in most productionseveral RANS turbulence models

differ in their usage of wall functions the number of additional variables solved for

and what these variables represent

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

27

Two equation turbulence models

Two-equation turbulence models are widely used in CFD as they give a good

compromise between computational power needed and accuracy The term lsquotwo

equationrsquo refers to the fact that these models solve for the velocity and length scales

using separate transport equations The turbulence velocity scale is obtained by

solving the transport equation The turbulent length scale is estimated from two

properties of the turbulence field namely the turbulent kinetic energy and the

dissipation rate The dissipation rate of the kinetic energy is obtained from its

transport equation The most widely used are k-ε and k-ω two equation models In

the next section the k-ε Wilcox k-ω BSL k-ω and SST models will be briefly

discussed

321 k-ε turbulence model

The k-epsilon model solves for two variables k the turbulent kinetic energy and

epsilon the rate of dissipation of kinetic energy Wall functions are used in this

model so the flow in the buffer region is not simulated The k-epsilon model is very

popular for industrial applications due to its good convergence rate and relatively

low memory requirements It does not very accurately compute flow fields that

exhibit adverse pressure gradients strong curvature to the flow or jet flow It does

perform well for external flow problems around complex geometries

The k-ε model introduces k (m2s2) as the turbulence kinetic energy and ε (m2s3) as

the turbulence eddy dissipation The continuity equation remains the same

ρ

+ nabla ∙ (ρ U) = 0 (35)

The momentum equation changes as shown by equation

δρ

δ + nabla ∙ (ρU⨂U) = minusnablapprime + nabla ∙ μ

nablaU + (nablaU) + S (36)

Where SM is the sum of body forces μeff is the effective viscosity accounting for

turbulence and prsquo is the modified pressure The k-ε model uses the concept of eddy

viscosity giving the equation for effective viscosity as shown by equation

μ

= μ + μ (37)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

28

μt is the turbulence viscosity is linked to the turbulence kinetic energy and

dissipation by the equation

μ = Cμρ

ε (38)

Where Cμ is a constant

The values for k and ε come from the differential transport equations for the

turbulence kinetic energy and the turbulence dissipation rate

The turbulence kinetic energy equation is given as equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus ρε (39)

The turbulence dissipation rate is given by equation

(ρε)

+ nabla ∙ (ρUε) = nabla ∙ μ +

μ

σε nablaε +

ε

(Cε (P + Pε ) minus Cε ρε) (310)

Where Cε1 Cε2 σk σε are constants

Pk is the turbulence production due to viscous forces and is modeled by the

equation

P = μ nablaU ∙ nablaU + nablaU minus

nabla ∙ U 3μ

nabla ∙ U + ρk (311)

If a buoyancy term is added to the previous equation if the full buoyancy model is

used

The buoyancy term Pkb is modeled as

P = minusμ

ρσρg ∙ nablaρ (312)

322 Wilcox k-ω turbulence model

The k-omega model is similar to k-epsilon instead however it solves for omega mdash

the specific rate of dissipation of kinetic energy It also uses wall functions and

therefore has comparable memory requirements It has more difficulty converging

and is quite sensitive to the initial guess at the solution Hence the k-epsilon model

is often used first to find an initial condition for solving the k-omega model The k-

omega model is useful in many cases where the k-epsilon model is not accurate

such as internal flows flows that exhibit strong curvature separated flows and jets

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

29

This model has an advantage over the k-ε model where it does not involve complex

linear damping functions for near wall calculations at low Reynolds numbers

The k-ω model assumes that the turbulence viscosity is related to the turbulence

kinetic energy k and the turbulent frequency ω by the equation

μ = ρ

ω (313)

The transport equation for k is given by the equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus βprimeρ k ω (314)

The transport equation for ω is shown as equation

(ρω)

δ + nabla ∙ (ρ U ω) = nabla ∙ μ +

μ

μω

nablaω + αω

P + Pω minus β ρ ω (315)

The production rate of turbulence (Pk) is calculated as shown previously in the k-ε

section

The recommended values for the constants in the above equations are

βprime = 009

α = 59

β = 0075

σ = 2

σω = 2

The Reynolds stress tensor ρ u⨂u is calculated by

minusρ u⨂u = μ nablaU + (nablaU) minus

part ρk + μ

nabla ∙ U (316)

323 Shear stress transport model

The disadvantage of the Wilcox model is the strong sensitivity to free-stream

conditions Therefore a blending of the k-ω model near the surface and the k-ε in

the outer region was made by Menter [23] which resulted in the formulation of the

BSL k-ω turbulence model It consists of a transformation of the k-ε model to a k-ω

formulation and subsequently adding the resulting equations The Wilcox model is

multiplied by a blending function F1 and the transformed k-ε by another function 1-

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

30

F1 F1 is a function of wall distance (being the value of one near the surface and zero

outside the boundary layer) Outside the boundary and on the edge of the boundary

layer the standard k-ε model is used [24]

SST model is a combination of the k-epsilon in the free stream and the k-omega

models near the walls It does not use wall functions and tends to be most accurate

when solving the flow near the wall The SST model does not always converge to

the solution quickly so the k-epsilon or k-omega models are often solved first to

give good initial conditions

The k-ω based Shear Stress Transport (SST) model of Menter was applied for

turbulence treatment The transport equations for the SST model are expressed

below where the turbulent kinetic energy lsquokrsquo and turbulent frequency or dissipation

per unit turbulent kinetic energy are computed by using the following relations

For Turbulence Kinetic Energy

kkPukt

kTkk Ntilde+Ntilde+-=Ntilde+

para

para)[()( nsnwb (317)

where Pk is the production limiter

For Specific Dissipation Rate

ww

swnsnbwaww

ww NtildeNtilde-+Ntilde+Ntilde+-=Ntilde+para

para

1)1(2])[()( 21

22 kFSut

T (318)

The first blending function F1 is calculated from

iumlthorn

iumlyacute

uuml

iumlicirc

iumliacute

igrave

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

uacuteuacuteucirc

ugrave

ecircecirceuml

eacute

dividedivideoslash

oumlccedilccedilegrave

aelig=

4

22

21

4500maxmintanh

yCD

k

yy

kF

kw

wrs

w

n

wb (319)

y is the distance to the nearest wall and v is the kinematic viscosity

divideoslash

oumlccedilegrave

aeligNtildeNtilde= -10

2 101

2max ww

rsww kCDk (320)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

31

And Kinematic eddy viscosity

)max( 21

1

SF

kT

wa

an = (321)

where F2 is a blending function which restricts the limiter to the wall boundary and

S is the invariant measure of the strain rate

uacuteuacute

ucirc

ugrave

ecircecirc

euml

eacute

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

dividedivideoslash

oumlccedilccedilegrave

aelig=

2

22

5002maxtanh

w

n

wb yy

kF (322)

33 Brief introduction of simulation process

All aspects of CFD study including modelling meshing solving and post-

processing of the results are briefly introduced in this chapter

331 Pre-processing technique

The speed and accuracy of simulation results depending fundamentally on meshing

and the ability to prepare model geometry correctly

The steps in pre-processing are explained as follows

1) Domain Definitions

The domain definition depends mainly on the physics involved in the problem and

objective of the problem

a) Geometry creation or geometry import

b) Geometry cleanup

i) Removing the parts not necessary for simulation under consideration

ii) Closing the gaps in the geometry

iii) Removing small surfaces curves to make meshing process simple

2)Grid generation

To solve the governing equations the numerical methods like Finite Difference

Finite Volume or Finite Element Method to divide the domain into small parts and

solve the equations on one part at a time are applied

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 23: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

11

Adverse hydraulic conditions that can affect pump performance include free and

submerged vortices swirl approaching the pump impeller flow separation at the

pump bell and a non-uniform axial velocity distribution at the suction

The purpose of model test is to examine these adverse hydraulic conditions and to

investigate measure against them

221 The scale effects

One possible difficulty in scale modeling is that it is impossible to reduce all

pertinent forces by the same factor With Froude-scale models the inertial and

gravitational forces are reduced similarly but the viscous and surface tension forces

cannot be simultaneously reduced The extra influence of these forces is known as

a ldquoscale effectrdquo A number of investigations have indicated that the scale effects for

the model are avoided if the Reynolds number based on the inlet flow and

submergence or intake diameter exceeds approximately 3x 104 and the Weber

number exceeds 150 [19]

In most engineering applications involving closed-conduit and open-channel flow

the Reynolds number limits are far exceeded and the flows are fully turbulent

In a sump model Froude number (Fr) and Reynolds number (Re) are the most

important non-dimensional parameters Some reduction in Re should be

introduced to the sump model However flow patterns are generally very similar

at high Re

No specific geometric scale rate is recommended but reasonable large

geometric scale to minimize viscous and surface tension scale effects and reproduce

the flow pattern

Re= uDν (22)

We= u2ρDσ (23)

where

u average axial velocity (ms)

ν kinematic viscosity of the liquid (m2s)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

12

ρ liquid density (kgm3)

σ surface tension of liquidair interface

According to the ANSIHI 98 the viscous scale effects on vortex may be negligible

if the resulting dimensionless numbers must meet these values Re>6times104 and

We>240

To allow practical visual observations of flow patterns accurate measurements of

swirl and velocity distribution and sufficient dimensional control the model scale

shall yield a bay width of at least 300mm a minimum liquid depth of at least

150mm and a pump throat or suction diameter of at least 80mm

222 Principle of similarity

In the similitude analysis the geometric and flow similarity requires the model and

the flow to be identical as the real model and flow This is to ensure the result

obtained from the model study is well presented in order to predict full scale

behavior Model involving a free surface are operated using Froude similarity since

the flow process is controlled by gravity and inertial forces

For similarity of the flow patterns the Froude number shall be equal in model and

prototype

Fr =

=

(24)

and as the scale ratio λ was selected in section 221 the following equation is

obtained

V = V times (

) = V times λ or Q = Q times (

) = Q times λ (25)

where Fr is the Froude number λ is the scale ratio of model dimensions to prototype

The test of 15 times Froude scaled flow should be conducted to ascertain the

potential scale affect on free surface vortices

Models of closed conduit piping system leading to a pump suction are not operated

based on Froude similitude However flow patterns are generally very similar at

high Re A minimum value of 1times105 for the Re at the pump suction is

recommended to ensure the flow patterns are correctly scaled

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

13

223 Preliminary model test

The initial design shall be tested first to identify any hydraulic problems

1) Vortex Formation Mechanisms

There are various investigations that provide insight into the fundamental processes

leading to the development of vortices both in experiment and numerical simulation

Shin et al [20] demonstrated that two basic mechanisms lead to inlet vortex

formation The first mechanism involves the development of an inlet vortex due to

the amplification of ambient vorticity in the approach flow as vortex lines

are convicted into the inlet The second mechanism involves the development of a

trailing vortex in the vicinity of the intake as a result of the variation in

circulation along the inlet For this second case a vortex can develop in a flow that

is irrotational upstream and the vortex development therefore does not depend

on the presence of ambient vorticity Shin et al investigation on kinematic

parameters indicate that the strength of an inlet-vortex or trailing vortex system

increases with decreasing distance from the surface However for an inlet in

an upstream irrotational flow two counter rotating vortices can still trail from

the rear of the inlet

2) Classification of vortex type

As illustrated in Fig 24 vortices in the vicinity of pump intakes may be adjacent to

the channel bottom or a channel wall (ie submerged vortices) or they may appear

adjacent to the free surface (ie free surface vortex)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

14

(a) Schematic diagram of vortices in a sump

(b) Experimental results of different vortex

Fig 24 Typical vortex in a sump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

15

For the model visual observation dye was injected around the intake structure near

the surface and submergence as Fig 24 (b) shows And the model test results of

vortex was classified as Fig 25

(a) Free Surface Vortices

(b) Submerged Vortices

Fig 25 Vortex classifications

There are a number of researches focused on the numerical theory to detect the

inception of a visible vortex As the diameters of vortices are usually much smaller

compared with the computing grid size to compensate for the lack of vortex

resolution in numerical simulation a stretching vortex model is applied to the local

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

16

flow field around semi-analytically identified vortex positions [21][22] The brief

explanations are as follows

a) Free surface vortex

The criteria expressed by equation 26 and 27 is applied to determine the

visible inception of a free surface vortex

≫ 1 (26)

=

gt 1 (27)

where P is the pressure drop at the vortex core and Ph is the static head at the

vortex element and f is the drag force acting at a unit length of water surface

vortex dimple based on Stokes approximation and caused by the down flow

Further f is the force acting on the bubble in the direction from the high

pressure region to the low pressure region like buoyancy

b) Submerged vortex

Equation 28 is applied to determine the visible inception of a submerged vortex

infin =

infin

infin gt 1 (28)

where is the static pressure at the vortex core infin is the atmospheric pressure

and is the critical cavitation inception pressure and nearly equal to the

saturated vapor pressure

3) Swirl angle

Swirl angle predicts the intensity of flow rotation Fig 26 shows experiment

calculation method with swirl meter and attempts of the CFD calculation method

were conducted as Fig 27 A four blade zero-pitch swirl meter which is supported

by a low friction bearing will be installed at about four suction pipe diameter (d)

downstream from the pump suction to measure swirl angle of pump approaching

flow For the swirl meter the tip to tip blade diameter is 075d and the length in

flow direction is 06 and one of the four blade is painted yellow as a reference to

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

17

count revolutions As flow swirl is usually unsteady the observation time of swirl

meter reading should be a continuous period of time

Fig 26 Typical swirl meter

(29)

(210)

(211)

Where qV flow rotational speed at swirl meter (ms)

zV average axial velocity at swirl meter (ms)

n Number of revolution per minute (rpm)

d Diameter of pipe at the swirl meter (m)

q flow rate (m3s)

zV

Vqq 1tan -=

)(60

smnd

Vtimestimes

=p

q

)(4

2

smd

qVz

times=

p

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

18

For the swirl angle calculating method with CFD the key point is to obtain the

average tangential velocity so the swirl check circles were created as illustrated in

Fig 27 The swirl check circle is located at the section of 4d height with diameter

of 025d 05d and 075d respectively

Swirl check circle

Fig 27 Schematic view of swirl angle numerical calculation

23

1 7501 501 250

acute++

=aringaringaring

N

i

N

i

NNVNVNV

Vi

q (212)

where

qV average tangential velocity (ms)

V circumferential velocity of check point in check circle

respectively (ms)

N check point total numbers in single check circle

4) Velocity

Velocity measurement at bell throat by Pitot tube was implemented by ANSIHI

method Typical Pitot tube configuration is shown in Fig 28 Eight measuring

points are defined as Fig 29 Time averaged velocity at eight points was obtained

and these eight data were again averaged to give throat sectional spatial velocity

average

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

19

Fig 28 Example of Pitot tube Fig 29 Measuring points at bell throat

224 Test criteria

There is the criteria for satisfactory performance of the sump model test and

summarized in Table 23

1) Free surface and submerged vortices

Free surface vortices entering the pump must be less severe than vortices with

coherent core Dye core vortices may be acceptable only if they occur less than 10

of the time or only for infrequent pump operating conditions

Submerged vortex which starts from the sump walls induce vibration and noise

problems Dye core submerged vortices extending into the pump bell mouth is not

allowed

2) Swirl angles

Swirl angle is used to check the flow rotation in the suction pipe The calculation

method is referred to section 223

Short-term maximum and long-term average values must be less than 5 degrees

Maximum short-term swirl angles up to 7 degrees may be acceptable only if they

occur less than 10 of the time or only for infrequent pump operating conditions

3)Velocity

Each point time-averaged velocity shall not be larger than 10 of sectional average

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

20

Standard deviation of time-varying velocity at each point was obtained and its

value shall not be larger than 10 of the time-averaged velocity at each point

1) Time-averaged

Variation[] = 100acute-

a

a

V

VV aV Averaged at section

2) Time-varying

Standard deviation= )1(

)(1i

2

-

-aring=

n

xxn

i

Variation[] = (Standard deviation X ) x 100

X Averaged at point

Table 23 Permission criteria of ANSIHI

Vortex occurence Swirl angle

Free surface vortex Submerged vortex

Observation

time Above 10 minutes Above 10 minutes

Above 10

minutes

Permission

criteria

Dye-core

Vortex(F3)

below 110 total

time

Dye-core

Vortex(S2)

below 110 total

time

Below

5degConditional

intermittent

below 7deg

225 Remedial measures for problem intakes

Usually in the preliminary test the worst flow condition is simulated And based on

the results remedial measures for problem intakes are discussed in this part

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

21

1) Overall approach flow patterns

The characteristics of flow approaching an intake structure is one of the foremost

considerations for the designer It is difficult to predict the effects of a given set of

flow conditions upstream from an intake structure Fig 210 depicts several typical

approach flow patterns of pump sump with different operating conditions

Fig 210 Examples of approach flow patterns

Generally speaking for the intake problems especially for the multiple pumps

sump there are the main measures using dividing walls placed between the pumps

as a partitioned structure trash racks with elongated bars guide piers or a number

of structural columns at the bay entrance can provide some assistance in controlling

cross-flow baffles flow distributors and screen mesh etc are applied to improve

the flow distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

22

2) Problems around pump intake

a) Free surface vortices

There are a number of factors that have an effect on the formation of surface

vortices To achieve a higher degree of certainty that objectionable surface vortices

do not form modifications can be made to intake structure to allow operation at

practical depths of submergence and the minimum submergence is referred in Table

22 the use of suction umbrellas vertical curtain walls and horizontal gratings

b) Submerged vortices

In the vicinity of intake bell there are the most complicated flow patterns and flow

makes the most changes in direction while maintaining a constant acceleration into

the pump bell to prevent local flow separation turbulence and vortex formation

Fig 211 represents a sample of various devices to inhibit the different submerged

vortex

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

23

Fig 211 Methods to reduce submerged vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

24

c) Preswirl

The approach flow distributions determine whether flow preswirl exists and a

sufficiently laterally skewed approach flow cause rotating around the intake bell

The most effective way of reducing preswirl is to establish a relatively uniform

approach flow within each pump bay by using the baffling schemes controlling

cross-flow and expanding concentrated flow discussed in the above part

To prevent non-uniform velocity distribution swirls and vortices necessary

modifications are conducted such as fore bay length baffle wall distributors

curtain wall mesh screen guide vane floor splitter wall fillet etc As the above

measures applied modification test will be conducted to check all the flow

conditions and compare with the preliminary results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

25

Chapter 3 Methodology

CFD (Computational fluid dynamics) is the technique which can design cycles to

better performance and reduce costs and time In CFD we solve the governing

equations of given physics (may be differential form or integral form) using some

numerical techniques like Finite Difference Method (FDM) Finite Element Method

(FEM) or Finite Volume Method (FVM) Although CFD plays a great part in the

design of turbomachinery CFD is not a replacement for experimental or analytical

approach In this study the numerical analysis was performed by the commercial

code ANSYS CFX 130 All aspects of the CFD study including modelling

meshing solving and post-processing of the results

31 Numerical modeling

ANSYS CFX solves the unsteady Navier-Stokes equations in their conservation

form The instantaneous equation of mass (continuity) in the stationary frame is

expressed as equation

ρ

δ + nabla ∙ (ρU) = 0 (31)

And the instantaneous equation for momentum is expressed as shown in equation

ρ

δ + nabla ∙ (ρU⨂U) = minusnablap + nabla ∙ τ + S (32)

Where τ is the stress tensor ( including both normal and shear components of the

stress)

These instantaneous equations are averaged for turbulent flows leading to additional

terms that need to be solved While the Navier-Stokes equations describe both

laminar and turbulent flows without addition terms realistic flows involve length

scales much smaller than the smallest finite volume mesh A Direct Numerical

Simulation of these flows would require significantly more computing power than

what is available now or in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

26

Therefore much research has been done to predict the effects of turbulence by

using turbulence models These models account for the effects of turbulence

without the use of a very fine mesh or direct numerical simulation

These turbulence models modify the transport equations by adding averaged and

fluctuating components The transport equations are changed to equations (33) and

(34)

ρ

δ + nabla ∙ (ρU) = 0 (33)

ρ

+ nabla ∙ ρ U ⨂U = minusnablap + nabla ∙ τ minus ρu⨂u + S (34)

The mass equation is not changed but the momentum equation contains extra terms

which are the Reynolds stresses ρ u⨂u and the Reynolds flux ρ u ϕ These

Reynolds stresses need to be modeled by additional equations to obtain closure

Obtaining closure implies that there are a sufficient number of equations to solve for

all the unknowns including the Reynolds stresses and Reynolds fluxes

Various turbulence models provide various ways to obtain closure In this

investigation Shear Stress Transport (SST) model and k-ε model were utilized The

advantage of SST model is that combines the advantages of other turbulence models

(the k-ε Wilcox k-ω) The 3 turbulence models will be discussed briefly

32 Turbulence models

Turbulence models are necessary because we cannot afford big enough computers

to directly capture every scale of motion There are three modelling frameworks

Direct Numerical Simulation (DNS) Reynolds-Averaged Navier-Stokes

Equations(RANS) Large Eddy Simulation(LES)

As RANS method is applied in most productionseveral RANS turbulence models

differ in their usage of wall functions the number of additional variables solved for

and what these variables represent

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

27

Two equation turbulence models

Two-equation turbulence models are widely used in CFD as they give a good

compromise between computational power needed and accuracy The term lsquotwo

equationrsquo refers to the fact that these models solve for the velocity and length scales

using separate transport equations The turbulence velocity scale is obtained by

solving the transport equation The turbulent length scale is estimated from two

properties of the turbulence field namely the turbulent kinetic energy and the

dissipation rate The dissipation rate of the kinetic energy is obtained from its

transport equation The most widely used are k-ε and k-ω two equation models In

the next section the k-ε Wilcox k-ω BSL k-ω and SST models will be briefly

discussed

321 k-ε turbulence model

The k-epsilon model solves for two variables k the turbulent kinetic energy and

epsilon the rate of dissipation of kinetic energy Wall functions are used in this

model so the flow in the buffer region is not simulated The k-epsilon model is very

popular for industrial applications due to its good convergence rate and relatively

low memory requirements It does not very accurately compute flow fields that

exhibit adverse pressure gradients strong curvature to the flow or jet flow It does

perform well for external flow problems around complex geometries

The k-ε model introduces k (m2s2) as the turbulence kinetic energy and ε (m2s3) as

the turbulence eddy dissipation The continuity equation remains the same

ρ

+ nabla ∙ (ρ U) = 0 (35)

The momentum equation changes as shown by equation

δρ

δ + nabla ∙ (ρU⨂U) = minusnablapprime + nabla ∙ μ

nablaU + (nablaU) + S (36)

Where SM is the sum of body forces μeff is the effective viscosity accounting for

turbulence and prsquo is the modified pressure The k-ε model uses the concept of eddy

viscosity giving the equation for effective viscosity as shown by equation

μ

= μ + μ (37)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

28

μt is the turbulence viscosity is linked to the turbulence kinetic energy and

dissipation by the equation

μ = Cμρ

ε (38)

Where Cμ is a constant

The values for k and ε come from the differential transport equations for the

turbulence kinetic energy and the turbulence dissipation rate

The turbulence kinetic energy equation is given as equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus ρε (39)

The turbulence dissipation rate is given by equation

(ρε)

+ nabla ∙ (ρUε) = nabla ∙ μ +

μ

σε nablaε +

ε

(Cε (P + Pε ) minus Cε ρε) (310)

Where Cε1 Cε2 σk σε are constants

Pk is the turbulence production due to viscous forces and is modeled by the

equation

P = μ nablaU ∙ nablaU + nablaU minus

nabla ∙ U 3μ

nabla ∙ U + ρk (311)

If a buoyancy term is added to the previous equation if the full buoyancy model is

used

The buoyancy term Pkb is modeled as

P = minusμ

ρσρg ∙ nablaρ (312)

322 Wilcox k-ω turbulence model

The k-omega model is similar to k-epsilon instead however it solves for omega mdash

the specific rate of dissipation of kinetic energy It also uses wall functions and

therefore has comparable memory requirements It has more difficulty converging

and is quite sensitive to the initial guess at the solution Hence the k-epsilon model

is often used first to find an initial condition for solving the k-omega model The k-

omega model is useful in many cases where the k-epsilon model is not accurate

such as internal flows flows that exhibit strong curvature separated flows and jets

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

29

This model has an advantage over the k-ε model where it does not involve complex

linear damping functions for near wall calculations at low Reynolds numbers

The k-ω model assumes that the turbulence viscosity is related to the turbulence

kinetic energy k and the turbulent frequency ω by the equation

μ = ρ

ω (313)

The transport equation for k is given by the equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus βprimeρ k ω (314)

The transport equation for ω is shown as equation

(ρω)

δ + nabla ∙ (ρ U ω) = nabla ∙ μ +

μ

μω

nablaω + αω

P + Pω minus β ρ ω (315)

The production rate of turbulence (Pk) is calculated as shown previously in the k-ε

section

The recommended values for the constants in the above equations are

βprime = 009

α = 59

β = 0075

σ = 2

σω = 2

The Reynolds stress tensor ρ u⨂u is calculated by

minusρ u⨂u = μ nablaU + (nablaU) minus

part ρk + μ

nabla ∙ U (316)

323 Shear stress transport model

The disadvantage of the Wilcox model is the strong sensitivity to free-stream

conditions Therefore a blending of the k-ω model near the surface and the k-ε in

the outer region was made by Menter [23] which resulted in the formulation of the

BSL k-ω turbulence model It consists of a transformation of the k-ε model to a k-ω

formulation and subsequently adding the resulting equations The Wilcox model is

multiplied by a blending function F1 and the transformed k-ε by another function 1-

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

30

F1 F1 is a function of wall distance (being the value of one near the surface and zero

outside the boundary layer) Outside the boundary and on the edge of the boundary

layer the standard k-ε model is used [24]

SST model is a combination of the k-epsilon in the free stream and the k-omega

models near the walls It does not use wall functions and tends to be most accurate

when solving the flow near the wall The SST model does not always converge to

the solution quickly so the k-epsilon or k-omega models are often solved first to

give good initial conditions

The k-ω based Shear Stress Transport (SST) model of Menter was applied for

turbulence treatment The transport equations for the SST model are expressed

below where the turbulent kinetic energy lsquokrsquo and turbulent frequency or dissipation

per unit turbulent kinetic energy are computed by using the following relations

For Turbulence Kinetic Energy

kkPukt

kTkk Ntilde+Ntilde+-=Ntilde+

para

para)[()( nsnwb (317)

where Pk is the production limiter

For Specific Dissipation Rate

ww

swnsnbwaww

ww NtildeNtilde-+Ntilde+Ntilde+-=Ntilde+para

para

1)1(2])[()( 21

22 kFSut

T (318)

The first blending function F1 is calculated from

iumlthorn

iumlyacute

uuml

iumlicirc

iumliacute

igrave

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

uacuteuacuteucirc

ugrave

ecircecirceuml

eacute

dividedivideoslash

oumlccedilccedilegrave

aelig=

4

22

21

4500maxmintanh

yCD

k

yy

kF

kw

wrs

w

n

wb (319)

y is the distance to the nearest wall and v is the kinematic viscosity

divideoslash

oumlccedilegrave

aeligNtildeNtilde= -10

2 101

2max ww

rsww kCDk (320)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

31

And Kinematic eddy viscosity

)max( 21

1

SF

kT

wa

an = (321)

where F2 is a blending function which restricts the limiter to the wall boundary and

S is the invariant measure of the strain rate

uacuteuacute

ucirc

ugrave

ecircecirc

euml

eacute

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

dividedivideoslash

oumlccedilccedilegrave

aelig=

2

22

5002maxtanh

w

n

wb yy

kF (322)

33 Brief introduction of simulation process

All aspects of CFD study including modelling meshing solving and post-

processing of the results are briefly introduced in this chapter

331 Pre-processing technique

The speed and accuracy of simulation results depending fundamentally on meshing

and the ability to prepare model geometry correctly

The steps in pre-processing are explained as follows

1) Domain Definitions

The domain definition depends mainly on the physics involved in the problem and

objective of the problem

a) Geometry creation or geometry import

b) Geometry cleanup

i) Removing the parts not necessary for simulation under consideration

ii) Closing the gaps in the geometry

iii) Removing small surfaces curves to make meshing process simple

2)Grid generation

To solve the governing equations the numerical methods like Finite Difference

Finite Volume or Finite Element Method to divide the domain into small parts and

solve the equations on one part at a time are applied

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 24: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

12

ρ liquid density (kgm3)

σ surface tension of liquidair interface

According to the ANSIHI 98 the viscous scale effects on vortex may be negligible

if the resulting dimensionless numbers must meet these values Re>6times104 and

We>240

To allow practical visual observations of flow patterns accurate measurements of

swirl and velocity distribution and sufficient dimensional control the model scale

shall yield a bay width of at least 300mm a minimum liquid depth of at least

150mm and a pump throat or suction diameter of at least 80mm

222 Principle of similarity

In the similitude analysis the geometric and flow similarity requires the model and

the flow to be identical as the real model and flow This is to ensure the result

obtained from the model study is well presented in order to predict full scale

behavior Model involving a free surface are operated using Froude similarity since

the flow process is controlled by gravity and inertial forces

For similarity of the flow patterns the Froude number shall be equal in model and

prototype

Fr =

=

(24)

and as the scale ratio λ was selected in section 221 the following equation is

obtained

V = V times (

) = V times λ or Q = Q times (

) = Q times λ (25)

where Fr is the Froude number λ is the scale ratio of model dimensions to prototype

The test of 15 times Froude scaled flow should be conducted to ascertain the

potential scale affect on free surface vortices

Models of closed conduit piping system leading to a pump suction are not operated

based on Froude similitude However flow patterns are generally very similar at

high Re A minimum value of 1times105 for the Re at the pump suction is

recommended to ensure the flow patterns are correctly scaled

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

13

223 Preliminary model test

The initial design shall be tested first to identify any hydraulic problems

1) Vortex Formation Mechanisms

There are various investigations that provide insight into the fundamental processes

leading to the development of vortices both in experiment and numerical simulation

Shin et al [20] demonstrated that two basic mechanisms lead to inlet vortex

formation The first mechanism involves the development of an inlet vortex due to

the amplification of ambient vorticity in the approach flow as vortex lines

are convicted into the inlet The second mechanism involves the development of a

trailing vortex in the vicinity of the intake as a result of the variation in

circulation along the inlet For this second case a vortex can develop in a flow that

is irrotational upstream and the vortex development therefore does not depend

on the presence of ambient vorticity Shin et al investigation on kinematic

parameters indicate that the strength of an inlet-vortex or trailing vortex system

increases with decreasing distance from the surface However for an inlet in

an upstream irrotational flow two counter rotating vortices can still trail from

the rear of the inlet

2) Classification of vortex type

As illustrated in Fig 24 vortices in the vicinity of pump intakes may be adjacent to

the channel bottom or a channel wall (ie submerged vortices) or they may appear

adjacent to the free surface (ie free surface vortex)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

14

(a) Schematic diagram of vortices in a sump

(b) Experimental results of different vortex

Fig 24 Typical vortex in a sump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

15

For the model visual observation dye was injected around the intake structure near

the surface and submergence as Fig 24 (b) shows And the model test results of

vortex was classified as Fig 25

(a) Free Surface Vortices

(b) Submerged Vortices

Fig 25 Vortex classifications

There are a number of researches focused on the numerical theory to detect the

inception of a visible vortex As the diameters of vortices are usually much smaller

compared with the computing grid size to compensate for the lack of vortex

resolution in numerical simulation a stretching vortex model is applied to the local

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

16

flow field around semi-analytically identified vortex positions [21][22] The brief

explanations are as follows

a) Free surface vortex

The criteria expressed by equation 26 and 27 is applied to determine the

visible inception of a free surface vortex

≫ 1 (26)

=

gt 1 (27)

where P is the pressure drop at the vortex core and Ph is the static head at the

vortex element and f is the drag force acting at a unit length of water surface

vortex dimple based on Stokes approximation and caused by the down flow

Further f is the force acting on the bubble in the direction from the high

pressure region to the low pressure region like buoyancy

b) Submerged vortex

Equation 28 is applied to determine the visible inception of a submerged vortex

infin =

infin

infin gt 1 (28)

where is the static pressure at the vortex core infin is the atmospheric pressure

and is the critical cavitation inception pressure and nearly equal to the

saturated vapor pressure

3) Swirl angle

Swirl angle predicts the intensity of flow rotation Fig 26 shows experiment

calculation method with swirl meter and attempts of the CFD calculation method

were conducted as Fig 27 A four blade zero-pitch swirl meter which is supported

by a low friction bearing will be installed at about four suction pipe diameter (d)

downstream from the pump suction to measure swirl angle of pump approaching

flow For the swirl meter the tip to tip blade diameter is 075d and the length in

flow direction is 06 and one of the four blade is painted yellow as a reference to

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

17

count revolutions As flow swirl is usually unsteady the observation time of swirl

meter reading should be a continuous period of time

Fig 26 Typical swirl meter

(29)

(210)

(211)

Where qV flow rotational speed at swirl meter (ms)

zV average axial velocity at swirl meter (ms)

n Number of revolution per minute (rpm)

d Diameter of pipe at the swirl meter (m)

q flow rate (m3s)

zV

Vqq 1tan -=

)(60

smnd

Vtimestimes

=p

q

)(4

2

smd

qVz

times=

p

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

18

For the swirl angle calculating method with CFD the key point is to obtain the

average tangential velocity so the swirl check circles were created as illustrated in

Fig 27 The swirl check circle is located at the section of 4d height with diameter

of 025d 05d and 075d respectively

Swirl check circle

Fig 27 Schematic view of swirl angle numerical calculation

23

1 7501 501 250

acute++

=aringaringaring

N

i

N

i

NNVNVNV

Vi

q (212)

where

qV average tangential velocity (ms)

V circumferential velocity of check point in check circle

respectively (ms)

N check point total numbers in single check circle

4) Velocity

Velocity measurement at bell throat by Pitot tube was implemented by ANSIHI

method Typical Pitot tube configuration is shown in Fig 28 Eight measuring

points are defined as Fig 29 Time averaged velocity at eight points was obtained

and these eight data were again averaged to give throat sectional spatial velocity

average

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

19

Fig 28 Example of Pitot tube Fig 29 Measuring points at bell throat

224 Test criteria

There is the criteria for satisfactory performance of the sump model test and

summarized in Table 23

1) Free surface and submerged vortices

Free surface vortices entering the pump must be less severe than vortices with

coherent core Dye core vortices may be acceptable only if they occur less than 10

of the time or only for infrequent pump operating conditions

Submerged vortex which starts from the sump walls induce vibration and noise

problems Dye core submerged vortices extending into the pump bell mouth is not

allowed

2) Swirl angles

Swirl angle is used to check the flow rotation in the suction pipe The calculation

method is referred to section 223

Short-term maximum and long-term average values must be less than 5 degrees

Maximum short-term swirl angles up to 7 degrees may be acceptable only if they

occur less than 10 of the time or only for infrequent pump operating conditions

3)Velocity

Each point time-averaged velocity shall not be larger than 10 of sectional average

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

20

Standard deviation of time-varying velocity at each point was obtained and its

value shall not be larger than 10 of the time-averaged velocity at each point

1) Time-averaged

Variation[] = 100acute-

a

a

V

VV aV Averaged at section

2) Time-varying

Standard deviation= )1(

)(1i

2

-

-aring=

n

xxn

i

Variation[] = (Standard deviation X ) x 100

X Averaged at point

Table 23 Permission criteria of ANSIHI

Vortex occurence Swirl angle

Free surface vortex Submerged vortex

Observation

time Above 10 minutes Above 10 minutes

Above 10

minutes

Permission

criteria

Dye-core

Vortex(F3)

below 110 total

time

Dye-core

Vortex(S2)

below 110 total

time

Below

5degConditional

intermittent

below 7deg

225 Remedial measures for problem intakes

Usually in the preliminary test the worst flow condition is simulated And based on

the results remedial measures for problem intakes are discussed in this part

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

21

1) Overall approach flow patterns

The characteristics of flow approaching an intake structure is one of the foremost

considerations for the designer It is difficult to predict the effects of a given set of

flow conditions upstream from an intake structure Fig 210 depicts several typical

approach flow patterns of pump sump with different operating conditions

Fig 210 Examples of approach flow patterns

Generally speaking for the intake problems especially for the multiple pumps

sump there are the main measures using dividing walls placed between the pumps

as a partitioned structure trash racks with elongated bars guide piers or a number

of structural columns at the bay entrance can provide some assistance in controlling

cross-flow baffles flow distributors and screen mesh etc are applied to improve

the flow distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

22

2) Problems around pump intake

a) Free surface vortices

There are a number of factors that have an effect on the formation of surface

vortices To achieve a higher degree of certainty that objectionable surface vortices

do not form modifications can be made to intake structure to allow operation at

practical depths of submergence and the minimum submergence is referred in Table

22 the use of suction umbrellas vertical curtain walls and horizontal gratings

b) Submerged vortices

In the vicinity of intake bell there are the most complicated flow patterns and flow

makes the most changes in direction while maintaining a constant acceleration into

the pump bell to prevent local flow separation turbulence and vortex formation

Fig 211 represents a sample of various devices to inhibit the different submerged

vortex

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

23

Fig 211 Methods to reduce submerged vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

24

c) Preswirl

The approach flow distributions determine whether flow preswirl exists and a

sufficiently laterally skewed approach flow cause rotating around the intake bell

The most effective way of reducing preswirl is to establish a relatively uniform

approach flow within each pump bay by using the baffling schemes controlling

cross-flow and expanding concentrated flow discussed in the above part

To prevent non-uniform velocity distribution swirls and vortices necessary

modifications are conducted such as fore bay length baffle wall distributors

curtain wall mesh screen guide vane floor splitter wall fillet etc As the above

measures applied modification test will be conducted to check all the flow

conditions and compare with the preliminary results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

25

Chapter 3 Methodology

CFD (Computational fluid dynamics) is the technique which can design cycles to

better performance and reduce costs and time In CFD we solve the governing

equations of given physics (may be differential form or integral form) using some

numerical techniques like Finite Difference Method (FDM) Finite Element Method

(FEM) or Finite Volume Method (FVM) Although CFD plays a great part in the

design of turbomachinery CFD is not a replacement for experimental or analytical

approach In this study the numerical analysis was performed by the commercial

code ANSYS CFX 130 All aspects of the CFD study including modelling

meshing solving and post-processing of the results

31 Numerical modeling

ANSYS CFX solves the unsteady Navier-Stokes equations in their conservation

form The instantaneous equation of mass (continuity) in the stationary frame is

expressed as equation

ρ

δ + nabla ∙ (ρU) = 0 (31)

And the instantaneous equation for momentum is expressed as shown in equation

ρ

δ + nabla ∙ (ρU⨂U) = minusnablap + nabla ∙ τ + S (32)

Where τ is the stress tensor ( including both normal and shear components of the

stress)

These instantaneous equations are averaged for turbulent flows leading to additional

terms that need to be solved While the Navier-Stokes equations describe both

laminar and turbulent flows without addition terms realistic flows involve length

scales much smaller than the smallest finite volume mesh A Direct Numerical

Simulation of these flows would require significantly more computing power than

what is available now or in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

26

Therefore much research has been done to predict the effects of turbulence by

using turbulence models These models account for the effects of turbulence

without the use of a very fine mesh or direct numerical simulation

These turbulence models modify the transport equations by adding averaged and

fluctuating components The transport equations are changed to equations (33) and

(34)

ρ

δ + nabla ∙ (ρU) = 0 (33)

ρ

+ nabla ∙ ρ U ⨂U = minusnablap + nabla ∙ τ minus ρu⨂u + S (34)

The mass equation is not changed but the momentum equation contains extra terms

which are the Reynolds stresses ρ u⨂u and the Reynolds flux ρ u ϕ These

Reynolds stresses need to be modeled by additional equations to obtain closure

Obtaining closure implies that there are a sufficient number of equations to solve for

all the unknowns including the Reynolds stresses and Reynolds fluxes

Various turbulence models provide various ways to obtain closure In this

investigation Shear Stress Transport (SST) model and k-ε model were utilized The

advantage of SST model is that combines the advantages of other turbulence models

(the k-ε Wilcox k-ω) The 3 turbulence models will be discussed briefly

32 Turbulence models

Turbulence models are necessary because we cannot afford big enough computers

to directly capture every scale of motion There are three modelling frameworks

Direct Numerical Simulation (DNS) Reynolds-Averaged Navier-Stokes

Equations(RANS) Large Eddy Simulation(LES)

As RANS method is applied in most productionseveral RANS turbulence models

differ in their usage of wall functions the number of additional variables solved for

and what these variables represent

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

27

Two equation turbulence models

Two-equation turbulence models are widely used in CFD as they give a good

compromise between computational power needed and accuracy The term lsquotwo

equationrsquo refers to the fact that these models solve for the velocity and length scales

using separate transport equations The turbulence velocity scale is obtained by

solving the transport equation The turbulent length scale is estimated from two

properties of the turbulence field namely the turbulent kinetic energy and the

dissipation rate The dissipation rate of the kinetic energy is obtained from its

transport equation The most widely used are k-ε and k-ω two equation models In

the next section the k-ε Wilcox k-ω BSL k-ω and SST models will be briefly

discussed

321 k-ε turbulence model

The k-epsilon model solves for two variables k the turbulent kinetic energy and

epsilon the rate of dissipation of kinetic energy Wall functions are used in this

model so the flow in the buffer region is not simulated The k-epsilon model is very

popular for industrial applications due to its good convergence rate and relatively

low memory requirements It does not very accurately compute flow fields that

exhibit adverse pressure gradients strong curvature to the flow or jet flow It does

perform well for external flow problems around complex geometries

The k-ε model introduces k (m2s2) as the turbulence kinetic energy and ε (m2s3) as

the turbulence eddy dissipation The continuity equation remains the same

ρ

+ nabla ∙ (ρ U) = 0 (35)

The momentum equation changes as shown by equation

δρ

δ + nabla ∙ (ρU⨂U) = minusnablapprime + nabla ∙ μ

nablaU + (nablaU) + S (36)

Where SM is the sum of body forces μeff is the effective viscosity accounting for

turbulence and prsquo is the modified pressure The k-ε model uses the concept of eddy

viscosity giving the equation for effective viscosity as shown by equation

μ

= μ + μ (37)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

28

μt is the turbulence viscosity is linked to the turbulence kinetic energy and

dissipation by the equation

μ = Cμρ

ε (38)

Where Cμ is a constant

The values for k and ε come from the differential transport equations for the

turbulence kinetic energy and the turbulence dissipation rate

The turbulence kinetic energy equation is given as equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus ρε (39)

The turbulence dissipation rate is given by equation

(ρε)

+ nabla ∙ (ρUε) = nabla ∙ μ +

μ

σε nablaε +

ε

(Cε (P + Pε ) minus Cε ρε) (310)

Where Cε1 Cε2 σk σε are constants

Pk is the turbulence production due to viscous forces and is modeled by the

equation

P = μ nablaU ∙ nablaU + nablaU minus

nabla ∙ U 3μ

nabla ∙ U + ρk (311)

If a buoyancy term is added to the previous equation if the full buoyancy model is

used

The buoyancy term Pkb is modeled as

P = minusμ

ρσρg ∙ nablaρ (312)

322 Wilcox k-ω turbulence model

The k-omega model is similar to k-epsilon instead however it solves for omega mdash

the specific rate of dissipation of kinetic energy It also uses wall functions and

therefore has comparable memory requirements It has more difficulty converging

and is quite sensitive to the initial guess at the solution Hence the k-epsilon model

is often used first to find an initial condition for solving the k-omega model The k-

omega model is useful in many cases where the k-epsilon model is not accurate

such as internal flows flows that exhibit strong curvature separated flows and jets

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

29

This model has an advantage over the k-ε model where it does not involve complex

linear damping functions for near wall calculations at low Reynolds numbers

The k-ω model assumes that the turbulence viscosity is related to the turbulence

kinetic energy k and the turbulent frequency ω by the equation

μ = ρ

ω (313)

The transport equation for k is given by the equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus βprimeρ k ω (314)

The transport equation for ω is shown as equation

(ρω)

δ + nabla ∙ (ρ U ω) = nabla ∙ μ +

μ

μω

nablaω + αω

P + Pω minus β ρ ω (315)

The production rate of turbulence (Pk) is calculated as shown previously in the k-ε

section

The recommended values for the constants in the above equations are

βprime = 009

α = 59

β = 0075

σ = 2

σω = 2

The Reynolds stress tensor ρ u⨂u is calculated by

minusρ u⨂u = μ nablaU + (nablaU) minus

part ρk + μ

nabla ∙ U (316)

323 Shear stress transport model

The disadvantage of the Wilcox model is the strong sensitivity to free-stream

conditions Therefore a blending of the k-ω model near the surface and the k-ε in

the outer region was made by Menter [23] which resulted in the formulation of the

BSL k-ω turbulence model It consists of a transformation of the k-ε model to a k-ω

formulation and subsequently adding the resulting equations The Wilcox model is

multiplied by a blending function F1 and the transformed k-ε by another function 1-

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

30

F1 F1 is a function of wall distance (being the value of one near the surface and zero

outside the boundary layer) Outside the boundary and on the edge of the boundary

layer the standard k-ε model is used [24]

SST model is a combination of the k-epsilon in the free stream and the k-omega

models near the walls It does not use wall functions and tends to be most accurate

when solving the flow near the wall The SST model does not always converge to

the solution quickly so the k-epsilon or k-omega models are often solved first to

give good initial conditions

The k-ω based Shear Stress Transport (SST) model of Menter was applied for

turbulence treatment The transport equations for the SST model are expressed

below where the turbulent kinetic energy lsquokrsquo and turbulent frequency or dissipation

per unit turbulent kinetic energy are computed by using the following relations

For Turbulence Kinetic Energy

kkPukt

kTkk Ntilde+Ntilde+-=Ntilde+

para

para)[()( nsnwb (317)

where Pk is the production limiter

For Specific Dissipation Rate

ww

swnsnbwaww

ww NtildeNtilde-+Ntilde+Ntilde+-=Ntilde+para

para

1)1(2])[()( 21

22 kFSut

T (318)

The first blending function F1 is calculated from

iumlthorn

iumlyacute

uuml

iumlicirc

iumliacute

igrave

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

uacuteuacuteucirc

ugrave

ecircecirceuml

eacute

dividedivideoslash

oumlccedilccedilegrave

aelig=

4

22

21

4500maxmintanh

yCD

k

yy

kF

kw

wrs

w

n

wb (319)

y is the distance to the nearest wall and v is the kinematic viscosity

divideoslash

oumlccedilegrave

aeligNtildeNtilde= -10

2 101

2max ww

rsww kCDk (320)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

31

And Kinematic eddy viscosity

)max( 21

1

SF

kT

wa

an = (321)

where F2 is a blending function which restricts the limiter to the wall boundary and

S is the invariant measure of the strain rate

uacuteuacute

ucirc

ugrave

ecircecirc

euml

eacute

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

dividedivideoslash

oumlccedilccedilegrave

aelig=

2

22

5002maxtanh

w

n

wb yy

kF (322)

33 Brief introduction of simulation process

All aspects of CFD study including modelling meshing solving and post-

processing of the results are briefly introduced in this chapter

331 Pre-processing technique

The speed and accuracy of simulation results depending fundamentally on meshing

and the ability to prepare model geometry correctly

The steps in pre-processing are explained as follows

1) Domain Definitions

The domain definition depends mainly on the physics involved in the problem and

objective of the problem

a) Geometry creation or geometry import

b) Geometry cleanup

i) Removing the parts not necessary for simulation under consideration

ii) Closing the gaps in the geometry

iii) Removing small surfaces curves to make meshing process simple

2)Grid generation

To solve the governing equations the numerical methods like Finite Difference

Finite Volume or Finite Element Method to divide the domain into small parts and

solve the equations on one part at a time are applied

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 25: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

13

223 Preliminary model test

The initial design shall be tested first to identify any hydraulic problems

1) Vortex Formation Mechanisms

There are various investigations that provide insight into the fundamental processes

leading to the development of vortices both in experiment and numerical simulation

Shin et al [20] demonstrated that two basic mechanisms lead to inlet vortex

formation The first mechanism involves the development of an inlet vortex due to

the amplification of ambient vorticity in the approach flow as vortex lines

are convicted into the inlet The second mechanism involves the development of a

trailing vortex in the vicinity of the intake as a result of the variation in

circulation along the inlet For this second case a vortex can develop in a flow that

is irrotational upstream and the vortex development therefore does not depend

on the presence of ambient vorticity Shin et al investigation on kinematic

parameters indicate that the strength of an inlet-vortex or trailing vortex system

increases with decreasing distance from the surface However for an inlet in

an upstream irrotational flow two counter rotating vortices can still trail from

the rear of the inlet

2) Classification of vortex type

As illustrated in Fig 24 vortices in the vicinity of pump intakes may be adjacent to

the channel bottom or a channel wall (ie submerged vortices) or they may appear

adjacent to the free surface (ie free surface vortex)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

14

(a) Schematic diagram of vortices in a sump

(b) Experimental results of different vortex

Fig 24 Typical vortex in a sump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

15

For the model visual observation dye was injected around the intake structure near

the surface and submergence as Fig 24 (b) shows And the model test results of

vortex was classified as Fig 25

(a) Free Surface Vortices

(b) Submerged Vortices

Fig 25 Vortex classifications

There are a number of researches focused on the numerical theory to detect the

inception of a visible vortex As the diameters of vortices are usually much smaller

compared with the computing grid size to compensate for the lack of vortex

resolution in numerical simulation a stretching vortex model is applied to the local

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

16

flow field around semi-analytically identified vortex positions [21][22] The brief

explanations are as follows

a) Free surface vortex

The criteria expressed by equation 26 and 27 is applied to determine the

visible inception of a free surface vortex

≫ 1 (26)

=

gt 1 (27)

where P is the pressure drop at the vortex core and Ph is the static head at the

vortex element and f is the drag force acting at a unit length of water surface

vortex dimple based on Stokes approximation and caused by the down flow

Further f is the force acting on the bubble in the direction from the high

pressure region to the low pressure region like buoyancy

b) Submerged vortex

Equation 28 is applied to determine the visible inception of a submerged vortex

infin =

infin

infin gt 1 (28)

where is the static pressure at the vortex core infin is the atmospheric pressure

and is the critical cavitation inception pressure and nearly equal to the

saturated vapor pressure

3) Swirl angle

Swirl angle predicts the intensity of flow rotation Fig 26 shows experiment

calculation method with swirl meter and attempts of the CFD calculation method

were conducted as Fig 27 A four blade zero-pitch swirl meter which is supported

by a low friction bearing will be installed at about four suction pipe diameter (d)

downstream from the pump suction to measure swirl angle of pump approaching

flow For the swirl meter the tip to tip blade diameter is 075d and the length in

flow direction is 06 and one of the four blade is painted yellow as a reference to

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

17

count revolutions As flow swirl is usually unsteady the observation time of swirl

meter reading should be a continuous period of time

Fig 26 Typical swirl meter

(29)

(210)

(211)

Where qV flow rotational speed at swirl meter (ms)

zV average axial velocity at swirl meter (ms)

n Number of revolution per minute (rpm)

d Diameter of pipe at the swirl meter (m)

q flow rate (m3s)

zV

Vqq 1tan -=

)(60

smnd

Vtimestimes

=p

q

)(4

2

smd

qVz

times=

p

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

18

For the swirl angle calculating method with CFD the key point is to obtain the

average tangential velocity so the swirl check circles were created as illustrated in

Fig 27 The swirl check circle is located at the section of 4d height with diameter

of 025d 05d and 075d respectively

Swirl check circle

Fig 27 Schematic view of swirl angle numerical calculation

23

1 7501 501 250

acute++

=aringaringaring

N

i

N

i

NNVNVNV

Vi

q (212)

where

qV average tangential velocity (ms)

V circumferential velocity of check point in check circle

respectively (ms)

N check point total numbers in single check circle

4) Velocity

Velocity measurement at bell throat by Pitot tube was implemented by ANSIHI

method Typical Pitot tube configuration is shown in Fig 28 Eight measuring

points are defined as Fig 29 Time averaged velocity at eight points was obtained

and these eight data were again averaged to give throat sectional spatial velocity

average

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

19

Fig 28 Example of Pitot tube Fig 29 Measuring points at bell throat

224 Test criteria

There is the criteria for satisfactory performance of the sump model test and

summarized in Table 23

1) Free surface and submerged vortices

Free surface vortices entering the pump must be less severe than vortices with

coherent core Dye core vortices may be acceptable only if they occur less than 10

of the time or only for infrequent pump operating conditions

Submerged vortex which starts from the sump walls induce vibration and noise

problems Dye core submerged vortices extending into the pump bell mouth is not

allowed

2) Swirl angles

Swirl angle is used to check the flow rotation in the suction pipe The calculation

method is referred to section 223

Short-term maximum and long-term average values must be less than 5 degrees

Maximum short-term swirl angles up to 7 degrees may be acceptable only if they

occur less than 10 of the time or only for infrequent pump operating conditions

3)Velocity

Each point time-averaged velocity shall not be larger than 10 of sectional average

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

20

Standard deviation of time-varying velocity at each point was obtained and its

value shall not be larger than 10 of the time-averaged velocity at each point

1) Time-averaged

Variation[] = 100acute-

a

a

V

VV aV Averaged at section

2) Time-varying

Standard deviation= )1(

)(1i

2

-

-aring=

n

xxn

i

Variation[] = (Standard deviation X ) x 100

X Averaged at point

Table 23 Permission criteria of ANSIHI

Vortex occurence Swirl angle

Free surface vortex Submerged vortex

Observation

time Above 10 minutes Above 10 minutes

Above 10

minutes

Permission

criteria

Dye-core

Vortex(F3)

below 110 total

time

Dye-core

Vortex(S2)

below 110 total

time

Below

5degConditional

intermittent

below 7deg

225 Remedial measures for problem intakes

Usually in the preliminary test the worst flow condition is simulated And based on

the results remedial measures for problem intakes are discussed in this part

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

21

1) Overall approach flow patterns

The characteristics of flow approaching an intake structure is one of the foremost

considerations for the designer It is difficult to predict the effects of a given set of

flow conditions upstream from an intake structure Fig 210 depicts several typical

approach flow patterns of pump sump with different operating conditions

Fig 210 Examples of approach flow patterns

Generally speaking for the intake problems especially for the multiple pumps

sump there are the main measures using dividing walls placed between the pumps

as a partitioned structure trash racks with elongated bars guide piers or a number

of structural columns at the bay entrance can provide some assistance in controlling

cross-flow baffles flow distributors and screen mesh etc are applied to improve

the flow distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

22

2) Problems around pump intake

a) Free surface vortices

There are a number of factors that have an effect on the formation of surface

vortices To achieve a higher degree of certainty that objectionable surface vortices

do not form modifications can be made to intake structure to allow operation at

practical depths of submergence and the minimum submergence is referred in Table

22 the use of suction umbrellas vertical curtain walls and horizontal gratings

b) Submerged vortices

In the vicinity of intake bell there are the most complicated flow patterns and flow

makes the most changes in direction while maintaining a constant acceleration into

the pump bell to prevent local flow separation turbulence and vortex formation

Fig 211 represents a sample of various devices to inhibit the different submerged

vortex

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

23

Fig 211 Methods to reduce submerged vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

24

c) Preswirl

The approach flow distributions determine whether flow preswirl exists and a

sufficiently laterally skewed approach flow cause rotating around the intake bell

The most effective way of reducing preswirl is to establish a relatively uniform

approach flow within each pump bay by using the baffling schemes controlling

cross-flow and expanding concentrated flow discussed in the above part

To prevent non-uniform velocity distribution swirls and vortices necessary

modifications are conducted such as fore bay length baffle wall distributors

curtain wall mesh screen guide vane floor splitter wall fillet etc As the above

measures applied modification test will be conducted to check all the flow

conditions and compare with the preliminary results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

25

Chapter 3 Methodology

CFD (Computational fluid dynamics) is the technique which can design cycles to

better performance and reduce costs and time In CFD we solve the governing

equations of given physics (may be differential form or integral form) using some

numerical techniques like Finite Difference Method (FDM) Finite Element Method

(FEM) or Finite Volume Method (FVM) Although CFD plays a great part in the

design of turbomachinery CFD is not a replacement for experimental or analytical

approach In this study the numerical analysis was performed by the commercial

code ANSYS CFX 130 All aspects of the CFD study including modelling

meshing solving and post-processing of the results

31 Numerical modeling

ANSYS CFX solves the unsteady Navier-Stokes equations in their conservation

form The instantaneous equation of mass (continuity) in the stationary frame is

expressed as equation

ρ

δ + nabla ∙ (ρU) = 0 (31)

And the instantaneous equation for momentum is expressed as shown in equation

ρ

δ + nabla ∙ (ρU⨂U) = minusnablap + nabla ∙ τ + S (32)

Where τ is the stress tensor ( including both normal and shear components of the

stress)

These instantaneous equations are averaged for turbulent flows leading to additional

terms that need to be solved While the Navier-Stokes equations describe both

laminar and turbulent flows without addition terms realistic flows involve length

scales much smaller than the smallest finite volume mesh A Direct Numerical

Simulation of these flows would require significantly more computing power than

what is available now or in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

26

Therefore much research has been done to predict the effects of turbulence by

using turbulence models These models account for the effects of turbulence

without the use of a very fine mesh or direct numerical simulation

These turbulence models modify the transport equations by adding averaged and

fluctuating components The transport equations are changed to equations (33) and

(34)

ρ

δ + nabla ∙ (ρU) = 0 (33)

ρ

+ nabla ∙ ρ U ⨂U = minusnablap + nabla ∙ τ minus ρu⨂u + S (34)

The mass equation is not changed but the momentum equation contains extra terms

which are the Reynolds stresses ρ u⨂u and the Reynolds flux ρ u ϕ These

Reynolds stresses need to be modeled by additional equations to obtain closure

Obtaining closure implies that there are a sufficient number of equations to solve for

all the unknowns including the Reynolds stresses and Reynolds fluxes

Various turbulence models provide various ways to obtain closure In this

investigation Shear Stress Transport (SST) model and k-ε model were utilized The

advantage of SST model is that combines the advantages of other turbulence models

(the k-ε Wilcox k-ω) The 3 turbulence models will be discussed briefly

32 Turbulence models

Turbulence models are necessary because we cannot afford big enough computers

to directly capture every scale of motion There are three modelling frameworks

Direct Numerical Simulation (DNS) Reynolds-Averaged Navier-Stokes

Equations(RANS) Large Eddy Simulation(LES)

As RANS method is applied in most productionseveral RANS turbulence models

differ in their usage of wall functions the number of additional variables solved for

and what these variables represent

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

27

Two equation turbulence models

Two-equation turbulence models are widely used in CFD as they give a good

compromise between computational power needed and accuracy The term lsquotwo

equationrsquo refers to the fact that these models solve for the velocity and length scales

using separate transport equations The turbulence velocity scale is obtained by

solving the transport equation The turbulent length scale is estimated from two

properties of the turbulence field namely the turbulent kinetic energy and the

dissipation rate The dissipation rate of the kinetic energy is obtained from its

transport equation The most widely used are k-ε and k-ω two equation models In

the next section the k-ε Wilcox k-ω BSL k-ω and SST models will be briefly

discussed

321 k-ε turbulence model

The k-epsilon model solves for two variables k the turbulent kinetic energy and

epsilon the rate of dissipation of kinetic energy Wall functions are used in this

model so the flow in the buffer region is not simulated The k-epsilon model is very

popular for industrial applications due to its good convergence rate and relatively

low memory requirements It does not very accurately compute flow fields that

exhibit adverse pressure gradients strong curvature to the flow or jet flow It does

perform well for external flow problems around complex geometries

The k-ε model introduces k (m2s2) as the turbulence kinetic energy and ε (m2s3) as

the turbulence eddy dissipation The continuity equation remains the same

ρ

+ nabla ∙ (ρ U) = 0 (35)

The momentum equation changes as shown by equation

δρ

δ + nabla ∙ (ρU⨂U) = minusnablapprime + nabla ∙ μ

nablaU + (nablaU) + S (36)

Where SM is the sum of body forces μeff is the effective viscosity accounting for

turbulence and prsquo is the modified pressure The k-ε model uses the concept of eddy

viscosity giving the equation for effective viscosity as shown by equation

μ

= μ + μ (37)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

28

μt is the turbulence viscosity is linked to the turbulence kinetic energy and

dissipation by the equation

μ = Cμρ

ε (38)

Where Cμ is a constant

The values for k and ε come from the differential transport equations for the

turbulence kinetic energy and the turbulence dissipation rate

The turbulence kinetic energy equation is given as equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus ρε (39)

The turbulence dissipation rate is given by equation

(ρε)

+ nabla ∙ (ρUε) = nabla ∙ μ +

μ

σε nablaε +

ε

(Cε (P + Pε ) minus Cε ρε) (310)

Where Cε1 Cε2 σk σε are constants

Pk is the turbulence production due to viscous forces and is modeled by the

equation

P = μ nablaU ∙ nablaU + nablaU minus

nabla ∙ U 3μ

nabla ∙ U + ρk (311)

If a buoyancy term is added to the previous equation if the full buoyancy model is

used

The buoyancy term Pkb is modeled as

P = minusμ

ρσρg ∙ nablaρ (312)

322 Wilcox k-ω turbulence model

The k-omega model is similar to k-epsilon instead however it solves for omega mdash

the specific rate of dissipation of kinetic energy It also uses wall functions and

therefore has comparable memory requirements It has more difficulty converging

and is quite sensitive to the initial guess at the solution Hence the k-epsilon model

is often used first to find an initial condition for solving the k-omega model The k-

omega model is useful in many cases where the k-epsilon model is not accurate

such as internal flows flows that exhibit strong curvature separated flows and jets

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

29

This model has an advantage over the k-ε model where it does not involve complex

linear damping functions for near wall calculations at low Reynolds numbers

The k-ω model assumes that the turbulence viscosity is related to the turbulence

kinetic energy k and the turbulent frequency ω by the equation

μ = ρ

ω (313)

The transport equation for k is given by the equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus βprimeρ k ω (314)

The transport equation for ω is shown as equation

(ρω)

δ + nabla ∙ (ρ U ω) = nabla ∙ μ +

μ

μω

nablaω + αω

P + Pω minus β ρ ω (315)

The production rate of turbulence (Pk) is calculated as shown previously in the k-ε

section

The recommended values for the constants in the above equations are

βprime = 009

α = 59

β = 0075

σ = 2

σω = 2

The Reynolds stress tensor ρ u⨂u is calculated by

minusρ u⨂u = μ nablaU + (nablaU) minus

part ρk + μ

nabla ∙ U (316)

323 Shear stress transport model

The disadvantage of the Wilcox model is the strong sensitivity to free-stream

conditions Therefore a blending of the k-ω model near the surface and the k-ε in

the outer region was made by Menter [23] which resulted in the formulation of the

BSL k-ω turbulence model It consists of a transformation of the k-ε model to a k-ω

formulation and subsequently adding the resulting equations The Wilcox model is

multiplied by a blending function F1 and the transformed k-ε by another function 1-

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

30

F1 F1 is a function of wall distance (being the value of one near the surface and zero

outside the boundary layer) Outside the boundary and on the edge of the boundary

layer the standard k-ε model is used [24]

SST model is a combination of the k-epsilon in the free stream and the k-omega

models near the walls It does not use wall functions and tends to be most accurate

when solving the flow near the wall The SST model does not always converge to

the solution quickly so the k-epsilon or k-omega models are often solved first to

give good initial conditions

The k-ω based Shear Stress Transport (SST) model of Menter was applied for

turbulence treatment The transport equations for the SST model are expressed

below where the turbulent kinetic energy lsquokrsquo and turbulent frequency or dissipation

per unit turbulent kinetic energy are computed by using the following relations

For Turbulence Kinetic Energy

kkPukt

kTkk Ntilde+Ntilde+-=Ntilde+

para

para)[()( nsnwb (317)

where Pk is the production limiter

For Specific Dissipation Rate

ww

swnsnbwaww

ww NtildeNtilde-+Ntilde+Ntilde+-=Ntilde+para

para

1)1(2])[()( 21

22 kFSut

T (318)

The first blending function F1 is calculated from

iumlthorn

iumlyacute

uuml

iumlicirc

iumliacute

igrave

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

uacuteuacuteucirc

ugrave

ecircecirceuml

eacute

dividedivideoslash

oumlccedilccedilegrave

aelig=

4

22

21

4500maxmintanh

yCD

k

yy

kF

kw

wrs

w

n

wb (319)

y is the distance to the nearest wall and v is the kinematic viscosity

divideoslash

oumlccedilegrave

aeligNtildeNtilde= -10

2 101

2max ww

rsww kCDk (320)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

31

And Kinematic eddy viscosity

)max( 21

1

SF

kT

wa

an = (321)

where F2 is a blending function which restricts the limiter to the wall boundary and

S is the invariant measure of the strain rate

uacuteuacute

ucirc

ugrave

ecircecirc

euml

eacute

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

dividedivideoslash

oumlccedilccedilegrave

aelig=

2

22

5002maxtanh

w

n

wb yy

kF (322)

33 Brief introduction of simulation process

All aspects of CFD study including modelling meshing solving and post-

processing of the results are briefly introduced in this chapter

331 Pre-processing technique

The speed and accuracy of simulation results depending fundamentally on meshing

and the ability to prepare model geometry correctly

The steps in pre-processing are explained as follows

1) Domain Definitions

The domain definition depends mainly on the physics involved in the problem and

objective of the problem

a) Geometry creation or geometry import

b) Geometry cleanup

i) Removing the parts not necessary for simulation under consideration

ii) Closing the gaps in the geometry

iii) Removing small surfaces curves to make meshing process simple

2)Grid generation

To solve the governing equations the numerical methods like Finite Difference

Finite Volume or Finite Element Method to divide the domain into small parts and

solve the equations on one part at a time are applied

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 26: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

14

(a) Schematic diagram of vortices in a sump

(b) Experimental results of different vortex

Fig 24 Typical vortex in a sump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

15

For the model visual observation dye was injected around the intake structure near

the surface and submergence as Fig 24 (b) shows And the model test results of

vortex was classified as Fig 25

(a) Free Surface Vortices

(b) Submerged Vortices

Fig 25 Vortex classifications

There are a number of researches focused on the numerical theory to detect the

inception of a visible vortex As the diameters of vortices are usually much smaller

compared with the computing grid size to compensate for the lack of vortex

resolution in numerical simulation a stretching vortex model is applied to the local

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

16

flow field around semi-analytically identified vortex positions [21][22] The brief

explanations are as follows

a) Free surface vortex

The criteria expressed by equation 26 and 27 is applied to determine the

visible inception of a free surface vortex

≫ 1 (26)

=

gt 1 (27)

where P is the pressure drop at the vortex core and Ph is the static head at the

vortex element and f is the drag force acting at a unit length of water surface

vortex dimple based on Stokes approximation and caused by the down flow

Further f is the force acting on the bubble in the direction from the high

pressure region to the low pressure region like buoyancy

b) Submerged vortex

Equation 28 is applied to determine the visible inception of a submerged vortex

infin =

infin

infin gt 1 (28)

where is the static pressure at the vortex core infin is the atmospheric pressure

and is the critical cavitation inception pressure and nearly equal to the

saturated vapor pressure

3) Swirl angle

Swirl angle predicts the intensity of flow rotation Fig 26 shows experiment

calculation method with swirl meter and attempts of the CFD calculation method

were conducted as Fig 27 A four blade zero-pitch swirl meter which is supported

by a low friction bearing will be installed at about four suction pipe diameter (d)

downstream from the pump suction to measure swirl angle of pump approaching

flow For the swirl meter the tip to tip blade diameter is 075d and the length in

flow direction is 06 and one of the four blade is painted yellow as a reference to

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

17

count revolutions As flow swirl is usually unsteady the observation time of swirl

meter reading should be a continuous period of time

Fig 26 Typical swirl meter

(29)

(210)

(211)

Where qV flow rotational speed at swirl meter (ms)

zV average axial velocity at swirl meter (ms)

n Number of revolution per minute (rpm)

d Diameter of pipe at the swirl meter (m)

q flow rate (m3s)

zV

Vqq 1tan -=

)(60

smnd

Vtimestimes

=p

q

)(4

2

smd

qVz

times=

p

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

18

For the swirl angle calculating method with CFD the key point is to obtain the

average tangential velocity so the swirl check circles were created as illustrated in

Fig 27 The swirl check circle is located at the section of 4d height with diameter

of 025d 05d and 075d respectively

Swirl check circle

Fig 27 Schematic view of swirl angle numerical calculation

23

1 7501 501 250

acute++

=aringaringaring

N

i

N

i

NNVNVNV

Vi

q (212)

where

qV average tangential velocity (ms)

V circumferential velocity of check point in check circle

respectively (ms)

N check point total numbers in single check circle

4) Velocity

Velocity measurement at bell throat by Pitot tube was implemented by ANSIHI

method Typical Pitot tube configuration is shown in Fig 28 Eight measuring

points are defined as Fig 29 Time averaged velocity at eight points was obtained

and these eight data were again averaged to give throat sectional spatial velocity

average

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

19

Fig 28 Example of Pitot tube Fig 29 Measuring points at bell throat

224 Test criteria

There is the criteria for satisfactory performance of the sump model test and

summarized in Table 23

1) Free surface and submerged vortices

Free surface vortices entering the pump must be less severe than vortices with

coherent core Dye core vortices may be acceptable only if they occur less than 10

of the time or only for infrequent pump operating conditions

Submerged vortex which starts from the sump walls induce vibration and noise

problems Dye core submerged vortices extending into the pump bell mouth is not

allowed

2) Swirl angles

Swirl angle is used to check the flow rotation in the suction pipe The calculation

method is referred to section 223

Short-term maximum and long-term average values must be less than 5 degrees

Maximum short-term swirl angles up to 7 degrees may be acceptable only if they

occur less than 10 of the time or only for infrequent pump operating conditions

3)Velocity

Each point time-averaged velocity shall not be larger than 10 of sectional average

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

20

Standard deviation of time-varying velocity at each point was obtained and its

value shall not be larger than 10 of the time-averaged velocity at each point

1) Time-averaged

Variation[] = 100acute-

a

a

V

VV aV Averaged at section

2) Time-varying

Standard deviation= )1(

)(1i

2

-

-aring=

n

xxn

i

Variation[] = (Standard deviation X ) x 100

X Averaged at point

Table 23 Permission criteria of ANSIHI

Vortex occurence Swirl angle

Free surface vortex Submerged vortex

Observation

time Above 10 minutes Above 10 minutes

Above 10

minutes

Permission

criteria

Dye-core

Vortex(F3)

below 110 total

time

Dye-core

Vortex(S2)

below 110 total

time

Below

5degConditional

intermittent

below 7deg

225 Remedial measures for problem intakes

Usually in the preliminary test the worst flow condition is simulated And based on

the results remedial measures for problem intakes are discussed in this part

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

21

1) Overall approach flow patterns

The characteristics of flow approaching an intake structure is one of the foremost

considerations for the designer It is difficult to predict the effects of a given set of

flow conditions upstream from an intake structure Fig 210 depicts several typical

approach flow patterns of pump sump with different operating conditions

Fig 210 Examples of approach flow patterns

Generally speaking for the intake problems especially for the multiple pumps

sump there are the main measures using dividing walls placed between the pumps

as a partitioned structure trash racks with elongated bars guide piers or a number

of structural columns at the bay entrance can provide some assistance in controlling

cross-flow baffles flow distributors and screen mesh etc are applied to improve

the flow distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

22

2) Problems around pump intake

a) Free surface vortices

There are a number of factors that have an effect on the formation of surface

vortices To achieve a higher degree of certainty that objectionable surface vortices

do not form modifications can be made to intake structure to allow operation at

practical depths of submergence and the minimum submergence is referred in Table

22 the use of suction umbrellas vertical curtain walls and horizontal gratings

b) Submerged vortices

In the vicinity of intake bell there are the most complicated flow patterns and flow

makes the most changes in direction while maintaining a constant acceleration into

the pump bell to prevent local flow separation turbulence and vortex formation

Fig 211 represents a sample of various devices to inhibit the different submerged

vortex

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

23

Fig 211 Methods to reduce submerged vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

24

c) Preswirl

The approach flow distributions determine whether flow preswirl exists and a

sufficiently laterally skewed approach flow cause rotating around the intake bell

The most effective way of reducing preswirl is to establish a relatively uniform

approach flow within each pump bay by using the baffling schemes controlling

cross-flow and expanding concentrated flow discussed in the above part

To prevent non-uniform velocity distribution swirls and vortices necessary

modifications are conducted such as fore bay length baffle wall distributors

curtain wall mesh screen guide vane floor splitter wall fillet etc As the above

measures applied modification test will be conducted to check all the flow

conditions and compare with the preliminary results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

25

Chapter 3 Methodology

CFD (Computational fluid dynamics) is the technique which can design cycles to

better performance and reduce costs and time In CFD we solve the governing

equations of given physics (may be differential form or integral form) using some

numerical techniques like Finite Difference Method (FDM) Finite Element Method

(FEM) or Finite Volume Method (FVM) Although CFD plays a great part in the

design of turbomachinery CFD is not a replacement for experimental or analytical

approach In this study the numerical analysis was performed by the commercial

code ANSYS CFX 130 All aspects of the CFD study including modelling

meshing solving and post-processing of the results

31 Numerical modeling

ANSYS CFX solves the unsteady Navier-Stokes equations in their conservation

form The instantaneous equation of mass (continuity) in the stationary frame is

expressed as equation

ρ

δ + nabla ∙ (ρU) = 0 (31)

And the instantaneous equation for momentum is expressed as shown in equation

ρ

δ + nabla ∙ (ρU⨂U) = minusnablap + nabla ∙ τ + S (32)

Where τ is the stress tensor ( including both normal and shear components of the

stress)

These instantaneous equations are averaged for turbulent flows leading to additional

terms that need to be solved While the Navier-Stokes equations describe both

laminar and turbulent flows without addition terms realistic flows involve length

scales much smaller than the smallest finite volume mesh A Direct Numerical

Simulation of these flows would require significantly more computing power than

what is available now or in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

26

Therefore much research has been done to predict the effects of turbulence by

using turbulence models These models account for the effects of turbulence

without the use of a very fine mesh or direct numerical simulation

These turbulence models modify the transport equations by adding averaged and

fluctuating components The transport equations are changed to equations (33) and

(34)

ρ

δ + nabla ∙ (ρU) = 0 (33)

ρ

+ nabla ∙ ρ U ⨂U = minusnablap + nabla ∙ τ minus ρu⨂u + S (34)

The mass equation is not changed but the momentum equation contains extra terms

which are the Reynolds stresses ρ u⨂u and the Reynolds flux ρ u ϕ These

Reynolds stresses need to be modeled by additional equations to obtain closure

Obtaining closure implies that there are a sufficient number of equations to solve for

all the unknowns including the Reynolds stresses and Reynolds fluxes

Various turbulence models provide various ways to obtain closure In this

investigation Shear Stress Transport (SST) model and k-ε model were utilized The

advantage of SST model is that combines the advantages of other turbulence models

(the k-ε Wilcox k-ω) The 3 turbulence models will be discussed briefly

32 Turbulence models

Turbulence models are necessary because we cannot afford big enough computers

to directly capture every scale of motion There are three modelling frameworks

Direct Numerical Simulation (DNS) Reynolds-Averaged Navier-Stokes

Equations(RANS) Large Eddy Simulation(LES)

As RANS method is applied in most productionseveral RANS turbulence models

differ in their usage of wall functions the number of additional variables solved for

and what these variables represent

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

27

Two equation turbulence models

Two-equation turbulence models are widely used in CFD as they give a good

compromise between computational power needed and accuracy The term lsquotwo

equationrsquo refers to the fact that these models solve for the velocity and length scales

using separate transport equations The turbulence velocity scale is obtained by

solving the transport equation The turbulent length scale is estimated from two

properties of the turbulence field namely the turbulent kinetic energy and the

dissipation rate The dissipation rate of the kinetic energy is obtained from its

transport equation The most widely used are k-ε and k-ω two equation models In

the next section the k-ε Wilcox k-ω BSL k-ω and SST models will be briefly

discussed

321 k-ε turbulence model

The k-epsilon model solves for two variables k the turbulent kinetic energy and

epsilon the rate of dissipation of kinetic energy Wall functions are used in this

model so the flow in the buffer region is not simulated The k-epsilon model is very

popular for industrial applications due to its good convergence rate and relatively

low memory requirements It does not very accurately compute flow fields that

exhibit adverse pressure gradients strong curvature to the flow or jet flow It does

perform well for external flow problems around complex geometries

The k-ε model introduces k (m2s2) as the turbulence kinetic energy and ε (m2s3) as

the turbulence eddy dissipation The continuity equation remains the same

ρ

+ nabla ∙ (ρ U) = 0 (35)

The momentum equation changes as shown by equation

δρ

δ + nabla ∙ (ρU⨂U) = minusnablapprime + nabla ∙ μ

nablaU + (nablaU) + S (36)

Where SM is the sum of body forces μeff is the effective viscosity accounting for

turbulence and prsquo is the modified pressure The k-ε model uses the concept of eddy

viscosity giving the equation for effective viscosity as shown by equation

μ

= μ + μ (37)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

28

μt is the turbulence viscosity is linked to the turbulence kinetic energy and

dissipation by the equation

μ = Cμρ

ε (38)

Where Cμ is a constant

The values for k and ε come from the differential transport equations for the

turbulence kinetic energy and the turbulence dissipation rate

The turbulence kinetic energy equation is given as equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus ρε (39)

The turbulence dissipation rate is given by equation

(ρε)

+ nabla ∙ (ρUε) = nabla ∙ μ +

μ

σε nablaε +

ε

(Cε (P + Pε ) minus Cε ρε) (310)

Where Cε1 Cε2 σk σε are constants

Pk is the turbulence production due to viscous forces and is modeled by the

equation

P = μ nablaU ∙ nablaU + nablaU minus

nabla ∙ U 3μ

nabla ∙ U + ρk (311)

If a buoyancy term is added to the previous equation if the full buoyancy model is

used

The buoyancy term Pkb is modeled as

P = minusμ

ρσρg ∙ nablaρ (312)

322 Wilcox k-ω turbulence model

The k-omega model is similar to k-epsilon instead however it solves for omega mdash

the specific rate of dissipation of kinetic energy It also uses wall functions and

therefore has comparable memory requirements It has more difficulty converging

and is quite sensitive to the initial guess at the solution Hence the k-epsilon model

is often used first to find an initial condition for solving the k-omega model The k-

omega model is useful in many cases where the k-epsilon model is not accurate

such as internal flows flows that exhibit strong curvature separated flows and jets

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

29

This model has an advantage over the k-ε model where it does not involve complex

linear damping functions for near wall calculations at low Reynolds numbers

The k-ω model assumes that the turbulence viscosity is related to the turbulence

kinetic energy k and the turbulent frequency ω by the equation

μ = ρ

ω (313)

The transport equation for k is given by the equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus βprimeρ k ω (314)

The transport equation for ω is shown as equation

(ρω)

δ + nabla ∙ (ρ U ω) = nabla ∙ μ +

μ

μω

nablaω + αω

P + Pω minus β ρ ω (315)

The production rate of turbulence (Pk) is calculated as shown previously in the k-ε

section

The recommended values for the constants in the above equations are

βprime = 009

α = 59

β = 0075

σ = 2

σω = 2

The Reynolds stress tensor ρ u⨂u is calculated by

minusρ u⨂u = μ nablaU + (nablaU) minus

part ρk + μ

nabla ∙ U (316)

323 Shear stress transport model

The disadvantage of the Wilcox model is the strong sensitivity to free-stream

conditions Therefore a blending of the k-ω model near the surface and the k-ε in

the outer region was made by Menter [23] which resulted in the formulation of the

BSL k-ω turbulence model It consists of a transformation of the k-ε model to a k-ω

formulation and subsequently adding the resulting equations The Wilcox model is

multiplied by a blending function F1 and the transformed k-ε by another function 1-

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

30

F1 F1 is a function of wall distance (being the value of one near the surface and zero

outside the boundary layer) Outside the boundary and on the edge of the boundary

layer the standard k-ε model is used [24]

SST model is a combination of the k-epsilon in the free stream and the k-omega

models near the walls It does not use wall functions and tends to be most accurate

when solving the flow near the wall The SST model does not always converge to

the solution quickly so the k-epsilon or k-omega models are often solved first to

give good initial conditions

The k-ω based Shear Stress Transport (SST) model of Menter was applied for

turbulence treatment The transport equations for the SST model are expressed

below where the turbulent kinetic energy lsquokrsquo and turbulent frequency or dissipation

per unit turbulent kinetic energy are computed by using the following relations

For Turbulence Kinetic Energy

kkPukt

kTkk Ntilde+Ntilde+-=Ntilde+

para

para)[()( nsnwb (317)

where Pk is the production limiter

For Specific Dissipation Rate

ww

swnsnbwaww

ww NtildeNtilde-+Ntilde+Ntilde+-=Ntilde+para

para

1)1(2])[()( 21

22 kFSut

T (318)

The first blending function F1 is calculated from

iumlthorn

iumlyacute

uuml

iumlicirc

iumliacute

igrave

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

uacuteuacuteucirc

ugrave

ecircecirceuml

eacute

dividedivideoslash

oumlccedilccedilegrave

aelig=

4

22

21

4500maxmintanh

yCD

k

yy

kF

kw

wrs

w

n

wb (319)

y is the distance to the nearest wall and v is the kinematic viscosity

divideoslash

oumlccedilegrave

aeligNtildeNtilde= -10

2 101

2max ww

rsww kCDk (320)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

31

And Kinematic eddy viscosity

)max( 21

1

SF

kT

wa

an = (321)

where F2 is a blending function which restricts the limiter to the wall boundary and

S is the invariant measure of the strain rate

uacuteuacute

ucirc

ugrave

ecircecirc

euml

eacute

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

dividedivideoslash

oumlccedilccedilegrave

aelig=

2

22

5002maxtanh

w

n

wb yy

kF (322)

33 Brief introduction of simulation process

All aspects of CFD study including modelling meshing solving and post-

processing of the results are briefly introduced in this chapter

331 Pre-processing technique

The speed and accuracy of simulation results depending fundamentally on meshing

and the ability to prepare model geometry correctly

The steps in pre-processing are explained as follows

1) Domain Definitions

The domain definition depends mainly on the physics involved in the problem and

objective of the problem

a) Geometry creation or geometry import

b) Geometry cleanup

i) Removing the parts not necessary for simulation under consideration

ii) Closing the gaps in the geometry

iii) Removing small surfaces curves to make meshing process simple

2)Grid generation

To solve the governing equations the numerical methods like Finite Difference

Finite Volume or Finite Element Method to divide the domain into small parts and

solve the equations on one part at a time are applied

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 27: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

15

For the model visual observation dye was injected around the intake structure near

the surface and submergence as Fig 24 (b) shows And the model test results of

vortex was classified as Fig 25

(a) Free Surface Vortices

(b) Submerged Vortices

Fig 25 Vortex classifications

There are a number of researches focused on the numerical theory to detect the

inception of a visible vortex As the diameters of vortices are usually much smaller

compared with the computing grid size to compensate for the lack of vortex

resolution in numerical simulation a stretching vortex model is applied to the local

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

16

flow field around semi-analytically identified vortex positions [21][22] The brief

explanations are as follows

a) Free surface vortex

The criteria expressed by equation 26 and 27 is applied to determine the

visible inception of a free surface vortex

≫ 1 (26)

=

gt 1 (27)

where P is the pressure drop at the vortex core and Ph is the static head at the

vortex element and f is the drag force acting at a unit length of water surface

vortex dimple based on Stokes approximation and caused by the down flow

Further f is the force acting on the bubble in the direction from the high

pressure region to the low pressure region like buoyancy

b) Submerged vortex

Equation 28 is applied to determine the visible inception of a submerged vortex

infin =

infin

infin gt 1 (28)

where is the static pressure at the vortex core infin is the atmospheric pressure

and is the critical cavitation inception pressure and nearly equal to the

saturated vapor pressure

3) Swirl angle

Swirl angle predicts the intensity of flow rotation Fig 26 shows experiment

calculation method with swirl meter and attempts of the CFD calculation method

were conducted as Fig 27 A four blade zero-pitch swirl meter which is supported

by a low friction bearing will be installed at about four suction pipe diameter (d)

downstream from the pump suction to measure swirl angle of pump approaching

flow For the swirl meter the tip to tip blade diameter is 075d and the length in

flow direction is 06 and one of the four blade is painted yellow as a reference to

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

17

count revolutions As flow swirl is usually unsteady the observation time of swirl

meter reading should be a continuous period of time

Fig 26 Typical swirl meter

(29)

(210)

(211)

Where qV flow rotational speed at swirl meter (ms)

zV average axial velocity at swirl meter (ms)

n Number of revolution per minute (rpm)

d Diameter of pipe at the swirl meter (m)

q flow rate (m3s)

zV

Vqq 1tan -=

)(60

smnd

Vtimestimes

=p

q

)(4

2

smd

qVz

times=

p

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

18

For the swirl angle calculating method with CFD the key point is to obtain the

average tangential velocity so the swirl check circles were created as illustrated in

Fig 27 The swirl check circle is located at the section of 4d height with diameter

of 025d 05d and 075d respectively

Swirl check circle

Fig 27 Schematic view of swirl angle numerical calculation

23

1 7501 501 250

acute++

=aringaringaring

N

i

N

i

NNVNVNV

Vi

q (212)

where

qV average tangential velocity (ms)

V circumferential velocity of check point in check circle

respectively (ms)

N check point total numbers in single check circle

4) Velocity

Velocity measurement at bell throat by Pitot tube was implemented by ANSIHI

method Typical Pitot tube configuration is shown in Fig 28 Eight measuring

points are defined as Fig 29 Time averaged velocity at eight points was obtained

and these eight data were again averaged to give throat sectional spatial velocity

average

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

19

Fig 28 Example of Pitot tube Fig 29 Measuring points at bell throat

224 Test criteria

There is the criteria for satisfactory performance of the sump model test and

summarized in Table 23

1) Free surface and submerged vortices

Free surface vortices entering the pump must be less severe than vortices with

coherent core Dye core vortices may be acceptable only if they occur less than 10

of the time or only for infrequent pump operating conditions

Submerged vortex which starts from the sump walls induce vibration and noise

problems Dye core submerged vortices extending into the pump bell mouth is not

allowed

2) Swirl angles

Swirl angle is used to check the flow rotation in the suction pipe The calculation

method is referred to section 223

Short-term maximum and long-term average values must be less than 5 degrees

Maximum short-term swirl angles up to 7 degrees may be acceptable only if they

occur less than 10 of the time or only for infrequent pump operating conditions

3)Velocity

Each point time-averaged velocity shall not be larger than 10 of sectional average

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

20

Standard deviation of time-varying velocity at each point was obtained and its

value shall not be larger than 10 of the time-averaged velocity at each point

1) Time-averaged

Variation[] = 100acute-

a

a

V

VV aV Averaged at section

2) Time-varying

Standard deviation= )1(

)(1i

2

-

-aring=

n

xxn

i

Variation[] = (Standard deviation X ) x 100

X Averaged at point

Table 23 Permission criteria of ANSIHI

Vortex occurence Swirl angle

Free surface vortex Submerged vortex

Observation

time Above 10 minutes Above 10 minutes

Above 10

minutes

Permission

criteria

Dye-core

Vortex(F3)

below 110 total

time

Dye-core

Vortex(S2)

below 110 total

time

Below

5degConditional

intermittent

below 7deg

225 Remedial measures for problem intakes

Usually in the preliminary test the worst flow condition is simulated And based on

the results remedial measures for problem intakes are discussed in this part

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

21

1) Overall approach flow patterns

The characteristics of flow approaching an intake structure is one of the foremost

considerations for the designer It is difficult to predict the effects of a given set of

flow conditions upstream from an intake structure Fig 210 depicts several typical

approach flow patterns of pump sump with different operating conditions

Fig 210 Examples of approach flow patterns

Generally speaking for the intake problems especially for the multiple pumps

sump there are the main measures using dividing walls placed between the pumps

as a partitioned structure trash racks with elongated bars guide piers or a number

of structural columns at the bay entrance can provide some assistance in controlling

cross-flow baffles flow distributors and screen mesh etc are applied to improve

the flow distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

22

2) Problems around pump intake

a) Free surface vortices

There are a number of factors that have an effect on the formation of surface

vortices To achieve a higher degree of certainty that objectionable surface vortices

do not form modifications can be made to intake structure to allow operation at

practical depths of submergence and the minimum submergence is referred in Table

22 the use of suction umbrellas vertical curtain walls and horizontal gratings

b) Submerged vortices

In the vicinity of intake bell there are the most complicated flow patterns and flow

makes the most changes in direction while maintaining a constant acceleration into

the pump bell to prevent local flow separation turbulence and vortex formation

Fig 211 represents a sample of various devices to inhibit the different submerged

vortex

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

23

Fig 211 Methods to reduce submerged vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

24

c) Preswirl

The approach flow distributions determine whether flow preswirl exists and a

sufficiently laterally skewed approach flow cause rotating around the intake bell

The most effective way of reducing preswirl is to establish a relatively uniform

approach flow within each pump bay by using the baffling schemes controlling

cross-flow and expanding concentrated flow discussed in the above part

To prevent non-uniform velocity distribution swirls and vortices necessary

modifications are conducted such as fore bay length baffle wall distributors

curtain wall mesh screen guide vane floor splitter wall fillet etc As the above

measures applied modification test will be conducted to check all the flow

conditions and compare with the preliminary results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

25

Chapter 3 Methodology

CFD (Computational fluid dynamics) is the technique which can design cycles to

better performance and reduce costs and time In CFD we solve the governing

equations of given physics (may be differential form or integral form) using some

numerical techniques like Finite Difference Method (FDM) Finite Element Method

(FEM) or Finite Volume Method (FVM) Although CFD plays a great part in the

design of turbomachinery CFD is not a replacement for experimental or analytical

approach In this study the numerical analysis was performed by the commercial

code ANSYS CFX 130 All aspects of the CFD study including modelling

meshing solving and post-processing of the results

31 Numerical modeling

ANSYS CFX solves the unsteady Navier-Stokes equations in their conservation

form The instantaneous equation of mass (continuity) in the stationary frame is

expressed as equation

ρ

δ + nabla ∙ (ρU) = 0 (31)

And the instantaneous equation for momentum is expressed as shown in equation

ρ

δ + nabla ∙ (ρU⨂U) = minusnablap + nabla ∙ τ + S (32)

Where τ is the stress tensor ( including both normal and shear components of the

stress)

These instantaneous equations are averaged for turbulent flows leading to additional

terms that need to be solved While the Navier-Stokes equations describe both

laminar and turbulent flows without addition terms realistic flows involve length

scales much smaller than the smallest finite volume mesh A Direct Numerical

Simulation of these flows would require significantly more computing power than

what is available now or in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

26

Therefore much research has been done to predict the effects of turbulence by

using turbulence models These models account for the effects of turbulence

without the use of a very fine mesh or direct numerical simulation

These turbulence models modify the transport equations by adding averaged and

fluctuating components The transport equations are changed to equations (33) and

(34)

ρ

δ + nabla ∙ (ρU) = 0 (33)

ρ

+ nabla ∙ ρ U ⨂U = minusnablap + nabla ∙ τ minus ρu⨂u + S (34)

The mass equation is not changed but the momentum equation contains extra terms

which are the Reynolds stresses ρ u⨂u and the Reynolds flux ρ u ϕ These

Reynolds stresses need to be modeled by additional equations to obtain closure

Obtaining closure implies that there are a sufficient number of equations to solve for

all the unknowns including the Reynolds stresses and Reynolds fluxes

Various turbulence models provide various ways to obtain closure In this

investigation Shear Stress Transport (SST) model and k-ε model were utilized The

advantage of SST model is that combines the advantages of other turbulence models

(the k-ε Wilcox k-ω) The 3 turbulence models will be discussed briefly

32 Turbulence models

Turbulence models are necessary because we cannot afford big enough computers

to directly capture every scale of motion There are three modelling frameworks

Direct Numerical Simulation (DNS) Reynolds-Averaged Navier-Stokes

Equations(RANS) Large Eddy Simulation(LES)

As RANS method is applied in most productionseveral RANS turbulence models

differ in their usage of wall functions the number of additional variables solved for

and what these variables represent

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

27

Two equation turbulence models

Two-equation turbulence models are widely used in CFD as they give a good

compromise between computational power needed and accuracy The term lsquotwo

equationrsquo refers to the fact that these models solve for the velocity and length scales

using separate transport equations The turbulence velocity scale is obtained by

solving the transport equation The turbulent length scale is estimated from two

properties of the turbulence field namely the turbulent kinetic energy and the

dissipation rate The dissipation rate of the kinetic energy is obtained from its

transport equation The most widely used are k-ε and k-ω two equation models In

the next section the k-ε Wilcox k-ω BSL k-ω and SST models will be briefly

discussed

321 k-ε turbulence model

The k-epsilon model solves for two variables k the turbulent kinetic energy and

epsilon the rate of dissipation of kinetic energy Wall functions are used in this

model so the flow in the buffer region is not simulated The k-epsilon model is very

popular for industrial applications due to its good convergence rate and relatively

low memory requirements It does not very accurately compute flow fields that

exhibit adverse pressure gradients strong curvature to the flow or jet flow It does

perform well for external flow problems around complex geometries

The k-ε model introduces k (m2s2) as the turbulence kinetic energy and ε (m2s3) as

the turbulence eddy dissipation The continuity equation remains the same

ρ

+ nabla ∙ (ρ U) = 0 (35)

The momentum equation changes as shown by equation

δρ

δ + nabla ∙ (ρU⨂U) = minusnablapprime + nabla ∙ μ

nablaU + (nablaU) + S (36)

Where SM is the sum of body forces μeff is the effective viscosity accounting for

turbulence and prsquo is the modified pressure The k-ε model uses the concept of eddy

viscosity giving the equation for effective viscosity as shown by equation

μ

= μ + μ (37)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

28

μt is the turbulence viscosity is linked to the turbulence kinetic energy and

dissipation by the equation

μ = Cμρ

ε (38)

Where Cμ is a constant

The values for k and ε come from the differential transport equations for the

turbulence kinetic energy and the turbulence dissipation rate

The turbulence kinetic energy equation is given as equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus ρε (39)

The turbulence dissipation rate is given by equation

(ρε)

+ nabla ∙ (ρUε) = nabla ∙ μ +

μ

σε nablaε +

ε

(Cε (P + Pε ) minus Cε ρε) (310)

Where Cε1 Cε2 σk σε are constants

Pk is the turbulence production due to viscous forces and is modeled by the

equation

P = μ nablaU ∙ nablaU + nablaU minus

nabla ∙ U 3μ

nabla ∙ U + ρk (311)

If a buoyancy term is added to the previous equation if the full buoyancy model is

used

The buoyancy term Pkb is modeled as

P = minusμ

ρσρg ∙ nablaρ (312)

322 Wilcox k-ω turbulence model

The k-omega model is similar to k-epsilon instead however it solves for omega mdash

the specific rate of dissipation of kinetic energy It also uses wall functions and

therefore has comparable memory requirements It has more difficulty converging

and is quite sensitive to the initial guess at the solution Hence the k-epsilon model

is often used first to find an initial condition for solving the k-omega model The k-

omega model is useful in many cases where the k-epsilon model is not accurate

such as internal flows flows that exhibit strong curvature separated flows and jets

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

29

This model has an advantage over the k-ε model where it does not involve complex

linear damping functions for near wall calculations at low Reynolds numbers

The k-ω model assumes that the turbulence viscosity is related to the turbulence

kinetic energy k and the turbulent frequency ω by the equation

μ = ρ

ω (313)

The transport equation for k is given by the equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus βprimeρ k ω (314)

The transport equation for ω is shown as equation

(ρω)

δ + nabla ∙ (ρ U ω) = nabla ∙ μ +

μ

μω

nablaω + αω

P + Pω minus β ρ ω (315)

The production rate of turbulence (Pk) is calculated as shown previously in the k-ε

section

The recommended values for the constants in the above equations are

βprime = 009

α = 59

β = 0075

σ = 2

σω = 2

The Reynolds stress tensor ρ u⨂u is calculated by

minusρ u⨂u = μ nablaU + (nablaU) minus

part ρk + μ

nabla ∙ U (316)

323 Shear stress transport model

The disadvantage of the Wilcox model is the strong sensitivity to free-stream

conditions Therefore a blending of the k-ω model near the surface and the k-ε in

the outer region was made by Menter [23] which resulted in the formulation of the

BSL k-ω turbulence model It consists of a transformation of the k-ε model to a k-ω

formulation and subsequently adding the resulting equations The Wilcox model is

multiplied by a blending function F1 and the transformed k-ε by another function 1-

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

30

F1 F1 is a function of wall distance (being the value of one near the surface and zero

outside the boundary layer) Outside the boundary and on the edge of the boundary

layer the standard k-ε model is used [24]

SST model is a combination of the k-epsilon in the free stream and the k-omega

models near the walls It does not use wall functions and tends to be most accurate

when solving the flow near the wall The SST model does not always converge to

the solution quickly so the k-epsilon or k-omega models are often solved first to

give good initial conditions

The k-ω based Shear Stress Transport (SST) model of Menter was applied for

turbulence treatment The transport equations for the SST model are expressed

below where the turbulent kinetic energy lsquokrsquo and turbulent frequency or dissipation

per unit turbulent kinetic energy are computed by using the following relations

For Turbulence Kinetic Energy

kkPukt

kTkk Ntilde+Ntilde+-=Ntilde+

para

para)[()( nsnwb (317)

where Pk is the production limiter

For Specific Dissipation Rate

ww

swnsnbwaww

ww NtildeNtilde-+Ntilde+Ntilde+-=Ntilde+para

para

1)1(2])[()( 21

22 kFSut

T (318)

The first blending function F1 is calculated from

iumlthorn

iumlyacute

uuml

iumlicirc

iumliacute

igrave

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

uacuteuacuteucirc

ugrave

ecircecirceuml

eacute

dividedivideoslash

oumlccedilccedilegrave

aelig=

4

22

21

4500maxmintanh

yCD

k

yy

kF

kw

wrs

w

n

wb (319)

y is the distance to the nearest wall and v is the kinematic viscosity

divideoslash

oumlccedilegrave

aeligNtildeNtilde= -10

2 101

2max ww

rsww kCDk (320)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

31

And Kinematic eddy viscosity

)max( 21

1

SF

kT

wa

an = (321)

where F2 is a blending function which restricts the limiter to the wall boundary and

S is the invariant measure of the strain rate

uacuteuacute

ucirc

ugrave

ecircecirc

euml

eacute

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

dividedivideoslash

oumlccedilccedilegrave

aelig=

2

22

5002maxtanh

w

n

wb yy

kF (322)

33 Brief introduction of simulation process

All aspects of CFD study including modelling meshing solving and post-

processing of the results are briefly introduced in this chapter

331 Pre-processing technique

The speed and accuracy of simulation results depending fundamentally on meshing

and the ability to prepare model geometry correctly

The steps in pre-processing are explained as follows

1) Domain Definitions

The domain definition depends mainly on the physics involved in the problem and

objective of the problem

a) Geometry creation or geometry import

b) Geometry cleanup

i) Removing the parts not necessary for simulation under consideration

ii) Closing the gaps in the geometry

iii) Removing small surfaces curves to make meshing process simple

2)Grid generation

To solve the governing equations the numerical methods like Finite Difference

Finite Volume or Finite Element Method to divide the domain into small parts and

solve the equations on one part at a time are applied

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 28: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

16

flow field around semi-analytically identified vortex positions [21][22] The brief

explanations are as follows

a) Free surface vortex

The criteria expressed by equation 26 and 27 is applied to determine the

visible inception of a free surface vortex

≫ 1 (26)

=

gt 1 (27)

where P is the pressure drop at the vortex core and Ph is the static head at the

vortex element and f is the drag force acting at a unit length of water surface

vortex dimple based on Stokes approximation and caused by the down flow

Further f is the force acting on the bubble in the direction from the high

pressure region to the low pressure region like buoyancy

b) Submerged vortex

Equation 28 is applied to determine the visible inception of a submerged vortex

infin =

infin

infin gt 1 (28)

where is the static pressure at the vortex core infin is the atmospheric pressure

and is the critical cavitation inception pressure and nearly equal to the

saturated vapor pressure

3) Swirl angle

Swirl angle predicts the intensity of flow rotation Fig 26 shows experiment

calculation method with swirl meter and attempts of the CFD calculation method

were conducted as Fig 27 A four blade zero-pitch swirl meter which is supported

by a low friction bearing will be installed at about four suction pipe diameter (d)

downstream from the pump suction to measure swirl angle of pump approaching

flow For the swirl meter the tip to tip blade diameter is 075d and the length in

flow direction is 06 and one of the four blade is painted yellow as a reference to

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

17

count revolutions As flow swirl is usually unsteady the observation time of swirl

meter reading should be a continuous period of time

Fig 26 Typical swirl meter

(29)

(210)

(211)

Where qV flow rotational speed at swirl meter (ms)

zV average axial velocity at swirl meter (ms)

n Number of revolution per minute (rpm)

d Diameter of pipe at the swirl meter (m)

q flow rate (m3s)

zV

Vqq 1tan -=

)(60

smnd

Vtimestimes

=p

q

)(4

2

smd

qVz

times=

p

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

18

For the swirl angle calculating method with CFD the key point is to obtain the

average tangential velocity so the swirl check circles were created as illustrated in

Fig 27 The swirl check circle is located at the section of 4d height with diameter

of 025d 05d and 075d respectively

Swirl check circle

Fig 27 Schematic view of swirl angle numerical calculation

23

1 7501 501 250

acute++

=aringaringaring

N

i

N

i

NNVNVNV

Vi

q (212)

where

qV average tangential velocity (ms)

V circumferential velocity of check point in check circle

respectively (ms)

N check point total numbers in single check circle

4) Velocity

Velocity measurement at bell throat by Pitot tube was implemented by ANSIHI

method Typical Pitot tube configuration is shown in Fig 28 Eight measuring

points are defined as Fig 29 Time averaged velocity at eight points was obtained

and these eight data were again averaged to give throat sectional spatial velocity

average

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

19

Fig 28 Example of Pitot tube Fig 29 Measuring points at bell throat

224 Test criteria

There is the criteria for satisfactory performance of the sump model test and

summarized in Table 23

1) Free surface and submerged vortices

Free surface vortices entering the pump must be less severe than vortices with

coherent core Dye core vortices may be acceptable only if they occur less than 10

of the time or only for infrequent pump operating conditions

Submerged vortex which starts from the sump walls induce vibration and noise

problems Dye core submerged vortices extending into the pump bell mouth is not

allowed

2) Swirl angles

Swirl angle is used to check the flow rotation in the suction pipe The calculation

method is referred to section 223

Short-term maximum and long-term average values must be less than 5 degrees

Maximum short-term swirl angles up to 7 degrees may be acceptable only if they

occur less than 10 of the time or only for infrequent pump operating conditions

3)Velocity

Each point time-averaged velocity shall not be larger than 10 of sectional average

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

20

Standard deviation of time-varying velocity at each point was obtained and its

value shall not be larger than 10 of the time-averaged velocity at each point

1) Time-averaged

Variation[] = 100acute-

a

a

V

VV aV Averaged at section

2) Time-varying

Standard deviation= )1(

)(1i

2

-

-aring=

n

xxn

i

Variation[] = (Standard deviation X ) x 100

X Averaged at point

Table 23 Permission criteria of ANSIHI

Vortex occurence Swirl angle

Free surface vortex Submerged vortex

Observation

time Above 10 minutes Above 10 minutes

Above 10

minutes

Permission

criteria

Dye-core

Vortex(F3)

below 110 total

time

Dye-core

Vortex(S2)

below 110 total

time

Below

5degConditional

intermittent

below 7deg

225 Remedial measures for problem intakes

Usually in the preliminary test the worst flow condition is simulated And based on

the results remedial measures for problem intakes are discussed in this part

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

21

1) Overall approach flow patterns

The characteristics of flow approaching an intake structure is one of the foremost

considerations for the designer It is difficult to predict the effects of a given set of

flow conditions upstream from an intake structure Fig 210 depicts several typical

approach flow patterns of pump sump with different operating conditions

Fig 210 Examples of approach flow patterns

Generally speaking for the intake problems especially for the multiple pumps

sump there are the main measures using dividing walls placed between the pumps

as a partitioned structure trash racks with elongated bars guide piers or a number

of structural columns at the bay entrance can provide some assistance in controlling

cross-flow baffles flow distributors and screen mesh etc are applied to improve

the flow distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

22

2) Problems around pump intake

a) Free surface vortices

There are a number of factors that have an effect on the formation of surface

vortices To achieve a higher degree of certainty that objectionable surface vortices

do not form modifications can be made to intake structure to allow operation at

practical depths of submergence and the minimum submergence is referred in Table

22 the use of suction umbrellas vertical curtain walls and horizontal gratings

b) Submerged vortices

In the vicinity of intake bell there are the most complicated flow patterns and flow

makes the most changes in direction while maintaining a constant acceleration into

the pump bell to prevent local flow separation turbulence and vortex formation

Fig 211 represents a sample of various devices to inhibit the different submerged

vortex

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

23

Fig 211 Methods to reduce submerged vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

24

c) Preswirl

The approach flow distributions determine whether flow preswirl exists and a

sufficiently laterally skewed approach flow cause rotating around the intake bell

The most effective way of reducing preswirl is to establish a relatively uniform

approach flow within each pump bay by using the baffling schemes controlling

cross-flow and expanding concentrated flow discussed in the above part

To prevent non-uniform velocity distribution swirls and vortices necessary

modifications are conducted such as fore bay length baffle wall distributors

curtain wall mesh screen guide vane floor splitter wall fillet etc As the above

measures applied modification test will be conducted to check all the flow

conditions and compare with the preliminary results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

25

Chapter 3 Methodology

CFD (Computational fluid dynamics) is the technique which can design cycles to

better performance and reduce costs and time In CFD we solve the governing

equations of given physics (may be differential form or integral form) using some

numerical techniques like Finite Difference Method (FDM) Finite Element Method

(FEM) or Finite Volume Method (FVM) Although CFD plays a great part in the

design of turbomachinery CFD is not a replacement for experimental or analytical

approach In this study the numerical analysis was performed by the commercial

code ANSYS CFX 130 All aspects of the CFD study including modelling

meshing solving and post-processing of the results

31 Numerical modeling

ANSYS CFX solves the unsteady Navier-Stokes equations in their conservation

form The instantaneous equation of mass (continuity) in the stationary frame is

expressed as equation

ρ

δ + nabla ∙ (ρU) = 0 (31)

And the instantaneous equation for momentum is expressed as shown in equation

ρ

δ + nabla ∙ (ρU⨂U) = minusnablap + nabla ∙ τ + S (32)

Where τ is the stress tensor ( including both normal and shear components of the

stress)

These instantaneous equations are averaged for turbulent flows leading to additional

terms that need to be solved While the Navier-Stokes equations describe both

laminar and turbulent flows without addition terms realistic flows involve length

scales much smaller than the smallest finite volume mesh A Direct Numerical

Simulation of these flows would require significantly more computing power than

what is available now or in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

26

Therefore much research has been done to predict the effects of turbulence by

using turbulence models These models account for the effects of turbulence

without the use of a very fine mesh or direct numerical simulation

These turbulence models modify the transport equations by adding averaged and

fluctuating components The transport equations are changed to equations (33) and

(34)

ρ

δ + nabla ∙ (ρU) = 0 (33)

ρ

+ nabla ∙ ρ U ⨂U = minusnablap + nabla ∙ τ minus ρu⨂u + S (34)

The mass equation is not changed but the momentum equation contains extra terms

which are the Reynolds stresses ρ u⨂u and the Reynolds flux ρ u ϕ These

Reynolds stresses need to be modeled by additional equations to obtain closure

Obtaining closure implies that there are a sufficient number of equations to solve for

all the unknowns including the Reynolds stresses and Reynolds fluxes

Various turbulence models provide various ways to obtain closure In this

investigation Shear Stress Transport (SST) model and k-ε model were utilized The

advantage of SST model is that combines the advantages of other turbulence models

(the k-ε Wilcox k-ω) The 3 turbulence models will be discussed briefly

32 Turbulence models

Turbulence models are necessary because we cannot afford big enough computers

to directly capture every scale of motion There are three modelling frameworks

Direct Numerical Simulation (DNS) Reynolds-Averaged Navier-Stokes

Equations(RANS) Large Eddy Simulation(LES)

As RANS method is applied in most productionseveral RANS turbulence models

differ in their usage of wall functions the number of additional variables solved for

and what these variables represent

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

27

Two equation turbulence models

Two-equation turbulence models are widely used in CFD as they give a good

compromise between computational power needed and accuracy The term lsquotwo

equationrsquo refers to the fact that these models solve for the velocity and length scales

using separate transport equations The turbulence velocity scale is obtained by

solving the transport equation The turbulent length scale is estimated from two

properties of the turbulence field namely the turbulent kinetic energy and the

dissipation rate The dissipation rate of the kinetic energy is obtained from its

transport equation The most widely used are k-ε and k-ω two equation models In

the next section the k-ε Wilcox k-ω BSL k-ω and SST models will be briefly

discussed

321 k-ε turbulence model

The k-epsilon model solves for two variables k the turbulent kinetic energy and

epsilon the rate of dissipation of kinetic energy Wall functions are used in this

model so the flow in the buffer region is not simulated The k-epsilon model is very

popular for industrial applications due to its good convergence rate and relatively

low memory requirements It does not very accurately compute flow fields that

exhibit adverse pressure gradients strong curvature to the flow or jet flow It does

perform well for external flow problems around complex geometries

The k-ε model introduces k (m2s2) as the turbulence kinetic energy and ε (m2s3) as

the turbulence eddy dissipation The continuity equation remains the same

ρ

+ nabla ∙ (ρ U) = 0 (35)

The momentum equation changes as shown by equation

δρ

δ + nabla ∙ (ρU⨂U) = minusnablapprime + nabla ∙ μ

nablaU + (nablaU) + S (36)

Where SM is the sum of body forces μeff is the effective viscosity accounting for

turbulence and prsquo is the modified pressure The k-ε model uses the concept of eddy

viscosity giving the equation for effective viscosity as shown by equation

μ

= μ + μ (37)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

28

μt is the turbulence viscosity is linked to the turbulence kinetic energy and

dissipation by the equation

μ = Cμρ

ε (38)

Where Cμ is a constant

The values for k and ε come from the differential transport equations for the

turbulence kinetic energy and the turbulence dissipation rate

The turbulence kinetic energy equation is given as equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus ρε (39)

The turbulence dissipation rate is given by equation

(ρε)

+ nabla ∙ (ρUε) = nabla ∙ μ +

μ

σε nablaε +

ε

(Cε (P + Pε ) minus Cε ρε) (310)

Where Cε1 Cε2 σk σε are constants

Pk is the turbulence production due to viscous forces and is modeled by the

equation

P = μ nablaU ∙ nablaU + nablaU minus

nabla ∙ U 3μ

nabla ∙ U + ρk (311)

If a buoyancy term is added to the previous equation if the full buoyancy model is

used

The buoyancy term Pkb is modeled as

P = minusμ

ρσρg ∙ nablaρ (312)

322 Wilcox k-ω turbulence model

The k-omega model is similar to k-epsilon instead however it solves for omega mdash

the specific rate of dissipation of kinetic energy It also uses wall functions and

therefore has comparable memory requirements It has more difficulty converging

and is quite sensitive to the initial guess at the solution Hence the k-epsilon model

is often used first to find an initial condition for solving the k-omega model The k-

omega model is useful in many cases where the k-epsilon model is not accurate

such as internal flows flows that exhibit strong curvature separated flows and jets

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

29

This model has an advantage over the k-ε model where it does not involve complex

linear damping functions for near wall calculations at low Reynolds numbers

The k-ω model assumes that the turbulence viscosity is related to the turbulence

kinetic energy k and the turbulent frequency ω by the equation

μ = ρ

ω (313)

The transport equation for k is given by the equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus βprimeρ k ω (314)

The transport equation for ω is shown as equation

(ρω)

δ + nabla ∙ (ρ U ω) = nabla ∙ μ +

μ

μω

nablaω + αω

P + Pω minus β ρ ω (315)

The production rate of turbulence (Pk) is calculated as shown previously in the k-ε

section

The recommended values for the constants in the above equations are

βprime = 009

α = 59

β = 0075

σ = 2

σω = 2

The Reynolds stress tensor ρ u⨂u is calculated by

minusρ u⨂u = μ nablaU + (nablaU) minus

part ρk + μ

nabla ∙ U (316)

323 Shear stress transport model

The disadvantage of the Wilcox model is the strong sensitivity to free-stream

conditions Therefore a blending of the k-ω model near the surface and the k-ε in

the outer region was made by Menter [23] which resulted in the formulation of the

BSL k-ω turbulence model It consists of a transformation of the k-ε model to a k-ω

formulation and subsequently adding the resulting equations The Wilcox model is

multiplied by a blending function F1 and the transformed k-ε by another function 1-

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

30

F1 F1 is a function of wall distance (being the value of one near the surface and zero

outside the boundary layer) Outside the boundary and on the edge of the boundary

layer the standard k-ε model is used [24]

SST model is a combination of the k-epsilon in the free stream and the k-omega

models near the walls It does not use wall functions and tends to be most accurate

when solving the flow near the wall The SST model does not always converge to

the solution quickly so the k-epsilon or k-omega models are often solved first to

give good initial conditions

The k-ω based Shear Stress Transport (SST) model of Menter was applied for

turbulence treatment The transport equations for the SST model are expressed

below where the turbulent kinetic energy lsquokrsquo and turbulent frequency or dissipation

per unit turbulent kinetic energy are computed by using the following relations

For Turbulence Kinetic Energy

kkPukt

kTkk Ntilde+Ntilde+-=Ntilde+

para

para)[()( nsnwb (317)

where Pk is the production limiter

For Specific Dissipation Rate

ww

swnsnbwaww

ww NtildeNtilde-+Ntilde+Ntilde+-=Ntilde+para

para

1)1(2])[()( 21

22 kFSut

T (318)

The first blending function F1 is calculated from

iumlthorn

iumlyacute

uuml

iumlicirc

iumliacute

igrave

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

uacuteuacuteucirc

ugrave

ecircecirceuml

eacute

dividedivideoslash

oumlccedilccedilegrave

aelig=

4

22

21

4500maxmintanh

yCD

k

yy

kF

kw

wrs

w

n

wb (319)

y is the distance to the nearest wall and v is the kinematic viscosity

divideoslash

oumlccedilegrave

aeligNtildeNtilde= -10

2 101

2max ww

rsww kCDk (320)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

31

And Kinematic eddy viscosity

)max( 21

1

SF

kT

wa

an = (321)

where F2 is a blending function which restricts the limiter to the wall boundary and

S is the invariant measure of the strain rate

uacuteuacute

ucirc

ugrave

ecircecirc

euml

eacute

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

dividedivideoslash

oumlccedilccedilegrave

aelig=

2

22

5002maxtanh

w

n

wb yy

kF (322)

33 Brief introduction of simulation process

All aspects of CFD study including modelling meshing solving and post-

processing of the results are briefly introduced in this chapter

331 Pre-processing technique

The speed and accuracy of simulation results depending fundamentally on meshing

and the ability to prepare model geometry correctly

The steps in pre-processing are explained as follows

1) Domain Definitions

The domain definition depends mainly on the physics involved in the problem and

objective of the problem

a) Geometry creation or geometry import

b) Geometry cleanup

i) Removing the parts not necessary for simulation under consideration

ii) Closing the gaps in the geometry

iii) Removing small surfaces curves to make meshing process simple

2)Grid generation

To solve the governing equations the numerical methods like Finite Difference

Finite Volume or Finite Element Method to divide the domain into small parts and

solve the equations on one part at a time are applied

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 29: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

17

count revolutions As flow swirl is usually unsteady the observation time of swirl

meter reading should be a continuous period of time

Fig 26 Typical swirl meter

(29)

(210)

(211)

Where qV flow rotational speed at swirl meter (ms)

zV average axial velocity at swirl meter (ms)

n Number of revolution per minute (rpm)

d Diameter of pipe at the swirl meter (m)

q flow rate (m3s)

zV

Vqq 1tan -=

)(60

smnd

Vtimestimes

=p

q

)(4

2

smd

qVz

times=

p

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

18

For the swirl angle calculating method with CFD the key point is to obtain the

average tangential velocity so the swirl check circles were created as illustrated in

Fig 27 The swirl check circle is located at the section of 4d height with diameter

of 025d 05d and 075d respectively

Swirl check circle

Fig 27 Schematic view of swirl angle numerical calculation

23

1 7501 501 250

acute++

=aringaringaring

N

i

N

i

NNVNVNV

Vi

q (212)

where

qV average tangential velocity (ms)

V circumferential velocity of check point in check circle

respectively (ms)

N check point total numbers in single check circle

4) Velocity

Velocity measurement at bell throat by Pitot tube was implemented by ANSIHI

method Typical Pitot tube configuration is shown in Fig 28 Eight measuring

points are defined as Fig 29 Time averaged velocity at eight points was obtained

and these eight data were again averaged to give throat sectional spatial velocity

average

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

19

Fig 28 Example of Pitot tube Fig 29 Measuring points at bell throat

224 Test criteria

There is the criteria for satisfactory performance of the sump model test and

summarized in Table 23

1) Free surface and submerged vortices

Free surface vortices entering the pump must be less severe than vortices with

coherent core Dye core vortices may be acceptable only if they occur less than 10

of the time or only for infrequent pump operating conditions

Submerged vortex which starts from the sump walls induce vibration and noise

problems Dye core submerged vortices extending into the pump bell mouth is not

allowed

2) Swirl angles

Swirl angle is used to check the flow rotation in the suction pipe The calculation

method is referred to section 223

Short-term maximum and long-term average values must be less than 5 degrees

Maximum short-term swirl angles up to 7 degrees may be acceptable only if they

occur less than 10 of the time or only for infrequent pump operating conditions

3)Velocity

Each point time-averaged velocity shall not be larger than 10 of sectional average

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

20

Standard deviation of time-varying velocity at each point was obtained and its

value shall not be larger than 10 of the time-averaged velocity at each point

1) Time-averaged

Variation[] = 100acute-

a

a

V

VV aV Averaged at section

2) Time-varying

Standard deviation= )1(

)(1i

2

-

-aring=

n

xxn

i

Variation[] = (Standard deviation X ) x 100

X Averaged at point

Table 23 Permission criteria of ANSIHI

Vortex occurence Swirl angle

Free surface vortex Submerged vortex

Observation

time Above 10 minutes Above 10 minutes

Above 10

minutes

Permission

criteria

Dye-core

Vortex(F3)

below 110 total

time

Dye-core

Vortex(S2)

below 110 total

time

Below

5degConditional

intermittent

below 7deg

225 Remedial measures for problem intakes

Usually in the preliminary test the worst flow condition is simulated And based on

the results remedial measures for problem intakes are discussed in this part

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

21

1) Overall approach flow patterns

The characteristics of flow approaching an intake structure is one of the foremost

considerations for the designer It is difficult to predict the effects of a given set of

flow conditions upstream from an intake structure Fig 210 depicts several typical

approach flow patterns of pump sump with different operating conditions

Fig 210 Examples of approach flow patterns

Generally speaking for the intake problems especially for the multiple pumps

sump there are the main measures using dividing walls placed between the pumps

as a partitioned structure trash racks with elongated bars guide piers or a number

of structural columns at the bay entrance can provide some assistance in controlling

cross-flow baffles flow distributors and screen mesh etc are applied to improve

the flow distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

22

2) Problems around pump intake

a) Free surface vortices

There are a number of factors that have an effect on the formation of surface

vortices To achieve a higher degree of certainty that objectionable surface vortices

do not form modifications can be made to intake structure to allow operation at

practical depths of submergence and the minimum submergence is referred in Table

22 the use of suction umbrellas vertical curtain walls and horizontal gratings

b) Submerged vortices

In the vicinity of intake bell there are the most complicated flow patterns and flow

makes the most changes in direction while maintaining a constant acceleration into

the pump bell to prevent local flow separation turbulence and vortex formation

Fig 211 represents a sample of various devices to inhibit the different submerged

vortex

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

23

Fig 211 Methods to reduce submerged vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

24

c) Preswirl

The approach flow distributions determine whether flow preswirl exists and a

sufficiently laterally skewed approach flow cause rotating around the intake bell

The most effective way of reducing preswirl is to establish a relatively uniform

approach flow within each pump bay by using the baffling schemes controlling

cross-flow and expanding concentrated flow discussed in the above part

To prevent non-uniform velocity distribution swirls and vortices necessary

modifications are conducted such as fore bay length baffle wall distributors

curtain wall mesh screen guide vane floor splitter wall fillet etc As the above

measures applied modification test will be conducted to check all the flow

conditions and compare with the preliminary results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

25

Chapter 3 Methodology

CFD (Computational fluid dynamics) is the technique which can design cycles to

better performance and reduce costs and time In CFD we solve the governing

equations of given physics (may be differential form or integral form) using some

numerical techniques like Finite Difference Method (FDM) Finite Element Method

(FEM) or Finite Volume Method (FVM) Although CFD plays a great part in the

design of turbomachinery CFD is not a replacement for experimental or analytical

approach In this study the numerical analysis was performed by the commercial

code ANSYS CFX 130 All aspects of the CFD study including modelling

meshing solving and post-processing of the results

31 Numerical modeling

ANSYS CFX solves the unsteady Navier-Stokes equations in their conservation

form The instantaneous equation of mass (continuity) in the stationary frame is

expressed as equation

ρ

δ + nabla ∙ (ρU) = 0 (31)

And the instantaneous equation for momentum is expressed as shown in equation

ρ

δ + nabla ∙ (ρU⨂U) = minusnablap + nabla ∙ τ + S (32)

Where τ is the stress tensor ( including both normal and shear components of the

stress)

These instantaneous equations are averaged for turbulent flows leading to additional

terms that need to be solved While the Navier-Stokes equations describe both

laminar and turbulent flows without addition terms realistic flows involve length

scales much smaller than the smallest finite volume mesh A Direct Numerical

Simulation of these flows would require significantly more computing power than

what is available now or in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

26

Therefore much research has been done to predict the effects of turbulence by

using turbulence models These models account for the effects of turbulence

without the use of a very fine mesh or direct numerical simulation

These turbulence models modify the transport equations by adding averaged and

fluctuating components The transport equations are changed to equations (33) and

(34)

ρ

δ + nabla ∙ (ρU) = 0 (33)

ρ

+ nabla ∙ ρ U ⨂U = minusnablap + nabla ∙ τ minus ρu⨂u + S (34)

The mass equation is not changed but the momentum equation contains extra terms

which are the Reynolds stresses ρ u⨂u and the Reynolds flux ρ u ϕ These

Reynolds stresses need to be modeled by additional equations to obtain closure

Obtaining closure implies that there are a sufficient number of equations to solve for

all the unknowns including the Reynolds stresses and Reynolds fluxes

Various turbulence models provide various ways to obtain closure In this

investigation Shear Stress Transport (SST) model and k-ε model were utilized The

advantage of SST model is that combines the advantages of other turbulence models

(the k-ε Wilcox k-ω) The 3 turbulence models will be discussed briefly

32 Turbulence models

Turbulence models are necessary because we cannot afford big enough computers

to directly capture every scale of motion There are three modelling frameworks

Direct Numerical Simulation (DNS) Reynolds-Averaged Navier-Stokes

Equations(RANS) Large Eddy Simulation(LES)

As RANS method is applied in most productionseveral RANS turbulence models

differ in their usage of wall functions the number of additional variables solved for

and what these variables represent

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

27

Two equation turbulence models

Two-equation turbulence models are widely used in CFD as they give a good

compromise between computational power needed and accuracy The term lsquotwo

equationrsquo refers to the fact that these models solve for the velocity and length scales

using separate transport equations The turbulence velocity scale is obtained by

solving the transport equation The turbulent length scale is estimated from two

properties of the turbulence field namely the turbulent kinetic energy and the

dissipation rate The dissipation rate of the kinetic energy is obtained from its

transport equation The most widely used are k-ε and k-ω two equation models In

the next section the k-ε Wilcox k-ω BSL k-ω and SST models will be briefly

discussed

321 k-ε turbulence model

The k-epsilon model solves for two variables k the turbulent kinetic energy and

epsilon the rate of dissipation of kinetic energy Wall functions are used in this

model so the flow in the buffer region is not simulated The k-epsilon model is very

popular for industrial applications due to its good convergence rate and relatively

low memory requirements It does not very accurately compute flow fields that

exhibit adverse pressure gradients strong curvature to the flow or jet flow It does

perform well for external flow problems around complex geometries

The k-ε model introduces k (m2s2) as the turbulence kinetic energy and ε (m2s3) as

the turbulence eddy dissipation The continuity equation remains the same

ρ

+ nabla ∙ (ρ U) = 0 (35)

The momentum equation changes as shown by equation

δρ

δ + nabla ∙ (ρU⨂U) = minusnablapprime + nabla ∙ μ

nablaU + (nablaU) + S (36)

Where SM is the sum of body forces μeff is the effective viscosity accounting for

turbulence and prsquo is the modified pressure The k-ε model uses the concept of eddy

viscosity giving the equation for effective viscosity as shown by equation

μ

= μ + μ (37)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

28

μt is the turbulence viscosity is linked to the turbulence kinetic energy and

dissipation by the equation

μ = Cμρ

ε (38)

Where Cμ is a constant

The values for k and ε come from the differential transport equations for the

turbulence kinetic energy and the turbulence dissipation rate

The turbulence kinetic energy equation is given as equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus ρε (39)

The turbulence dissipation rate is given by equation

(ρε)

+ nabla ∙ (ρUε) = nabla ∙ μ +

μ

σε nablaε +

ε

(Cε (P + Pε ) minus Cε ρε) (310)

Where Cε1 Cε2 σk σε are constants

Pk is the turbulence production due to viscous forces and is modeled by the

equation

P = μ nablaU ∙ nablaU + nablaU minus

nabla ∙ U 3μ

nabla ∙ U + ρk (311)

If a buoyancy term is added to the previous equation if the full buoyancy model is

used

The buoyancy term Pkb is modeled as

P = minusμ

ρσρg ∙ nablaρ (312)

322 Wilcox k-ω turbulence model

The k-omega model is similar to k-epsilon instead however it solves for omega mdash

the specific rate of dissipation of kinetic energy It also uses wall functions and

therefore has comparable memory requirements It has more difficulty converging

and is quite sensitive to the initial guess at the solution Hence the k-epsilon model

is often used first to find an initial condition for solving the k-omega model The k-

omega model is useful in many cases where the k-epsilon model is not accurate

such as internal flows flows that exhibit strong curvature separated flows and jets

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

29

This model has an advantage over the k-ε model where it does not involve complex

linear damping functions for near wall calculations at low Reynolds numbers

The k-ω model assumes that the turbulence viscosity is related to the turbulence

kinetic energy k and the turbulent frequency ω by the equation

μ = ρ

ω (313)

The transport equation for k is given by the equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus βprimeρ k ω (314)

The transport equation for ω is shown as equation

(ρω)

δ + nabla ∙ (ρ U ω) = nabla ∙ μ +

μ

μω

nablaω + αω

P + Pω minus β ρ ω (315)

The production rate of turbulence (Pk) is calculated as shown previously in the k-ε

section

The recommended values for the constants in the above equations are

βprime = 009

α = 59

β = 0075

σ = 2

σω = 2

The Reynolds stress tensor ρ u⨂u is calculated by

minusρ u⨂u = μ nablaU + (nablaU) minus

part ρk + μ

nabla ∙ U (316)

323 Shear stress transport model

The disadvantage of the Wilcox model is the strong sensitivity to free-stream

conditions Therefore a blending of the k-ω model near the surface and the k-ε in

the outer region was made by Menter [23] which resulted in the formulation of the

BSL k-ω turbulence model It consists of a transformation of the k-ε model to a k-ω

formulation and subsequently adding the resulting equations The Wilcox model is

multiplied by a blending function F1 and the transformed k-ε by another function 1-

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

30

F1 F1 is a function of wall distance (being the value of one near the surface and zero

outside the boundary layer) Outside the boundary and on the edge of the boundary

layer the standard k-ε model is used [24]

SST model is a combination of the k-epsilon in the free stream and the k-omega

models near the walls It does not use wall functions and tends to be most accurate

when solving the flow near the wall The SST model does not always converge to

the solution quickly so the k-epsilon or k-omega models are often solved first to

give good initial conditions

The k-ω based Shear Stress Transport (SST) model of Menter was applied for

turbulence treatment The transport equations for the SST model are expressed

below where the turbulent kinetic energy lsquokrsquo and turbulent frequency or dissipation

per unit turbulent kinetic energy are computed by using the following relations

For Turbulence Kinetic Energy

kkPukt

kTkk Ntilde+Ntilde+-=Ntilde+

para

para)[()( nsnwb (317)

where Pk is the production limiter

For Specific Dissipation Rate

ww

swnsnbwaww

ww NtildeNtilde-+Ntilde+Ntilde+-=Ntilde+para

para

1)1(2])[()( 21

22 kFSut

T (318)

The first blending function F1 is calculated from

iumlthorn

iumlyacute

uuml

iumlicirc

iumliacute

igrave

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

uacuteuacuteucirc

ugrave

ecircecirceuml

eacute

dividedivideoslash

oumlccedilccedilegrave

aelig=

4

22

21

4500maxmintanh

yCD

k

yy

kF

kw

wrs

w

n

wb (319)

y is the distance to the nearest wall and v is the kinematic viscosity

divideoslash

oumlccedilegrave

aeligNtildeNtilde= -10

2 101

2max ww

rsww kCDk (320)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

31

And Kinematic eddy viscosity

)max( 21

1

SF

kT

wa

an = (321)

where F2 is a blending function which restricts the limiter to the wall boundary and

S is the invariant measure of the strain rate

uacuteuacute

ucirc

ugrave

ecircecirc

euml

eacute

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

dividedivideoslash

oumlccedilccedilegrave

aelig=

2

22

5002maxtanh

w

n

wb yy

kF (322)

33 Brief introduction of simulation process

All aspects of CFD study including modelling meshing solving and post-

processing of the results are briefly introduced in this chapter

331 Pre-processing technique

The speed and accuracy of simulation results depending fundamentally on meshing

and the ability to prepare model geometry correctly

The steps in pre-processing are explained as follows

1) Domain Definitions

The domain definition depends mainly on the physics involved in the problem and

objective of the problem

a) Geometry creation or geometry import

b) Geometry cleanup

i) Removing the parts not necessary for simulation under consideration

ii) Closing the gaps in the geometry

iii) Removing small surfaces curves to make meshing process simple

2)Grid generation

To solve the governing equations the numerical methods like Finite Difference

Finite Volume or Finite Element Method to divide the domain into small parts and

solve the equations on one part at a time are applied

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 30: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

18

For the swirl angle calculating method with CFD the key point is to obtain the

average tangential velocity so the swirl check circles were created as illustrated in

Fig 27 The swirl check circle is located at the section of 4d height with diameter

of 025d 05d and 075d respectively

Swirl check circle

Fig 27 Schematic view of swirl angle numerical calculation

23

1 7501 501 250

acute++

=aringaringaring

N

i

N

i

NNVNVNV

Vi

q (212)

where

qV average tangential velocity (ms)

V circumferential velocity of check point in check circle

respectively (ms)

N check point total numbers in single check circle

4) Velocity

Velocity measurement at bell throat by Pitot tube was implemented by ANSIHI

method Typical Pitot tube configuration is shown in Fig 28 Eight measuring

points are defined as Fig 29 Time averaged velocity at eight points was obtained

and these eight data were again averaged to give throat sectional spatial velocity

average

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

19

Fig 28 Example of Pitot tube Fig 29 Measuring points at bell throat

224 Test criteria

There is the criteria for satisfactory performance of the sump model test and

summarized in Table 23

1) Free surface and submerged vortices

Free surface vortices entering the pump must be less severe than vortices with

coherent core Dye core vortices may be acceptable only if they occur less than 10

of the time or only for infrequent pump operating conditions

Submerged vortex which starts from the sump walls induce vibration and noise

problems Dye core submerged vortices extending into the pump bell mouth is not

allowed

2) Swirl angles

Swirl angle is used to check the flow rotation in the suction pipe The calculation

method is referred to section 223

Short-term maximum and long-term average values must be less than 5 degrees

Maximum short-term swirl angles up to 7 degrees may be acceptable only if they

occur less than 10 of the time or only for infrequent pump operating conditions

3)Velocity

Each point time-averaged velocity shall not be larger than 10 of sectional average

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

20

Standard deviation of time-varying velocity at each point was obtained and its

value shall not be larger than 10 of the time-averaged velocity at each point

1) Time-averaged

Variation[] = 100acute-

a

a

V

VV aV Averaged at section

2) Time-varying

Standard deviation= )1(

)(1i

2

-

-aring=

n

xxn

i

Variation[] = (Standard deviation X ) x 100

X Averaged at point

Table 23 Permission criteria of ANSIHI

Vortex occurence Swirl angle

Free surface vortex Submerged vortex

Observation

time Above 10 minutes Above 10 minutes

Above 10

minutes

Permission

criteria

Dye-core

Vortex(F3)

below 110 total

time

Dye-core

Vortex(S2)

below 110 total

time

Below

5degConditional

intermittent

below 7deg

225 Remedial measures for problem intakes

Usually in the preliminary test the worst flow condition is simulated And based on

the results remedial measures for problem intakes are discussed in this part

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

21

1) Overall approach flow patterns

The characteristics of flow approaching an intake structure is one of the foremost

considerations for the designer It is difficult to predict the effects of a given set of

flow conditions upstream from an intake structure Fig 210 depicts several typical

approach flow patterns of pump sump with different operating conditions

Fig 210 Examples of approach flow patterns

Generally speaking for the intake problems especially for the multiple pumps

sump there are the main measures using dividing walls placed between the pumps

as a partitioned structure trash racks with elongated bars guide piers or a number

of structural columns at the bay entrance can provide some assistance in controlling

cross-flow baffles flow distributors and screen mesh etc are applied to improve

the flow distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

22

2) Problems around pump intake

a) Free surface vortices

There are a number of factors that have an effect on the formation of surface

vortices To achieve a higher degree of certainty that objectionable surface vortices

do not form modifications can be made to intake structure to allow operation at

practical depths of submergence and the minimum submergence is referred in Table

22 the use of suction umbrellas vertical curtain walls and horizontal gratings

b) Submerged vortices

In the vicinity of intake bell there are the most complicated flow patterns and flow

makes the most changes in direction while maintaining a constant acceleration into

the pump bell to prevent local flow separation turbulence and vortex formation

Fig 211 represents a sample of various devices to inhibit the different submerged

vortex

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

23

Fig 211 Methods to reduce submerged vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

24

c) Preswirl

The approach flow distributions determine whether flow preswirl exists and a

sufficiently laterally skewed approach flow cause rotating around the intake bell

The most effective way of reducing preswirl is to establish a relatively uniform

approach flow within each pump bay by using the baffling schemes controlling

cross-flow and expanding concentrated flow discussed in the above part

To prevent non-uniform velocity distribution swirls and vortices necessary

modifications are conducted such as fore bay length baffle wall distributors

curtain wall mesh screen guide vane floor splitter wall fillet etc As the above

measures applied modification test will be conducted to check all the flow

conditions and compare with the preliminary results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

25

Chapter 3 Methodology

CFD (Computational fluid dynamics) is the technique which can design cycles to

better performance and reduce costs and time In CFD we solve the governing

equations of given physics (may be differential form or integral form) using some

numerical techniques like Finite Difference Method (FDM) Finite Element Method

(FEM) or Finite Volume Method (FVM) Although CFD plays a great part in the

design of turbomachinery CFD is not a replacement for experimental or analytical

approach In this study the numerical analysis was performed by the commercial

code ANSYS CFX 130 All aspects of the CFD study including modelling

meshing solving and post-processing of the results

31 Numerical modeling

ANSYS CFX solves the unsteady Navier-Stokes equations in their conservation

form The instantaneous equation of mass (continuity) in the stationary frame is

expressed as equation

ρ

δ + nabla ∙ (ρU) = 0 (31)

And the instantaneous equation for momentum is expressed as shown in equation

ρ

δ + nabla ∙ (ρU⨂U) = minusnablap + nabla ∙ τ + S (32)

Where τ is the stress tensor ( including both normal and shear components of the

stress)

These instantaneous equations are averaged for turbulent flows leading to additional

terms that need to be solved While the Navier-Stokes equations describe both

laminar and turbulent flows without addition terms realistic flows involve length

scales much smaller than the smallest finite volume mesh A Direct Numerical

Simulation of these flows would require significantly more computing power than

what is available now or in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

26

Therefore much research has been done to predict the effects of turbulence by

using turbulence models These models account for the effects of turbulence

without the use of a very fine mesh or direct numerical simulation

These turbulence models modify the transport equations by adding averaged and

fluctuating components The transport equations are changed to equations (33) and

(34)

ρ

δ + nabla ∙ (ρU) = 0 (33)

ρ

+ nabla ∙ ρ U ⨂U = minusnablap + nabla ∙ τ minus ρu⨂u + S (34)

The mass equation is not changed but the momentum equation contains extra terms

which are the Reynolds stresses ρ u⨂u and the Reynolds flux ρ u ϕ These

Reynolds stresses need to be modeled by additional equations to obtain closure

Obtaining closure implies that there are a sufficient number of equations to solve for

all the unknowns including the Reynolds stresses and Reynolds fluxes

Various turbulence models provide various ways to obtain closure In this

investigation Shear Stress Transport (SST) model and k-ε model were utilized The

advantage of SST model is that combines the advantages of other turbulence models

(the k-ε Wilcox k-ω) The 3 turbulence models will be discussed briefly

32 Turbulence models

Turbulence models are necessary because we cannot afford big enough computers

to directly capture every scale of motion There are three modelling frameworks

Direct Numerical Simulation (DNS) Reynolds-Averaged Navier-Stokes

Equations(RANS) Large Eddy Simulation(LES)

As RANS method is applied in most productionseveral RANS turbulence models

differ in their usage of wall functions the number of additional variables solved for

and what these variables represent

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

27

Two equation turbulence models

Two-equation turbulence models are widely used in CFD as they give a good

compromise between computational power needed and accuracy The term lsquotwo

equationrsquo refers to the fact that these models solve for the velocity and length scales

using separate transport equations The turbulence velocity scale is obtained by

solving the transport equation The turbulent length scale is estimated from two

properties of the turbulence field namely the turbulent kinetic energy and the

dissipation rate The dissipation rate of the kinetic energy is obtained from its

transport equation The most widely used are k-ε and k-ω two equation models In

the next section the k-ε Wilcox k-ω BSL k-ω and SST models will be briefly

discussed

321 k-ε turbulence model

The k-epsilon model solves for two variables k the turbulent kinetic energy and

epsilon the rate of dissipation of kinetic energy Wall functions are used in this

model so the flow in the buffer region is not simulated The k-epsilon model is very

popular for industrial applications due to its good convergence rate and relatively

low memory requirements It does not very accurately compute flow fields that

exhibit adverse pressure gradients strong curvature to the flow or jet flow It does

perform well for external flow problems around complex geometries

The k-ε model introduces k (m2s2) as the turbulence kinetic energy and ε (m2s3) as

the turbulence eddy dissipation The continuity equation remains the same

ρ

+ nabla ∙ (ρ U) = 0 (35)

The momentum equation changes as shown by equation

δρ

δ + nabla ∙ (ρU⨂U) = minusnablapprime + nabla ∙ μ

nablaU + (nablaU) + S (36)

Where SM is the sum of body forces μeff is the effective viscosity accounting for

turbulence and prsquo is the modified pressure The k-ε model uses the concept of eddy

viscosity giving the equation for effective viscosity as shown by equation

μ

= μ + μ (37)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

28

μt is the turbulence viscosity is linked to the turbulence kinetic energy and

dissipation by the equation

μ = Cμρ

ε (38)

Where Cμ is a constant

The values for k and ε come from the differential transport equations for the

turbulence kinetic energy and the turbulence dissipation rate

The turbulence kinetic energy equation is given as equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus ρε (39)

The turbulence dissipation rate is given by equation

(ρε)

+ nabla ∙ (ρUε) = nabla ∙ μ +

μ

σε nablaε +

ε

(Cε (P + Pε ) minus Cε ρε) (310)

Where Cε1 Cε2 σk σε are constants

Pk is the turbulence production due to viscous forces and is modeled by the

equation

P = μ nablaU ∙ nablaU + nablaU minus

nabla ∙ U 3μ

nabla ∙ U + ρk (311)

If a buoyancy term is added to the previous equation if the full buoyancy model is

used

The buoyancy term Pkb is modeled as

P = minusμ

ρσρg ∙ nablaρ (312)

322 Wilcox k-ω turbulence model

The k-omega model is similar to k-epsilon instead however it solves for omega mdash

the specific rate of dissipation of kinetic energy It also uses wall functions and

therefore has comparable memory requirements It has more difficulty converging

and is quite sensitive to the initial guess at the solution Hence the k-epsilon model

is often used first to find an initial condition for solving the k-omega model The k-

omega model is useful in many cases where the k-epsilon model is not accurate

such as internal flows flows that exhibit strong curvature separated flows and jets

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

29

This model has an advantage over the k-ε model where it does not involve complex

linear damping functions for near wall calculations at low Reynolds numbers

The k-ω model assumes that the turbulence viscosity is related to the turbulence

kinetic energy k and the turbulent frequency ω by the equation

μ = ρ

ω (313)

The transport equation for k is given by the equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus βprimeρ k ω (314)

The transport equation for ω is shown as equation

(ρω)

δ + nabla ∙ (ρ U ω) = nabla ∙ μ +

μ

μω

nablaω + αω

P + Pω minus β ρ ω (315)

The production rate of turbulence (Pk) is calculated as shown previously in the k-ε

section

The recommended values for the constants in the above equations are

βprime = 009

α = 59

β = 0075

σ = 2

σω = 2

The Reynolds stress tensor ρ u⨂u is calculated by

minusρ u⨂u = μ nablaU + (nablaU) minus

part ρk + μ

nabla ∙ U (316)

323 Shear stress transport model

The disadvantage of the Wilcox model is the strong sensitivity to free-stream

conditions Therefore a blending of the k-ω model near the surface and the k-ε in

the outer region was made by Menter [23] which resulted in the formulation of the

BSL k-ω turbulence model It consists of a transformation of the k-ε model to a k-ω

formulation and subsequently adding the resulting equations The Wilcox model is

multiplied by a blending function F1 and the transformed k-ε by another function 1-

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

30

F1 F1 is a function of wall distance (being the value of one near the surface and zero

outside the boundary layer) Outside the boundary and on the edge of the boundary

layer the standard k-ε model is used [24]

SST model is a combination of the k-epsilon in the free stream and the k-omega

models near the walls It does not use wall functions and tends to be most accurate

when solving the flow near the wall The SST model does not always converge to

the solution quickly so the k-epsilon or k-omega models are often solved first to

give good initial conditions

The k-ω based Shear Stress Transport (SST) model of Menter was applied for

turbulence treatment The transport equations for the SST model are expressed

below where the turbulent kinetic energy lsquokrsquo and turbulent frequency or dissipation

per unit turbulent kinetic energy are computed by using the following relations

For Turbulence Kinetic Energy

kkPukt

kTkk Ntilde+Ntilde+-=Ntilde+

para

para)[()( nsnwb (317)

where Pk is the production limiter

For Specific Dissipation Rate

ww

swnsnbwaww

ww NtildeNtilde-+Ntilde+Ntilde+-=Ntilde+para

para

1)1(2])[()( 21

22 kFSut

T (318)

The first blending function F1 is calculated from

iumlthorn

iumlyacute

uuml

iumlicirc

iumliacute

igrave

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

uacuteuacuteucirc

ugrave

ecircecirceuml

eacute

dividedivideoslash

oumlccedilccedilegrave

aelig=

4

22

21

4500maxmintanh

yCD

k

yy

kF

kw

wrs

w

n

wb (319)

y is the distance to the nearest wall and v is the kinematic viscosity

divideoslash

oumlccedilegrave

aeligNtildeNtilde= -10

2 101

2max ww

rsww kCDk (320)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

31

And Kinematic eddy viscosity

)max( 21

1

SF

kT

wa

an = (321)

where F2 is a blending function which restricts the limiter to the wall boundary and

S is the invariant measure of the strain rate

uacuteuacute

ucirc

ugrave

ecircecirc

euml

eacute

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

dividedivideoslash

oumlccedilccedilegrave

aelig=

2

22

5002maxtanh

w

n

wb yy

kF (322)

33 Brief introduction of simulation process

All aspects of CFD study including modelling meshing solving and post-

processing of the results are briefly introduced in this chapter

331 Pre-processing technique

The speed and accuracy of simulation results depending fundamentally on meshing

and the ability to prepare model geometry correctly

The steps in pre-processing are explained as follows

1) Domain Definitions

The domain definition depends mainly on the physics involved in the problem and

objective of the problem

a) Geometry creation or geometry import

b) Geometry cleanup

i) Removing the parts not necessary for simulation under consideration

ii) Closing the gaps in the geometry

iii) Removing small surfaces curves to make meshing process simple

2)Grid generation

To solve the governing equations the numerical methods like Finite Difference

Finite Volume or Finite Element Method to divide the domain into small parts and

solve the equations on one part at a time are applied

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 31: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

19

Fig 28 Example of Pitot tube Fig 29 Measuring points at bell throat

224 Test criteria

There is the criteria for satisfactory performance of the sump model test and

summarized in Table 23

1) Free surface and submerged vortices

Free surface vortices entering the pump must be less severe than vortices with

coherent core Dye core vortices may be acceptable only if they occur less than 10

of the time or only for infrequent pump operating conditions

Submerged vortex which starts from the sump walls induce vibration and noise

problems Dye core submerged vortices extending into the pump bell mouth is not

allowed

2) Swirl angles

Swirl angle is used to check the flow rotation in the suction pipe The calculation

method is referred to section 223

Short-term maximum and long-term average values must be less than 5 degrees

Maximum short-term swirl angles up to 7 degrees may be acceptable only if they

occur less than 10 of the time or only for infrequent pump operating conditions

3)Velocity

Each point time-averaged velocity shall not be larger than 10 of sectional average

velocity

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

20

Standard deviation of time-varying velocity at each point was obtained and its

value shall not be larger than 10 of the time-averaged velocity at each point

1) Time-averaged

Variation[] = 100acute-

a

a

V

VV aV Averaged at section

2) Time-varying

Standard deviation= )1(

)(1i

2

-

-aring=

n

xxn

i

Variation[] = (Standard deviation X ) x 100

X Averaged at point

Table 23 Permission criteria of ANSIHI

Vortex occurence Swirl angle

Free surface vortex Submerged vortex

Observation

time Above 10 minutes Above 10 minutes

Above 10

minutes

Permission

criteria

Dye-core

Vortex(F3)

below 110 total

time

Dye-core

Vortex(S2)

below 110 total

time

Below

5degConditional

intermittent

below 7deg

225 Remedial measures for problem intakes

Usually in the preliminary test the worst flow condition is simulated And based on

the results remedial measures for problem intakes are discussed in this part

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

21

1) Overall approach flow patterns

The characteristics of flow approaching an intake structure is one of the foremost

considerations for the designer It is difficult to predict the effects of a given set of

flow conditions upstream from an intake structure Fig 210 depicts several typical

approach flow patterns of pump sump with different operating conditions

Fig 210 Examples of approach flow patterns

Generally speaking for the intake problems especially for the multiple pumps

sump there are the main measures using dividing walls placed between the pumps

as a partitioned structure trash racks with elongated bars guide piers or a number

of structural columns at the bay entrance can provide some assistance in controlling

cross-flow baffles flow distributors and screen mesh etc are applied to improve

the flow distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

22

2) Problems around pump intake

a) Free surface vortices

There are a number of factors that have an effect on the formation of surface

vortices To achieve a higher degree of certainty that objectionable surface vortices

do not form modifications can be made to intake structure to allow operation at

practical depths of submergence and the minimum submergence is referred in Table

22 the use of suction umbrellas vertical curtain walls and horizontal gratings

b) Submerged vortices

In the vicinity of intake bell there are the most complicated flow patterns and flow

makes the most changes in direction while maintaining a constant acceleration into

the pump bell to prevent local flow separation turbulence and vortex formation

Fig 211 represents a sample of various devices to inhibit the different submerged

vortex

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

23

Fig 211 Methods to reduce submerged vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

24

c) Preswirl

The approach flow distributions determine whether flow preswirl exists and a

sufficiently laterally skewed approach flow cause rotating around the intake bell

The most effective way of reducing preswirl is to establish a relatively uniform

approach flow within each pump bay by using the baffling schemes controlling

cross-flow and expanding concentrated flow discussed in the above part

To prevent non-uniform velocity distribution swirls and vortices necessary

modifications are conducted such as fore bay length baffle wall distributors

curtain wall mesh screen guide vane floor splitter wall fillet etc As the above

measures applied modification test will be conducted to check all the flow

conditions and compare with the preliminary results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

25

Chapter 3 Methodology

CFD (Computational fluid dynamics) is the technique which can design cycles to

better performance and reduce costs and time In CFD we solve the governing

equations of given physics (may be differential form or integral form) using some

numerical techniques like Finite Difference Method (FDM) Finite Element Method

(FEM) or Finite Volume Method (FVM) Although CFD plays a great part in the

design of turbomachinery CFD is not a replacement for experimental or analytical

approach In this study the numerical analysis was performed by the commercial

code ANSYS CFX 130 All aspects of the CFD study including modelling

meshing solving and post-processing of the results

31 Numerical modeling

ANSYS CFX solves the unsteady Navier-Stokes equations in their conservation

form The instantaneous equation of mass (continuity) in the stationary frame is

expressed as equation

ρ

δ + nabla ∙ (ρU) = 0 (31)

And the instantaneous equation for momentum is expressed as shown in equation

ρ

δ + nabla ∙ (ρU⨂U) = minusnablap + nabla ∙ τ + S (32)

Where τ is the stress tensor ( including both normal and shear components of the

stress)

These instantaneous equations are averaged for turbulent flows leading to additional

terms that need to be solved While the Navier-Stokes equations describe both

laminar and turbulent flows without addition terms realistic flows involve length

scales much smaller than the smallest finite volume mesh A Direct Numerical

Simulation of these flows would require significantly more computing power than

what is available now or in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

26

Therefore much research has been done to predict the effects of turbulence by

using turbulence models These models account for the effects of turbulence

without the use of a very fine mesh or direct numerical simulation

These turbulence models modify the transport equations by adding averaged and

fluctuating components The transport equations are changed to equations (33) and

(34)

ρ

δ + nabla ∙ (ρU) = 0 (33)

ρ

+ nabla ∙ ρ U ⨂U = minusnablap + nabla ∙ τ minus ρu⨂u + S (34)

The mass equation is not changed but the momentum equation contains extra terms

which are the Reynolds stresses ρ u⨂u and the Reynolds flux ρ u ϕ These

Reynolds stresses need to be modeled by additional equations to obtain closure

Obtaining closure implies that there are a sufficient number of equations to solve for

all the unknowns including the Reynolds stresses and Reynolds fluxes

Various turbulence models provide various ways to obtain closure In this

investigation Shear Stress Transport (SST) model and k-ε model were utilized The

advantage of SST model is that combines the advantages of other turbulence models

(the k-ε Wilcox k-ω) The 3 turbulence models will be discussed briefly

32 Turbulence models

Turbulence models are necessary because we cannot afford big enough computers

to directly capture every scale of motion There are three modelling frameworks

Direct Numerical Simulation (DNS) Reynolds-Averaged Navier-Stokes

Equations(RANS) Large Eddy Simulation(LES)

As RANS method is applied in most productionseveral RANS turbulence models

differ in their usage of wall functions the number of additional variables solved for

and what these variables represent

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

27

Two equation turbulence models

Two-equation turbulence models are widely used in CFD as they give a good

compromise between computational power needed and accuracy The term lsquotwo

equationrsquo refers to the fact that these models solve for the velocity and length scales

using separate transport equations The turbulence velocity scale is obtained by

solving the transport equation The turbulent length scale is estimated from two

properties of the turbulence field namely the turbulent kinetic energy and the

dissipation rate The dissipation rate of the kinetic energy is obtained from its

transport equation The most widely used are k-ε and k-ω two equation models In

the next section the k-ε Wilcox k-ω BSL k-ω and SST models will be briefly

discussed

321 k-ε turbulence model

The k-epsilon model solves for two variables k the turbulent kinetic energy and

epsilon the rate of dissipation of kinetic energy Wall functions are used in this

model so the flow in the buffer region is not simulated The k-epsilon model is very

popular for industrial applications due to its good convergence rate and relatively

low memory requirements It does not very accurately compute flow fields that

exhibit adverse pressure gradients strong curvature to the flow or jet flow It does

perform well for external flow problems around complex geometries

The k-ε model introduces k (m2s2) as the turbulence kinetic energy and ε (m2s3) as

the turbulence eddy dissipation The continuity equation remains the same

ρ

+ nabla ∙ (ρ U) = 0 (35)

The momentum equation changes as shown by equation

δρ

δ + nabla ∙ (ρU⨂U) = minusnablapprime + nabla ∙ μ

nablaU + (nablaU) + S (36)

Where SM is the sum of body forces μeff is the effective viscosity accounting for

turbulence and prsquo is the modified pressure The k-ε model uses the concept of eddy

viscosity giving the equation for effective viscosity as shown by equation

μ

= μ + μ (37)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

28

μt is the turbulence viscosity is linked to the turbulence kinetic energy and

dissipation by the equation

μ = Cμρ

ε (38)

Where Cμ is a constant

The values for k and ε come from the differential transport equations for the

turbulence kinetic energy and the turbulence dissipation rate

The turbulence kinetic energy equation is given as equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus ρε (39)

The turbulence dissipation rate is given by equation

(ρε)

+ nabla ∙ (ρUε) = nabla ∙ μ +

μ

σε nablaε +

ε

(Cε (P + Pε ) minus Cε ρε) (310)

Where Cε1 Cε2 σk σε are constants

Pk is the turbulence production due to viscous forces and is modeled by the

equation

P = μ nablaU ∙ nablaU + nablaU minus

nabla ∙ U 3μ

nabla ∙ U + ρk (311)

If a buoyancy term is added to the previous equation if the full buoyancy model is

used

The buoyancy term Pkb is modeled as

P = minusμ

ρσρg ∙ nablaρ (312)

322 Wilcox k-ω turbulence model

The k-omega model is similar to k-epsilon instead however it solves for omega mdash

the specific rate of dissipation of kinetic energy It also uses wall functions and

therefore has comparable memory requirements It has more difficulty converging

and is quite sensitive to the initial guess at the solution Hence the k-epsilon model

is often used first to find an initial condition for solving the k-omega model The k-

omega model is useful in many cases where the k-epsilon model is not accurate

such as internal flows flows that exhibit strong curvature separated flows and jets

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

29

This model has an advantage over the k-ε model where it does not involve complex

linear damping functions for near wall calculations at low Reynolds numbers

The k-ω model assumes that the turbulence viscosity is related to the turbulence

kinetic energy k and the turbulent frequency ω by the equation

μ = ρ

ω (313)

The transport equation for k is given by the equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus βprimeρ k ω (314)

The transport equation for ω is shown as equation

(ρω)

δ + nabla ∙ (ρ U ω) = nabla ∙ μ +

μ

μω

nablaω + αω

P + Pω minus β ρ ω (315)

The production rate of turbulence (Pk) is calculated as shown previously in the k-ε

section

The recommended values for the constants in the above equations are

βprime = 009

α = 59

β = 0075

σ = 2

σω = 2

The Reynolds stress tensor ρ u⨂u is calculated by

minusρ u⨂u = μ nablaU + (nablaU) minus

part ρk + μ

nabla ∙ U (316)

323 Shear stress transport model

The disadvantage of the Wilcox model is the strong sensitivity to free-stream

conditions Therefore a blending of the k-ω model near the surface and the k-ε in

the outer region was made by Menter [23] which resulted in the formulation of the

BSL k-ω turbulence model It consists of a transformation of the k-ε model to a k-ω

formulation and subsequently adding the resulting equations The Wilcox model is

multiplied by a blending function F1 and the transformed k-ε by another function 1-

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

30

F1 F1 is a function of wall distance (being the value of one near the surface and zero

outside the boundary layer) Outside the boundary and on the edge of the boundary

layer the standard k-ε model is used [24]

SST model is a combination of the k-epsilon in the free stream and the k-omega

models near the walls It does not use wall functions and tends to be most accurate

when solving the flow near the wall The SST model does not always converge to

the solution quickly so the k-epsilon or k-omega models are often solved first to

give good initial conditions

The k-ω based Shear Stress Transport (SST) model of Menter was applied for

turbulence treatment The transport equations for the SST model are expressed

below where the turbulent kinetic energy lsquokrsquo and turbulent frequency or dissipation

per unit turbulent kinetic energy are computed by using the following relations

For Turbulence Kinetic Energy

kkPukt

kTkk Ntilde+Ntilde+-=Ntilde+

para

para)[()( nsnwb (317)

where Pk is the production limiter

For Specific Dissipation Rate

ww

swnsnbwaww

ww NtildeNtilde-+Ntilde+Ntilde+-=Ntilde+para

para

1)1(2])[()( 21

22 kFSut

T (318)

The first blending function F1 is calculated from

iumlthorn

iumlyacute

uuml

iumlicirc

iumliacute

igrave

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

uacuteuacuteucirc

ugrave

ecircecirceuml

eacute

dividedivideoslash

oumlccedilccedilegrave

aelig=

4

22

21

4500maxmintanh

yCD

k

yy

kF

kw

wrs

w

n

wb (319)

y is the distance to the nearest wall and v is the kinematic viscosity

divideoslash

oumlccedilegrave

aeligNtildeNtilde= -10

2 101

2max ww

rsww kCDk (320)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

31

And Kinematic eddy viscosity

)max( 21

1

SF

kT

wa

an = (321)

where F2 is a blending function which restricts the limiter to the wall boundary and

S is the invariant measure of the strain rate

uacuteuacute

ucirc

ugrave

ecircecirc

euml

eacute

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

dividedivideoslash

oumlccedilccedilegrave

aelig=

2

22

5002maxtanh

w

n

wb yy

kF (322)

33 Brief introduction of simulation process

All aspects of CFD study including modelling meshing solving and post-

processing of the results are briefly introduced in this chapter

331 Pre-processing technique

The speed and accuracy of simulation results depending fundamentally on meshing

and the ability to prepare model geometry correctly

The steps in pre-processing are explained as follows

1) Domain Definitions

The domain definition depends mainly on the physics involved in the problem and

objective of the problem

a) Geometry creation or geometry import

b) Geometry cleanup

i) Removing the parts not necessary for simulation under consideration

ii) Closing the gaps in the geometry

iii) Removing small surfaces curves to make meshing process simple

2)Grid generation

To solve the governing equations the numerical methods like Finite Difference

Finite Volume or Finite Element Method to divide the domain into small parts and

solve the equations on one part at a time are applied

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 32: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

20

Standard deviation of time-varying velocity at each point was obtained and its

value shall not be larger than 10 of the time-averaged velocity at each point

1) Time-averaged

Variation[] = 100acute-

a

a

V

VV aV Averaged at section

2) Time-varying

Standard deviation= )1(

)(1i

2

-

-aring=

n

xxn

i

Variation[] = (Standard deviation X ) x 100

X Averaged at point

Table 23 Permission criteria of ANSIHI

Vortex occurence Swirl angle

Free surface vortex Submerged vortex

Observation

time Above 10 minutes Above 10 minutes

Above 10

minutes

Permission

criteria

Dye-core

Vortex(F3)

below 110 total

time

Dye-core

Vortex(S2)

below 110 total

time

Below

5degConditional

intermittent

below 7deg

225 Remedial measures for problem intakes

Usually in the preliminary test the worst flow condition is simulated And based on

the results remedial measures for problem intakes are discussed in this part

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

21

1) Overall approach flow patterns

The characteristics of flow approaching an intake structure is one of the foremost

considerations for the designer It is difficult to predict the effects of a given set of

flow conditions upstream from an intake structure Fig 210 depicts several typical

approach flow patterns of pump sump with different operating conditions

Fig 210 Examples of approach flow patterns

Generally speaking for the intake problems especially for the multiple pumps

sump there are the main measures using dividing walls placed between the pumps

as a partitioned structure trash racks with elongated bars guide piers or a number

of structural columns at the bay entrance can provide some assistance in controlling

cross-flow baffles flow distributors and screen mesh etc are applied to improve

the flow distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

22

2) Problems around pump intake

a) Free surface vortices

There are a number of factors that have an effect on the formation of surface

vortices To achieve a higher degree of certainty that objectionable surface vortices

do not form modifications can be made to intake structure to allow operation at

practical depths of submergence and the minimum submergence is referred in Table

22 the use of suction umbrellas vertical curtain walls and horizontal gratings

b) Submerged vortices

In the vicinity of intake bell there are the most complicated flow patterns and flow

makes the most changes in direction while maintaining a constant acceleration into

the pump bell to prevent local flow separation turbulence and vortex formation

Fig 211 represents a sample of various devices to inhibit the different submerged

vortex

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

23

Fig 211 Methods to reduce submerged vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

24

c) Preswirl

The approach flow distributions determine whether flow preswirl exists and a

sufficiently laterally skewed approach flow cause rotating around the intake bell

The most effective way of reducing preswirl is to establish a relatively uniform

approach flow within each pump bay by using the baffling schemes controlling

cross-flow and expanding concentrated flow discussed in the above part

To prevent non-uniform velocity distribution swirls and vortices necessary

modifications are conducted such as fore bay length baffle wall distributors

curtain wall mesh screen guide vane floor splitter wall fillet etc As the above

measures applied modification test will be conducted to check all the flow

conditions and compare with the preliminary results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

25

Chapter 3 Methodology

CFD (Computational fluid dynamics) is the technique which can design cycles to

better performance and reduce costs and time In CFD we solve the governing

equations of given physics (may be differential form or integral form) using some

numerical techniques like Finite Difference Method (FDM) Finite Element Method

(FEM) or Finite Volume Method (FVM) Although CFD plays a great part in the

design of turbomachinery CFD is not a replacement for experimental or analytical

approach In this study the numerical analysis was performed by the commercial

code ANSYS CFX 130 All aspects of the CFD study including modelling

meshing solving and post-processing of the results

31 Numerical modeling

ANSYS CFX solves the unsteady Navier-Stokes equations in their conservation

form The instantaneous equation of mass (continuity) in the stationary frame is

expressed as equation

ρ

δ + nabla ∙ (ρU) = 0 (31)

And the instantaneous equation for momentum is expressed as shown in equation

ρ

δ + nabla ∙ (ρU⨂U) = minusnablap + nabla ∙ τ + S (32)

Where τ is the stress tensor ( including both normal and shear components of the

stress)

These instantaneous equations are averaged for turbulent flows leading to additional

terms that need to be solved While the Navier-Stokes equations describe both

laminar and turbulent flows without addition terms realistic flows involve length

scales much smaller than the smallest finite volume mesh A Direct Numerical

Simulation of these flows would require significantly more computing power than

what is available now or in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

26

Therefore much research has been done to predict the effects of turbulence by

using turbulence models These models account for the effects of turbulence

without the use of a very fine mesh or direct numerical simulation

These turbulence models modify the transport equations by adding averaged and

fluctuating components The transport equations are changed to equations (33) and

(34)

ρ

δ + nabla ∙ (ρU) = 0 (33)

ρ

+ nabla ∙ ρ U ⨂U = minusnablap + nabla ∙ τ minus ρu⨂u + S (34)

The mass equation is not changed but the momentum equation contains extra terms

which are the Reynolds stresses ρ u⨂u and the Reynolds flux ρ u ϕ These

Reynolds stresses need to be modeled by additional equations to obtain closure

Obtaining closure implies that there are a sufficient number of equations to solve for

all the unknowns including the Reynolds stresses and Reynolds fluxes

Various turbulence models provide various ways to obtain closure In this

investigation Shear Stress Transport (SST) model and k-ε model were utilized The

advantage of SST model is that combines the advantages of other turbulence models

(the k-ε Wilcox k-ω) The 3 turbulence models will be discussed briefly

32 Turbulence models

Turbulence models are necessary because we cannot afford big enough computers

to directly capture every scale of motion There are three modelling frameworks

Direct Numerical Simulation (DNS) Reynolds-Averaged Navier-Stokes

Equations(RANS) Large Eddy Simulation(LES)

As RANS method is applied in most productionseveral RANS turbulence models

differ in their usage of wall functions the number of additional variables solved for

and what these variables represent

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

27

Two equation turbulence models

Two-equation turbulence models are widely used in CFD as they give a good

compromise between computational power needed and accuracy The term lsquotwo

equationrsquo refers to the fact that these models solve for the velocity and length scales

using separate transport equations The turbulence velocity scale is obtained by

solving the transport equation The turbulent length scale is estimated from two

properties of the turbulence field namely the turbulent kinetic energy and the

dissipation rate The dissipation rate of the kinetic energy is obtained from its

transport equation The most widely used are k-ε and k-ω two equation models In

the next section the k-ε Wilcox k-ω BSL k-ω and SST models will be briefly

discussed

321 k-ε turbulence model

The k-epsilon model solves for two variables k the turbulent kinetic energy and

epsilon the rate of dissipation of kinetic energy Wall functions are used in this

model so the flow in the buffer region is not simulated The k-epsilon model is very

popular for industrial applications due to its good convergence rate and relatively

low memory requirements It does not very accurately compute flow fields that

exhibit adverse pressure gradients strong curvature to the flow or jet flow It does

perform well for external flow problems around complex geometries

The k-ε model introduces k (m2s2) as the turbulence kinetic energy and ε (m2s3) as

the turbulence eddy dissipation The continuity equation remains the same

ρ

+ nabla ∙ (ρ U) = 0 (35)

The momentum equation changes as shown by equation

δρ

δ + nabla ∙ (ρU⨂U) = minusnablapprime + nabla ∙ μ

nablaU + (nablaU) + S (36)

Where SM is the sum of body forces μeff is the effective viscosity accounting for

turbulence and prsquo is the modified pressure The k-ε model uses the concept of eddy

viscosity giving the equation for effective viscosity as shown by equation

μ

= μ + μ (37)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

28

μt is the turbulence viscosity is linked to the turbulence kinetic energy and

dissipation by the equation

μ = Cμρ

ε (38)

Where Cμ is a constant

The values for k and ε come from the differential transport equations for the

turbulence kinetic energy and the turbulence dissipation rate

The turbulence kinetic energy equation is given as equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus ρε (39)

The turbulence dissipation rate is given by equation

(ρε)

+ nabla ∙ (ρUε) = nabla ∙ μ +

μ

σε nablaε +

ε

(Cε (P + Pε ) minus Cε ρε) (310)

Where Cε1 Cε2 σk σε are constants

Pk is the turbulence production due to viscous forces and is modeled by the

equation

P = μ nablaU ∙ nablaU + nablaU minus

nabla ∙ U 3μ

nabla ∙ U + ρk (311)

If a buoyancy term is added to the previous equation if the full buoyancy model is

used

The buoyancy term Pkb is modeled as

P = minusμ

ρσρg ∙ nablaρ (312)

322 Wilcox k-ω turbulence model

The k-omega model is similar to k-epsilon instead however it solves for omega mdash

the specific rate of dissipation of kinetic energy It also uses wall functions and

therefore has comparable memory requirements It has more difficulty converging

and is quite sensitive to the initial guess at the solution Hence the k-epsilon model

is often used first to find an initial condition for solving the k-omega model The k-

omega model is useful in many cases where the k-epsilon model is not accurate

such as internal flows flows that exhibit strong curvature separated flows and jets

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

29

This model has an advantage over the k-ε model where it does not involve complex

linear damping functions for near wall calculations at low Reynolds numbers

The k-ω model assumes that the turbulence viscosity is related to the turbulence

kinetic energy k and the turbulent frequency ω by the equation

μ = ρ

ω (313)

The transport equation for k is given by the equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus βprimeρ k ω (314)

The transport equation for ω is shown as equation

(ρω)

δ + nabla ∙ (ρ U ω) = nabla ∙ μ +

μ

μω

nablaω + αω

P + Pω minus β ρ ω (315)

The production rate of turbulence (Pk) is calculated as shown previously in the k-ε

section

The recommended values for the constants in the above equations are

βprime = 009

α = 59

β = 0075

σ = 2

σω = 2

The Reynolds stress tensor ρ u⨂u is calculated by

minusρ u⨂u = μ nablaU + (nablaU) minus

part ρk + μ

nabla ∙ U (316)

323 Shear stress transport model

The disadvantage of the Wilcox model is the strong sensitivity to free-stream

conditions Therefore a blending of the k-ω model near the surface and the k-ε in

the outer region was made by Menter [23] which resulted in the formulation of the

BSL k-ω turbulence model It consists of a transformation of the k-ε model to a k-ω

formulation and subsequently adding the resulting equations The Wilcox model is

multiplied by a blending function F1 and the transformed k-ε by another function 1-

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

30

F1 F1 is a function of wall distance (being the value of one near the surface and zero

outside the boundary layer) Outside the boundary and on the edge of the boundary

layer the standard k-ε model is used [24]

SST model is a combination of the k-epsilon in the free stream and the k-omega

models near the walls It does not use wall functions and tends to be most accurate

when solving the flow near the wall The SST model does not always converge to

the solution quickly so the k-epsilon or k-omega models are often solved first to

give good initial conditions

The k-ω based Shear Stress Transport (SST) model of Menter was applied for

turbulence treatment The transport equations for the SST model are expressed

below where the turbulent kinetic energy lsquokrsquo and turbulent frequency or dissipation

per unit turbulent kinetic energy are computed by using the following relations

For Turbulence Kinetic Energy

kkPukt

kTkk Ntilde+Ntilde+-=Ntilde+

para

para)[()( nsnwb (317)

where Pk is the production limiter

For Specific Dissipation Rate

ww

swnsnbwaww

ww NtildeNtilde-+Ntilde+Ntilde+-=Ntilde+para

para

1)1(2])[()( 21

22 kFSut

T (318)

The first blending function F1 is calculated from

iumlthorn

iumlyacute

uuml

iumlicirc

iumliacute

igrave

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

uacuteuacuteucirc

ugrave

ecircecirceuml

eacute

dividedivideoslash

oumlccedilccedilegrave

aelig=

4

22

21

4500maxmintanh

yCD

k

yy

kF

kw

wrs

w

n

wb (319)

y is the distance to the nearest wall and v is the kinematic viscosity

divideoslash

oumlccedilegrave

aeligNtildeNtilde= -10

2 101

2max ww

rsww kCDk (320)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

31

And Kinematic eddy viscosity

)max( 21

1

SF

kT

wa

an = (321)

where F2 is a blending function which restricts the limiter to the wall boundary and

S is the invariant measure of the strain rate

uacuteuacute

ucirc

ugrave

ecircecirc

euml

eacute

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

dividedivideoslash

oumlccedilccedilegrave

aelig=

2

22

5002maxtanh

w

n

wb yy

kF (322)

33 Brief introduction of simulation process

All aspects of CFD study including modelling meshing solving and post-

processing of the results are briefly introduced in this chapter

331 Pre-processing technique

The speed and accuracy of simulation results depending fundamentally on meshing

and the ability to prepare model geometry correctly

The steps in pre-processing are explained as follows

1) Domain Definitions

The domain definition depends mainly on the physics involved in the problem and

objective of the problem

a) Geometry creation or geometry import

b) Geometry cleanup

i) Removing the parts not necessary for simulation under consideration

ii) Closing the gaps in the geometry

iii) Removing small surfaces curves to make meshing process simple

2)Grid generation

To solve the governing equations the numerical methods like Finite Difference

Finite Volume or Finite Element Method to divide the domain into small parts and

solve the equations on one part at a time are applied

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 33: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

21

1) Overall approach flow patterns

The characteristics of flow approaching an intake structure is one of the foremost

considerations for the designer It is difficult to predict the effects of a given set of

flow conditions upstream from an intake structure Fig 210 depicts several typical

approach flow patterns of pump sump with different operating conditions

Fig 210 Examples of approach flow patterns

Generally speaking for the intake problems especially for the multiple pumps

sump there are the main measures using dividing walls placed between the pumps

as a partitioned structure trash racks with elongated bars guide piers or a number

of structural columns at the bay entrance can provide some assistance in controlling

cross-flow baffles flow distributors and screen mesh etc are applied to improve

the flow distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

22

2) Problems around pump intake

a) Free surface vortices

There are a number of factors that have an effect on the formation of surface

vortices To achieve a higher degree of certainty that objectionable surface vortices

do not form modifications can be made to intake structure to allow operation at

practical depths of submergence and the minimum submergence is referred in Table

22 the use of suction umbrellas vertical curtain walls and horizontal gratings

b) Submerged vortices

In the vicinity of intake bell there are the most complicated flow patterns and flow

makes the most changes in direction while maintaining a constant acceleration into

the pump bell to prevent local flow separation turbulence and vortex formation

Fig 211 represents a sample of various devices to inhibit the different submerged

vortex

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

23

Fig 211 Methods to reduce submerged vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

24

c) Preswirl

The approach flow distributions determine whether flow preswirl exists and a

sufficiently laterally skewed approach flow cause rotating around the intake bell

The most effective way of reducing preswirl is to establish a relatively uniform

approach flow within each pump bay by using the baffling schemes controlling

cross-flow and expanding concentrated flow discussed in the above part

To prevent non-uniform velocity distribution swirls and vortices necessary

modifications are conducted such as fore bay length baffle wall distributors

curtain wall mesh screen guide vane floor splitter wall fillet etc As the above

measures applied modification test will be conducted to check all the flow

conditions and compare with the preliminary results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

25

Chapter 3 Methodology

CFD (Computational fluid dynamics) is the technique which can design cycles to

better performance and reduce costs and time In CFD we solve the governing

equations of given physics (may be differential form or integral form) using some

numerical techniques like Finite Difference Method (FDM) Finite Element Method

(FEM) or Finite Volume Method (FVM) Although CFD plays a great part in the

design of turbomachinery CFD is not a replacement for experimental or analytical

approach In this study the numerical analysis was performed by the commercial

code ANSYS CFX 130 All aspects of the CFD study including modelling

meshing solving and post-processing of the results

31 Numerical modeling

ANSYS CFX solves the unsteady Navier-Stokes equations in their conservation

form The instantaneous equation of mass (continuity) in the stationary frame is

expressed as equation

ρ

δ + nabla ∙ (ρU) = 0 (31)

And the instantaneous equation for momentum is expressed as shown in equation

ρ

δ + nabla ∙ (ρU⨂U) = minusnablap + nabla ∙ τ + S (32)

Where τ is the stress tensor ( including both normal and shear components of the

stress)

These instantaneous equations are averaged for turbulent flows leading to additional

terms that need to be solved While the Navier-Stokes equations describe both

laminar and turbulent flows without addition terms realistic flows involve length

scales much smaller than the smallest finite volume mesh A Direct Numerical

Simulation of these flows would require significantly more computing power than

what is available now or in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

26

Therefore much research has been done to predict the effects of turbulence by

using turbulence models These models account for the effects of turbulence

without the use of a very fine mesh or direct numerical simulation

These turbulence models modify the transport equations by adding averaged and

fluctuating components The transport equations are changed to equations (33) and

(34)

ρ

δ + nabla ∙ (ρU) = 0 (33)

ρ

+ nabla ∙ ρ U ⨂U = minusnablap + nabla ∙ τ minus ρu⨂u + S (34)

The mass equation is not changed but the momentum equation contains extra terms

which are the Reynolds stresses ρ u⨂u and the Reynolds flux ρ u ϕ These

Reynolds stresses need to be modeled by additional equations to obtain closure

Obtaining closure implies that there are a sufficient number of equations to solve for

all the unknowns including the Reynolds stresses and Reynolds fluxes

Various turbulence models provide various ways to obtain closure In this

investigation Shear Stress Transport (SST) model and k-ε model were utilized The

advantage of SST model is that combines the advantages of other turbulence models

(the k-ε Wilcox k-ω) The 3 turbulence models will be discussed briefly

32 Turbulence models

Turbulence models are necessary because we cannot afford big enough computers

to directly capture every scale of motion There are three modelling frameworks

Direct Numerical Simulation (DNS) Reynolds-Averaged Navier-Stokes

Equations(RANS) Large Eddy Simulation(LES)

As RANS method is applied in most productionseveral RANS turbulence models

differ in their usage of wall functions the number of additional variables solved for

and what these variables represent

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

27

Two equation turbulence models

Two-equation turbulence models are widely used in CFD as they give a good

compromise between computational power needed and accuracy The term lsquotwo

equationrsquo refers to the fact that these models solve for the velocity and length scales

using separate transport equations The turbulence velocity scale is obtained by

solving the transport equation The turbulent length scale is estimated from two

properties of the turbulence field namely the turbulent kinetic energy and the

dissipation rate The dissipation rate of the kinetic energy is obtained from its

transport equation The most widely used are k-ε and k-ω two equation models In

the next section the k-ε Wilcox k-ω BSL k-ω and SST models will be briefly

discussed

321 k-ε turbulence model

The k-epsilon model solves for two variables k the turbulent kinetic energy and

epsilon the rate of dissipation of kinetic energy Wall functions are used in this

model so the flow in the buffer region is not simulated The k-epsilon model is very

popular for industrial applications due to its good convergence rate and relatively

low memory requirements It does not very accurately compute flow fields that

exhibit adverse pressure gradients strong curvature to the flow or jet flow It does

perform well for external flow problems around complex geometries

The k-ε model introduces k (m2s2) as the turbulence kinetic energy and ε (m2s3) as

the turbulence eddy dissipation The continuity equation remains the same

ρ

+ nabla ∙ (ρ U) = 0 (35)

The momentum equation changes as shown by equation

δρ

δ + nabla ∙ (ρU⨂U) = minusnablapprime + nabla ∙ μ

nablaU + (nablaU) + S (36)

Where SM is the sum of body forces μeff is the effective viscosity accounting for

turbulence and prsquo is the modified pressure The k-ε model uses the concept of eddy

viscosity giving the equation for effective viscosity as shown by equation

μ

= μ + μ (37)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

28

μt is the turbulence viscosity is linked to the turbulence kinetic energy and

dissipation by the equation

μ = Cμρ

ε (38)

Where Cμ is a constant

The values for k and ε come from the differential transport equations for the

turbulence kinetic energy and the turbulence dissipation rate

The turbulence kinetic energy equation is given as equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus ρε (39)

The turbulence dissipation rate is given by equation

(ρε)

+ nabla ∙ (ρUε) = nabla ∙ μ +

μ

σε nablaε +

ε

(Cε (P + Pε ) minus Cε ρε) (310)

Where Cε1 Cε2 σk σε are constants

Pk is the turbulence production due to viscous forces and is modeled by the

equation

P = μ nablaU ∙ nablaU + nablaU minus

nabla ∙ U 3μ

nabla ∙ U + ρk (311)

If a buoyancy term is added to the previous equation if the full buoyancy model is

used

The buoyancy term Pkb is modeled as

P = minusμ

ρσρg ∙ nablaρ (312)

322 Wilcox k-ω turbulence model

The k-omega model is similar to k-epsilon instead however it solves for omega mdash

the specific rate of dissipation of kinetic energy It also uses wall functions and

therefore has comparable memory requirements It has more difficulty converging

and is quite sensitive to the initial guess at the solution Hence the k-epsilon model

is often used first to find an initial condition for solving the k-omega model The k-

omega model is useful in many cases where the k-epsilon model is not accurate

such as internal flows flows that exhibit strong curvature separated flows and jets

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

29

This model has an advantage over the k-ε model where it does not involve complex

linear damping functions for near wall calculations at low Reynolds numbers

The k-ω model assumes that the turbulence viscosity is related to the turbulence

kinetic energy k and the turbulent frequency ω by the equation

μ = ρ

ω (313)

The transport equation for k is given by the equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus βprimeρ k ω (314)

The transport equation for ω is shown as equation

(ρω)

δ + nabla ∙ (ρ U ω) = nabla ∙ μ +

μ

μω

nablaω + αω

P + Pω minus β ρ ω (315)

The production rate of turbulence (Pk) is calculated as shown previously in the k-ε

section

The recommended values for the constants in the above equations are

βprime = 009

α = 59

β = 0075

σ = 2

σω = 2

The Reynolds stress tensor ρ u⨂u is calculated by

minusρ u⨂u = μ nablaU + (nablaU) minus

part ρk + μ

nabla ∙ U (316)

323 Shear stress transport model

The disadvantage of the Wilcox model is the strong sensitivity to free-stream

conditions Therefore a blending of the k-ω model near the surface and the k-ε in

the outer region was made by Menter [23] which resulted in the formulation of the

BSL k-ω turbulence model It consists of a transformation of the k-ε model to a k-ω

formulation and subsequently adding the resulting equations The Wilcox model is

multiplied by a blending function F1 and the transformed k-ε by another function 1-

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

30

F1 F1 is a function of wall distance (being the value of one near the surface and zero

outside the boundary layer) Outside the boundary and on the edge of the boundary

layer the standard k-ε model is used [24]

SST model is a combination of the k-epsilon in the free stream and the k-omega

models near the walls It does not use wall functions and tends to be most accurate

when solving the flow near the wall The SST model does not always converge to

the solution quickly so the k-epsilon or k-omega models are often solved first to

give good initial conditions

The k-ω based Shear Stress Transport (SST) model of Menter was applied for

turbulence treatment The transport equations for the SST model are expressed

below where the turbulent kinetic energy lsquokrsquo and turbulent frequency or dissipation

per unit turbulent kinetic energy are computed by using the following relations

For Turbulence Kinetic Energy

kkPukt

kTkk Ntilde+Ntilde+-=Ntilde+

para

para)[()( nsnwb (317)

where Pk is the production limiter

For Specific Dissipation Rate

ww

swnsnbwaww

ww NtildeNtilde-+Ntilde+Ntilde+-=Ntilde+para

para

1)1(2])[()( 21

22 kFSut

T (318)

The first blending function F1 is calculated from

iumlthorn

iumlyacute

uuml

iumlicirc

iumliacute

igrave

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

uacuteuacuteucirc

ugrave

ecircecirceuml

eacute

dividedivideoslash

oumlccedilccedilegrave

aelig=

4

22

21

4500maxmintanh

yCD

k

yy

kF

kw

wrs

w

n

wb (319)

y is the distance to the nearest wall and v is the kinematic viscosity

divideoslash

oumlccedilegrave

aeligNtildeNtilde= -10

2 101

2max ww

rsww kCDk (320)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

31

And Kinematic eddy viscosity

)max( 21

1

SF

kT

wa

an = (321)

where F2 is a blending function which restricts the limiter to the wall boundary and

S is the invariant measure of the strain rate

uacuteuacute

ucirc

ugrave

ecircecirc

euml

eacute

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

dividedivideoslash

oumlccedilccedilegrave

aelig=

2

22

5002maxtanh

w

n

wb yy

kF (322)

33 Brief introduction of simulation process

All aspects of CFD study including modelling meshing solving and post-

processing of the results are briefly introduced in this chapter

331 Pre-processing technique

The speed and accuracy of simulation results depending fundamentally on meshing

and the ability to prepare model geometry correctly

The steps in pre-processing are explained as follows

1) Domain Definitions

The domain definition depends mainly on the physics involved in the problem and

objective of the problem

a) Geometry creation or geometry import

b) Geometry cleanup

i) Removing the parts not necessary for simulation under consideration

ii) Closing the gaps in the geometry

iii) Removing small surfaces curves to make meshing process simple

2)Grid generation

To solve the governing equations the numerical methods like Finite Difference

Finite Volume or Finite Element Method to divide the domain into small parts and

solve the equations on one part at a time are applied

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 34: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

22

2) Problems around pump intake

a) Free surface vortices

There are a number of factors that have an effect on the formation of surface

vortices To achieve a higher degree of certainty that objectionable surface vortices

do not form modifications can be made to intake structure to allow operation at

practical depths of submergence and the minimum submergence is referred in Table

22 the use of suction umbrellas vertical curtain walls and horizontal gratings

b) Submerged vortices

In the vicinity of intake bell there are the most complicated flow patterns and flow

makes the most changes in direction while maintaining a constant acceleration into

the pump bell to prevent local flow separation turbulence and vortex formation

Fig 211 represents a sample of various devices to inhibit the different submerged

vortex

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

23

Fig 211 Methods to reduce submerged vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

24

c) Preswirl

The approach flow distributions determine whether flow preswirl exists and a

sufficiently laterally skewed approach flow cause rotating around the intake bell

The most effective way of reducing preswirl is to establish a relatively uniform

approach flow within each pump bay by using the baffling schemes controlling

cross-flow and expanding concentrated flow discussed in the above part

To prevent non-uniform velocity distribution swirls and vortices necessary

modifications are conducted such as fore bay length baffle wall distributors

curtain wall mesh screen guide vane floor splitter wall fillet etc As the above

measures applied modification test will be conducted to check all the flow

conditions and compare with the preliminary results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

25

Chapter 3 Methodology

CFD (Computational fluid dynamics) is the technique which can design cycles to

better performance and reduce costs and time In CFD we solve the governing

equations of given physics (may be differential form or integral form) using some

numerical techniques like Finite Difference Method (FDM) Finite Element Method

(FEM) or Finite Volume Method (FVM) Although CFD plays a great part in the

design of turbomachinery CFD is not a replacement for experimental or analytical

approach In this study the numerical analysis was performed by the commercial

code ANSYS CFX 130 All aspects of the CFD study including modelling

meshing solving and post-processing of the results

31 Numerical modeling

ANSYS CFX solves the unsteady Navier-Stokes equations in their conservation

form The instantaneous equation of mass (continuity) in the stationary frame is

expressed as equation

ρ

δ + nabla ∙ (ρU) = 0 (31)

And the instantaneous equation for momentum is expressed as shown in equation

ρ

δ + nabla ∙ (ρU⨂U) = minusnablap + nabla ∙ τ + S (32)

Where τ is the stress tensor ( including both normal and shear components of the

stress)

These instantaneous equations are averaged for turbulent flows leading to additional

terms that need to be solved While the Navier-Stokes equations describe both

laminar and turbulent flows without addition terms realistic flows involve length

scales much smaller than the smallest finite volume mesh A Direct Numerical

Simulation of these flows would require significantly more computing power than

what is available now or in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

26

Therefore much research has been done to predict the effects of turbulence by

using turbulence models These models account for the effects of turbulence

without the use of a very fine mesh or direct numerical simulation

These turbulence models modify the transport equations by adding averaged and

fluctuating components The transport equations are changed to equations (33) and

(34)

ρ

δ + nabla ∙ (ρU) = 0 (33)

ρ

+ nabla ∙ ρ U ⨂U = minusnablap + nabla ∙ τ minus ρu⨂u + S (34)

The mass equation is not changed but the momentum equation contains extra terms

which are the Reynolds stresses ρ u⨂u and the Reynolds flux ρ u ϕ These

Reynolds stresses need to be modeled by additional equations to obtain closure

Obtaining closure implies that there are a sufficient number of equations to solve for

all the unknowns including the Reynolds stresses and Reynolds fluxes

Various turbulence models provide various ways to obtain closure In this

investigation Shear Stress Transport (SST) model and k-ε model were utilized The

advantage of SST model is that combines the advantages of other turbulence models

(the k-ε Wilcox k-ω) The 3 turbulence models will be discussed briefly

32 Turbulence models

Turbulence models are necessary because we cannot afford big enough computers

to directly capture every scale of motion There are three modelling frameworks

Direct Numerical Simulation (DNS) Reynolds-Averaged Navier-Stokes

Equations(RANS) Large Eddy Simulation(LES)

As RANS method is applied in most productionseveral RANS turbulence models

differ in their usage of wall functions the number of additional variables solved for

and what these variables represent

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

27

Two equation turbulence models

Two-equation turbulence models are widely used in CFD as they give a good

compromise between computational power needed and accuracy The term lsquotwo

equationrsquo refers to the fact that these models solve for the velocity and length scales

using separate transport equations The turbulence velocity scale is obtained by

solving the transport equation The turbulent length scale is estimated from two

properties of the turbulence field namely the turbulent kinetic energy and the

dissipation rate The dissipation rate of the kinetic energy is obtained from its

transport equation The most widely used are k-ε and k-ω two equation models In

the next section the k-ε Wilcox k-ω BSL k-ω and SST models will be briefly

discussed

321 k-ε turbulence model

The k-epsilon model solves for two variables k the turbulent kinetic energy and

epsilon the rate of dissipation of kinetic energy Wall functions are used in this

model so the flow in the buffer region is not simulated The k-epsilon model is very

popular for industrial applications due to its good convergence rate and relatively

low memory requirements It does not very accurately compute flow fields that

exhibit adverse pressure gradients strong curvature to the flow or jet flow It does

perform well for external flow problems around complex geometries

The k-ε model introduces k (m2s2) as the turbulence kinetic energy and ε (m2s3) as

the turbulence eddy dissipation The continuity equation remains the same

ρ

+ nabla ∙ (ρ U) = 0 (35)

The momentum equation changes as shown by equation

δρ

δ + nabla ∙ (ρU⨂U) = minusnablapprime + nabla ∙ μ

nablaU + (nablaU) + S (36)

Where SM is the sum of body forces μeff is the effective viscosity accounting for

turbulence and prsquo is the modified pressure The k-ε model uses the concept of eddy

viscosity giving the equation for effective viscosity as shown by equation

μ

= μ + μ (37)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

28

μt is the turbulence viscosity is linked to the turbulence kinetic energy and

dissipation by the equation

μ = Cμρ

ε (38)

Where Cμ is a constant

The values for k and ε come from the differential transport equations for the

turbulence kinetic energy and the turbulence dissipation rate

The turbulence kinetic energy equation is given as equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus ρε (39)

The turbulence dissipation rate is given by equation

(ρε)

+ nabla ∙ (ρUε) = nabla ∙ μ +

μ

σε nablaε +

ε

(Cε (P + Pε ) minus Cε ρε) (310)

Where Cε1 Cε2 σk σε are constants

Pk is the turbulence production due to viscous forces and is modeled by the

equation

P = μ nablaU ∙ nablaU + nablaU minus

nabla ∙ U 3μ

nabla ∙ U + ρk (311)

If a buoyancy term is added to the previous equation if the full buoyancy model is

used

The buoyancy term Pkb is modeled as

P = minusμ

ρσρg ∙ nablaρ (312)

322 Wilcox k-ω turbulence model

The k-omega model is similar to k-epsilon instead however it solves for omega mdash

the specific rate of dissipation of kinetic energy It also uses wall functions and

therefore has comparable memory requirements It has more difficulty converging

and is quite sensitive to the initial guess at the solution Hence the k-epsilon model

is often used first to find an initial condition for solving the k-omega model The k-

omega model is useful in many cases where the k-epsilon model is not accurate

such as internal flows flows that exhibit strong curvature separated flows and jets

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

29

This model has an advantage over the k-ε model where it does not involve complex

linear damping functions for near wall calculations at low Reynolds numbers

The k-ω model assumes that the turbulence viscosity is related to the turbulence

kinetic energy k and the turbulent frequency ω by the equation

μ = ρ

ω (313)

The transport equation for k is given by the equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus βprimeρ k ω (314)

The transport equation for ω is shown as equation

(ρω)

δ + nabla ∙ (ρ U ω) = nabla ∙ μ +

μ

μω

nablaω + αω

P + Pω minus β ρ ω (315)

The production rate of turbulence (Pk) is calculated as shown previously in the k-ε

section

The recommended values for the constants in the above equations are

βprime = 009

α = 59

β = 0075

σ = 2

σω = 2

The Reynolds stress tensor ρ u⨂u is calculated by

minusρ u⨂u = μ nablaU + (nablaU) minus

part ρk + μ

nabla ∙ U (316)

323 Shear stress transport model

The disadvantage of the Wilcox model is the strong sensitivity to free-stream

conditions Therefore a blending of the k-ω model near the surface and the k-ε in

the outer region was made by Menter [23] which resulted in the formulation of the

BSL k-ω turbulence model It consists of a transformation of the k-ε model to a k-ω

formulation and subsequently adding the resulting equations The Wilcox model is

multiplied by a blending function F1 and the transformed k-ε by another function 1-

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

30

F1 F1 is a function of wall distance (being the value of one near the surface and zero

outside the boundary layer) Outside the boundary and on the edge of the boundary

layer the standard k-ε model is used [24]

SST model is a combination of the k-epsilon in the free stream and the k-omega

models near the walls It does not use wall functions and tends to be most accurate

when solving the flow near the wall The SST model does not always converge to

the solution quickly so the k-epsilon or k-omega models are often solved first to

give good initial conditions

The k-ω based Shear Stress Transport (SST) model of Menter was applied for

turbulence treatment The transport equations for the SST model are expressed

below where the turbulent kinetic energy lsquokrsquo and turbulent frequency or dissipation

per unit turbulent kinetic energy are computed by using the following relations

For Turbulence Kinetic Energy

kkPukt

kTkk Ntilde+Ntilde+-=Ntilde+

para

para)[()( nsnwb (317)

where Pk is the production limiter

For Specific Dissipation Rate

ww

swnsnbwaww

ww NtildeNtilde-+Ntilde+Ntilde+-=Ntilde+para

para

1)1(2])[()( 21

22 kFSut

T (318)

The first blending function F1 is calculated from

iumlthorn

iumlyacute

uuml

iumlicirc

iumliacute

igrave

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

uacuteuacuteucirc

ugrave

ecircecirceuml

eacute

dividedivideoslash

oumlccedilccedilegrave

aelig=

4

22

21

4500maxmintanh

yCD

k

yy

kF

kw

wrs

w

n

wb (319)

y is the distance to the nearest wall and v is the kinematic viscosity

divideoslash

oumlccedilegrave

aeligNtildeNtilde= -10

2 101

2max ww

rsww kCDk (320)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

31

And Kinematic eddy viscosity

)max( 21

1

SF

kT

wa

an = (321)

where F2 is a blending function which restricts the limiter to the wall boundary and

S is the invariant measure of the strain rate

uacuteuacute

ucirc

ugrave

ecircecirc

euml

eacute

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

dividedivideoslash

oumlccedilccedilegrave

aelig=

2

22

5002maxtanh

w

n

wb yy

kF (322)

33 Brief introduction of simulation process

All aspects of CFD study including modelling meshing solving and post-

processing of the results are briefly introduced in this chapter

331 Pre-processing technique

The speed and accuracy of simulation results depending fundamentally on meshing

and the ability to prepare model geometry correctly

The steps in pre-processing are explained as follows

1) Domain Definitions

The domain definition depends mainly on the physics involved in the problem and

objective of the problem

a) Geometry creation or geometry import

b) Geometry cleanup

i) Removing the parts not necessary for simulation under consideration

ii) Closing the gaps in the geometry

iii) Removing small surfaces curves to make meshing process simple

2)Grid generation

To solve the governing equations the numerical methods like Finite Difference

Finite Volume or Finite Element Method to divide the domain into small parts and

solve the equations on one part at a time are applied

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 35: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

23

Fig 211 Methods to reduce submerged vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

24

c) Preswirl

The approach flow distributions determine whether flow preswirl exists and a

sufficiently laterally skewed approach flow cause rotating around the intake bell

The most effective way of reducing preswirl is to establish a relatively uniform

approach flow within each pump bay by using the baffling schemes controlling

cross-flow and expanding concentrated flow discussed in the above part

To prevent non-uniform velocity distribution swirls and vortices necessary

modifications are conducted such as fore bay length baffle wall distributors

curtain wall mesh screen guide vane floor splitter wall fillet etc As the above

measures applied modification test will be conducted to check all the flow

conditions and compare with the preliminary results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

25

Chapter 3 Methodology

CFD (Computational fluid dynamics) is the technique which can design cycles to

better performance and reduce costs and time In CFD we solve the governing

equations of given physics (may be differential form or integral form) using some

numerical techniques like Finite Difference Method (FDM) Finite Element Method

(FEM) or Finite Volume Method (FVM) Although CFD plays a great part in the

design of turbomachinery CFD is not a replacement for experimental or analytical

approach In this study the numerical analysis was performed by the commercial

code ANSYS CFX 130 All aspects of the CFD study including modelling

meshing solving and post-processing of the results

31 Numerical modeling

ANSYS CFX solves the unsteady Navier-Stokes equations in their conservation

form The instantaneous equation of mass (continuity) in the stationary frame is

expressed as equation

ρ

δ + nabla ∙ (ρU) = 0 (31)

And the instantaneous equation for momentum is expressed as shown in equation

ρ

δ + nabla ∙ (ρU⨂U) = minusnablap + nabla ∙ τ + S (32)

Where τ is the stress tensor ( including both normal and shear components of the

stress)

These instantaneous equations are averaged for turbulent flows leading to additional

terms that need to be solved While the Navier-Stokes equations describe both

laminar and turbulent flows without addition terms realistic flows involve length

scales much smaller than the smallest finite volume mesh A Direct Numerical

Simulation of these flows would require significantly more computing power than

what is available now or in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

26

Therefore much research has been done to predict the effects of turbulence by

using turbulence models These models account for the effects of turbulence

without the use of a very fine mesh or direct numerical simulation

These turbulence models modify the transport equations by adding averaged and

fluctuating components The transport equations are changed to equations (33) and

(34)

ρ

δ + nabla ∙ (ρU) = 0 (33)

ρ

+ nabla ∙ ρ U ⨂U = minusnablap + nabla ∙ τ minus ρu⨂u + S (34)

The mass equation is not changed but the momentum equation contains extra terms

which are the Reynolds stresses ρ u⨂u and the Reynolds flux ρ u ϕ These

Reynolds stresses need to be modeled by additional equations to obtain closure

Obtaining closure implies that there are a sufficient number of equations to solve for

all the unknowns including the Reynolds stresses and Reynolds fluxes

Various turbulence models provide various ways to obtain closure In this

investigation Shear Stress Transport (SST) model and k-ε model were utilized The

advantage of SST model is that combines the advantages of other turbulence models

(the k-ε Wilcox k-ω) The 3 turbulence models will be discussed briefly

32 Turbulence models

Turbulence models are necessary because we cannot afford big enough computers

to directly capture every scale of motion There are three modelling frameworks

Direct Numerical Simulation (DNS) Reynolds-Averaged Navier-Stokes

Equations(RANS) Large Eddy Simulation(LES)

As RANS method is applied in most productionseveral RANS turbulence models

differ in their usage of wall functions the number of additional variables solved for

and what these variables represent

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

27

Two equation turbulence models

Two-equation turbulence models are widely used in CFD as they give a good

compromise between computational power needed and accuracy The term lsquotwo

equationrsquo refers to the fact that these models solve for the velocity and length scales

using separate transport equations The turbulence velocity scale is obtained by

solving the transport equation The turbulent length scale is estimated from two

properties of the turbulence field namely the turbulent kinetic energy and the

dissipation rate The dissipation rate of the kinetic energy is obtained from its

transport equation The most widely used are k-ε and k-ω two equation models In

the next section the k-ε Wilcox k-ω BSL k-ω and SST models will be briefly

discussed

321 k-ε turbulence model

The k-epsilon model solves for two variables k the turbulent kinetic energy and

epsilon the rate of dissipation of kinetic energy Wall functions are used in this

model so the flow in the buffer region is not simulated The k-epsilon model is very

popular for industrial applications due to its good convergence rate and relatively

low memory requirements It does not very accurately compute flow fields that

exhibit adverse pressure gradients strong curvature to the flow or jet flow It does

perform well for external flow problems around complex geometries

The k-ε model introduces k (m2s2) as the turbulence kinetic energy and ε (m2s3) as

the turbulence eddy dissipation The continuity equation remains the same

ρ

+ nabla ∙ (ρ U) = 0 (35)

The momentum equation changes as shown by equation

δρ

δ + nabla ∙ (ρU⨂U) = minusnablapprime + nabla ∙ μ

nablaU + (nablaU) + S (36)

Where SM is the sum of body forces μeff is the effective viscosity accounting for

turbulence and prsquo is the modified pressure The k-ε model uses the concept of eddy

viscosity giving the equation for effective viscosity as shown by equation

μ

= μ + μ (37)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

28

μt is the turbulence viscosity is linked to the turbulence kinetic energy and

dissipation by the equation

μ = Cμρ

ε (38)

Where Cμ is a constant

The values for k and ε come from the differential transport equations for the

turbulence kinetic energy and the turbulence dissipation rate

The turbulence kinetic energy equation is given as equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus ρε (39)

The turbulence dissipation rate is given by equation

(ρε)

+ nabla ∙ (ρUε) = nabla ∙ μ +

μ

σε nablaε +

ε

(Cε (P + Pε ) minus Cε ρε) (310)

Where Cε1 Cε2 σk σε are constants

Pk is the turbulence production due to viscous forces and is modeled by the

equation

P = μ nablaU ∙ nablaU + nablaU minus

nabla ∙ U 3μ

nabla ∙ U + ρk (311)

If a buoyancy term is added to the previous equation if the full buoyancy model is

used

The buoyancy term Pkb is modeled as

P = minusμ

ρσρg ∙ nablaρ (312)

322 Wilcox k-ω turbulence model

The k-omega model is similar to k-epsilon instead however it solves for omega mdash

the specific rate of dissipation of kinetic energy It also uses wall functions and

therefore has comparable memory requirements It has more difficulty converging

and is quite sensitive to the initial guess at the solution Hence the k-epsilon model

is often used first to find an initial condition for solving the k-omega model The k-

omega model is useful in many cases where the k-epsilon model is not accurate

such as internal flows flows that exhibit strong curvature separated flows and jets

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

29

This model has an advantage over the k-ε model where it does not involve complex

linear damping functions for near wall calculations at low Reynolds numbers

The k-ω model assumes that the turbulence viscosity is related to the turbulence

kinetic energy k and the turbulent frequency ω by the equation

μ = ρ

ω (313)

The transport equation for k is given by the equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus βprimeρ k ω (314)

The transport equation for ω is shown as equation

(ρω)

δ + nabla ∙ (ρ U ω) = nabla ∙ μ +

μ

μω

nablaω + αω

P + Pω minus β ρ ω (315)

The production rate of turbulence (Pk) is calculated as shown previously in the k-ε

section

The recommended values for the constants in the above equations are

βprime = 009

α = 59

β = 0075

σ = 2

σω = 2

The Reynolds stress tensor ρ u⨂u is calculated by

minusρ u⨂u = μ nablaU + (nablaU) minus

part ρk + μ

nabla ∙ U (316)

323 Shear stress transport model

The disadvantage of the Wilcox model is the strong sensitivity to free-stream

conditions Therefore a blending of the k-ω model near the surface and the k-ε in

the outer region was made by Menter [23] which resulted in the formulation of the

BSL k-ω turbulence model It consists of a transformation of the k-ε model to a k-ω

formulation and subsequently adding the resulting equations The Wilcox model is

multiplied by a blending function F1 and the transformed k-ε by another function 1-

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

30

F1 F1 is a function of wall distance (being the value of one near the surface and zero

outside the boundary layer) Outside the boundary and on the edge of the boundary

layer the standard k-ε model is used [24]

SST model is a combination of the k-epsilon in the free stream and the k-omega

models near the walls It does not use wall functions and tends to be most accurate

when solving the flow near the wall The SST model does not always converge to

the solution quickly so the k-epsilon or k-omega models are often solved first to

give good initial conditions

The k-ω based Shear Stress Transport (SST) model of Menter was applied for

turbulence treatment The transport equations for the SST model are expressed

below where the turbulent kinetic energy lsquokrsquo and turbulent frequency or dissipation

per unit turbulent kinetic energy are computed by using the following relations

For Turbulence Kinetic Energy

kkPukt

kTkk Ntilde+Ntilde+-=Ntilde+

para

para)[()( nsnwb (317)

where Pk is the production limiter

For Specific Dissipation Rate

ww

swnsnbwaww

ww NtildeNtilde-+Ntilde+Ntilde+-=Ntilde+para

para

1)1(2])[()( 21

22 kFSut

T (318)

The first blending function F1 is calculated from

iumlthorn

iumlyacute

uuml

iumlicirc

iumliacute

igrave

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

uacuteuacuteucirc

ugrave

ecircecirceuml

eacute

dividedivideoslash

oumlccedilccedilegrave

aelig=

4

22

21

4500maxmintanh

yCD

k

yy

kF

kw

wrs

w

n

wb (319)

y is the distance to the nearest wall and v is the kinematic viscosity

divideoslash

oumlccedilegrave

aeligNtildeNtilde= -10

2 101

2max ww

rsww kCDk (320)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

31

And Kinematic eddy viscosity

)max( 21

1

SF

kT

wa

an = (321)

where F2 is a blending function which restricts the limiter to the wall boundary and

S is the invariant measure of the strain rate

uacuteuacute

ucirc

ugrave

ecircecirc

euml

eacute

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

dividedivideoslash

oumlccedilccedilegrave

aelig=

2

22

5002maxtanh

w

n

wb yy

kF (322)

33 Brief introduction of simulation process

All aspects of CFD study including modelling meshing solving and post-

processing of the results are briefly introduced in this chapter

331 Pre-processing technique

The speed and accuracy of simulation results depending fundamentally on meshing

and the ability to prepare model geometry correctly

The steps in pre-processing are explained as follows

1) Domain Definitions

The domain definition depends mainly on the physics involved in the problem and

objective of the problem

a) Geometry creation or geometry import

b) Geometry cleanup

i) Removing the parts not necessary for simulation under consideration

ii) Closing the gaps in the geometry

iii) Removing small surfaces curves to make meshing process simple

2)Grid generation

To solve the governing equations the numerical methods like Finite Difference

Finite Volume or Finite Element Method to divide the domain into small parts and

solve the equations on one part at a time are applied

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 36: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

24

c) Preswirl

The approach flow distributions determine whether flow preswirl exists and a

sufficiently laterally skewed approach flow cause rotating around the intake bell

The most effective way of reducing preswirl is to establish a relatively uniform

approach flow within each pump bay by using the baffling schemes controlling

cross-flow and expanding concentrated flow discussed in the above part

To prevent non-uniform velocity distribution swirls and vortices necessary

modifications are conducted such as fore bay length baffle wall distributors

curtain wall mesh screen guide vane floor splitter wall fillet etc As the above

measures applied modification test will be conducted to check all the flow

conditions and compare with the preliminary results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

25

Chapter 3 Methodology

CFD (Computational fluid dynamics) is the technique which can design cycles to

better performance and reduce costs and time In CFD we solve the governing

equations of given physics (may be differential form or integral form) using some

numerical techniques like Finite Difference Method (FDM) Finite Element Method

(FEM) or Finite Volume Method (FVM) Although CFD plays a great part in the

design of turbomachinery CFD is not a replacement for experimental or analytical

approach In this study the numerical analysis was performed by the commercial

code ANSYS CFX 130 All aspects of the CFD study including modelling

meshing solving and post-processing of the results

31 Numerical modeling

ANSYS CFX solves the unsteady Navier-Stokes equations in their conservation

form The instantaneous equation of mass (continuity) in the stationary frame is

expressed as equation

ρ

δ + nabla ∙ (ρU) = 0 (31)

And the instantaneous equation for momentum is expressed as shown in equation

ρ

δ + nabla ∙ (ρU⨂U) = minusnablap + nabla ∙ τ + S (32)

Where τ is the stress tensor ( including both normal and shear components of the

stress)

These instantaneous equations are averaged for turbulent flows leading to additional

terms that need to be solved While the Navier-Stokes equations describe both

laminar and turbulent flows without addition terms realistic flows involve length

scales much smaller than the smallest finite volume mesh A Direct Numerical

Simulation of these flows would require significantly more computing power than

what is available now or in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

26

Therefore much research has been done to predict the effects of turbulence by

using turbulence models These models account for the effects of turbulence

without the use of a very fine mesh or direct numerical simulation

These turbulence models modify the transport equations by adding averaged and

fluctuating components The transport equations are changed to equations (33) and

(34)

ρ

δ + nabla ∙ (ρU) = 0 (33)

ρ

+ nabla ∙ ρ U ⨂U = minusnablap + nabla ∙ τ minus ρu⨂u + S (34)

The mass equation is not changed but the momentum equation contains extra terms

which are the Reynolds stresses ρ u⨂u and the Reynolds flux ρ u ϕ These

Reynolds stresses need to be modeled by additional equations to obtain closure

Obtaining closure implies that there are a sufficient number of equations to solve for

all the unknowns including the Reynolds stresses and Reynolds fluxes

Various turbulence models provide various ways to obtain closure In this

investigation Shear Stress Transport (SST) model and k-ε model were utilized The

advantage of SST model is that combines the advantages of other turbulence models

(the k-ε Wilcox k-ω) The 3 turbulence models will be discussed briefly

32 Turbulence models

Turbulence models are necessary because we cannot afford big enough computers

to directly capture every scale of motion There are three modelling frameworks

Direct Numerical Simulation (DNS) Reynolds-Averaged Navier-Stokes

Equations(RANS) Large Eddy Simulation(LES)

As RANS method is applied in most productionseveral RANS turbulence models

differ in their usage of wall functions the number of additional variables solved for

and what these variables represent

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

27

Two equation turbulence models

Two-equation turbulence models are widely used in CFD as they give a good

compromise between computational power needed and accuracy The term lsquotwo

equationrsquo refers to the fact that these models solve for the velocity and length scales

using separate transport equations The turbulence velocity scale is obtained by

solving the transport equation The turbulent length scale is estimated from two

properties of the turbulence field namely the turbulent kinetic energy and the

dissipation rate The dissipation rate of the kinetic energy is obtained from its

transport equation The most widely used are k-ε and k-ω two equation models In

the next section the k-ε Wilcox k-ω BSL k-ω and SST models will be briefly

discussed

321 k-ε turbulence model

The k-epsilon model solves for two variables k the turbulent kinetic energy and

epsilon the rate of dissipation of kinetic energy Wall functions are used in this

model so the flow in the buffer region is not simulated The k-epsilon model is very

popular for industrial applications due to its good convergence rate and relatively

low memory requirements It does not very accurately compute flow fields that

exhibit adverse pressure gradients strong curvature to the flow or jet flow It does

perform well for external flow problems around complex geometries

The k-ε model introduces k (m2s2) as the turbulence kinetic energy and ε (m2s3) as

the turbulence eddy dissipation The continuity equation remains the same

ρ

+ nabla ∙ (ρ U) = 0 (35)

The momentum equation changes as shown by equation

δρ

δ + nabla ∙ (ρU⨂U) = minusnablapprime + nabla ∙ μ

nablaU + (nablaU) + S (36)

Where SM is the sum of body forces μeff is the effective viscosity accounting for

turbulence and prsquo is the modified pressure The k-ε model uses the concept of eddy

viscosity giving the equation for effective viscosity as shown by equation

μ

= μ + μ (37)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

28

μt is the turbulence viscosity is linked to the turbulence kinetic energy and

dissipation by the equation

μ = Cμρ

ε (38)

Where Cμ is a constant

The values for k and ε come from the differential transport equations for the

turbulence kinetic energy and the turbulence dissipation rate

The turbulence kinetic energy equation is given as equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus ρε (39)

The turbulence dissipation rate is given by equation

(ρε)

+ nabla ∙ (ρUε) = nabla ∙ μ +

μ

σε nablaε +

ε

(Cε (P + Pε ) minus Cε ρε) (310)

Where Cε1 Cε2 σk σε are constants

Pk is the turbulence production due to viscous forces and is modeled by the

equation

P = μ nablaU ∙ nablaU + nablaU minus

nabla ∙ U 3μ

nabla ∙ U + ρk (311)

If a buoyancy term is added to the previous equation if the full buoyancy model is

used

The buoyancy term Pkb is modeled as

P = minusμ

ρσρg ∙ nablaρ (312)

322 Wilcox k-ω turbulence model

The k-omega model is similar to k-epsilon instead however it solves for omega mdash

the specific rate of dissipation of kinetic energy It also uses wall functions and

therefore has comparable memory requirements It has more difficulty converging

and is quite sensitive to the initial guess at the solution Hence the k-epsilon model

is often used first to find an initial condition for solving the k-omega model The k-

omega model is useful in many cases where the k-epsilon model is not accurate

such as internal flows flows that exhibit strong curvature separated flows and jets

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

29

This model has an advantage over the k-ε model where it does not involve complex

linear damping functions for near wall calculations at low Reynolds numbers

The k-ω model assumes that the turbulence viscosity is related to the turbulence

kinetic energy k and the turbulent frequency ω by the equation

μ = ρ

ω (313)

The transport equation for k is given by the equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus βprimeρ k ω (314)

The transport equation for ω is shown as equation

(ρω)

δ + nabla ∙ (ρ U ω) = nabla ∙ μ +

μ

μω

nablaω + αω

P + Pω minus β ρ ω (315)

The production rate of turbulence (Pk) is calculated as shown previously in the k-ε

section

The recommended values for the constants in the above equations are

βprime = 009

α = 59

β = 0075

σ = 2

σω = 2

The Reynolds stress tensor ρ u⨂u is calculated by

minusρ u⨂u = μ nablaU + (nablaU) minus

part ρk + μ

nabla ∙ U (316)

323 Shear stress transport model

The disadvantage of the Wilcox model is the strong sensitivity to free-stream

conditions Therefore a blending of the k-ω model near the surface and the k-ε in

the outer region was made by Menter [23] which resulted in the formulation of the

BSL k-ω turbulence model It consists of a transformation of the k-ε model to a k-ω

formulation and subsequently adding the resulting equations The Wilcox model is

multiplied by a blending function F1 and the transformed k-ε by another function 1-

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

30

F1 F1 is a function of wall distance (being the value of one near the surface and zero

outside the boundary layer) Outside the boundary and on the edge of the boundary

layer the standard k-ε model is used [24]

SST model is a combination of the k-epsilon in the free stream and the k-omega

models near the walls It does not use wall functions and tends to be most accurate

when solving the flow near the wall The SST model does not always converge to

the solution quickly so the k-epsilon or k-omega models are often solved first to

give good initial conditions

The k-ω based Shear Stress Transport (SST) model of Menter was applied for

turbulence treatment The transport equations for the SST model are expressed

below where the turbulent kinetic energy lsquokrsquo and turbulent frequency or dissipation

per unit turbulent kinetic energy are computed by using the following relations

For Turbulence Kinetic Energy

kkPukt

kTkk Ntilde+Ntilde+-=Ntilde+

para

para)[()( nsnwb (317)

where Pk is the production limiter

For Specific Dissipation Rate

ww

swnsnbwaww

ww NtildeNtilde-+Ntilde+Ntilde+-=Ntilde+para

para

1)1(2])[()( 21

22 kFSut

T (318)

The first blending function F1 is calculated from

iumlthorn

iumlyacute

uuml

iumlicirc

iumliacute

igrave

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

uacuteuacuteucirc

ugrave

ecircecirceuml

eacute

dividedivideoslash

oumlccedilccedilegrave

aelig=

4

22

21

4500maxmintanh

yCD

k

yy

kF

kw

wrs

w

n

wb (319)

y is the distance to the nearest wall and v is the kinematic viscosity

divideoslash

oumlccedilegrave

aeligNtildeNtilde= -10

2 101

2max ww

rsww kCDk (320)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

31

And Kinematic eddy viscosity

)max( 21

1

SF

kT

wa

an = (321)

where F2 is a blending function which restricts the limiter to the wall boundary and

S is the invariant measure of the strain rate

uacuteuacute

ucirc

ugrave

ecircecirc

euml

eacute

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

dividedivideoslash

oumlccedilccedilegrave

aelig=

2

22

5002maxtanh

w

n

wb yy

kF (322)

33 Brief introduction of simulation process

All aspects of CFD study including modelling meshing solving and post-

processing of the results are briefly introduced in this chapter

331 Pre-processing technique

The speed and accuracy of simulation results depending fundamentally on meshing

and the ability to prepare model geometry correctly

The steps in pre-processing are explained as follows

1) Domain Definitions

The domain definition depends mainly on the physics involved in the problem and

objective of the problem

a) Geometry creation or geometry import

b) Geometry cleanup

i) Removing the parts not necessary for simulation under consideration

ii) Closing the gaps in the geometry

iii) Removing small surfaces curves to make meshing process simple

2)Grid generation

To solve the governing equations the numerical methods like Finite Difference

Finite Volume or Finite Element Method to divide the domain into small parts and

solve the equations on one part at a time are applied

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 37: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

25

Chapter 3 Methodology

CFD (Computational fluid dynamics) is the technique which can design cycles to

better performance and reduce costs and time In CFD we solve the governing

equations of given physics (may be differential form or integral form) using some

numerical techniques like Finite Difference Method (FDM) Finite Element Method

(FEM) or Finite Volume Method (FVM) Although CFD plays a great part in the

design of turbomachinery CFD is not a replacement for experimental or analytical

approach In this study the numerical analysis was performed by the commercial

code ANSYS CFX 130 All aspects of the CFD study including modelling

meshing solving and post-processing of the results

31 Numerical modeling

ANSYS CFX solves the unsteady Navier-Stokes equations in their conservation

form The instantaneous equation of mass (continuity) in the stationary frame is

expressed as equation

ρ

δ + nabla ∙ (ρU) = 0 (31)

And the instantaneous equation for momentum is expressed as shown in equation

ρ

δ + nabla ∙ (ρU⨂U) = minusnablap + nabla ∙ τ + S (32)

Where τ is the stress tensor ( including both normal and shear components of the

stress)

These instantaneous equations are averaged for turbulent flows leading to additional

terms that need to be solved While the Navier-Stokes equations describe both

laminar and turbulent flows without addition terms realistic flows involve length

scales much smaller than the smallest finite volume mesh A Direct Numerical

Simulation of these flows would require significantly more computing power than

what is available now or in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

26

Therefore much research has been done to predict the effects of turbulence by

using turbulence models These models account for the effects of turbulence

without the use of a very fine mesh or direct numerical simulation

These turbulence models modify the transport equations by adding averaged and

fluctuating components The transport equations are changed to equations (33) and

(34)

ρ

δ + nabla ∙ (ρU) = 0 (33)

ρ

+ nabla ∙ ρ U ⨂U = minusnablap + nabla ∙ τ minus ρu⨂u + S (34)

The mass equation is not changed but the momentum equation contains extra terms

which are the Reynolds stresses ρ u⨂u and the Reynolds flux ρ u ϕ These

Reynolds stresses need to be modeled by additional equations to obtain closure

Obtaining closure implies that there are a sufficient number of equations to solve for

all the unknowns including the Reynolds stresses and Reynolds fluxes

Various turbulence models provide various ways to obtain closure In this

investigation Shear Stress Transport (SST) model and k-ε model were utilized The

advantage of SST model is that combines the advantages of other turbulence models

(the k-ε Wilcox k-ω) The 3 turbulence models will be discussed briefly

32 Turbulence models

Turbulence models are necessary because we cannot afford big enough computers

to directly capture every scale of motion There are three modelling frameworks

Direct Numerical Simulation (DNS) Reynolds-Averaged Navier-Stokes

Equations(RANS) Large Eddy Simulation(LES)

As RANS method is applied in most productionseveral RANS turbulence models

differ in their usage of wall functions the number of additional variables solved for

and what these variables represent

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

27

Two equation turbulence models

Two-equation turbulence models are widely used in CFD as they give a good

compromise between computational power needed and accuracy The term lsquotwo

equationrsquo refers to the fact that these models solve for the velocity and length scales

using separate transport equations The turbulence velocity scale is obtained by

solving the transport equation The turbulent length scale is estimated from two

properties of the turbulence field namely the turbulent kinetic energy and the

dissipation rate The dissipation rate of the kinetic energy is obtained from its

transport equation The most widely used are k-ε and k-ω two equation models In

the next section the k-ε Wilcox k-ω BSL k-ω and SST models will be briefly

discussed

321 k-ε turbulence model

The k-epsilon model solves for two variables k the turbulent kinetic energy and

epsilon the rate of dissipation of kinetic energy Wall functions are used in this

model so the flow in the buffer region is not simulated The k-epsilon model is very

popular for industrial applications due to its good convergence rate and relatively

low memory requirements It does not very accurately compute flow fields that

exhibit adverse pressure gradients strong curvature to the flow or jet flow It does

perform well for external flow problems around complex geometries

The k-ε model introduces k (m2s2) as the turbulence kinetic energy and ε (m2s3) as

the turbulence eddy dissipation The continuity equation remains the same

ρ

+ nabla ∙ (ρ U) = 0 (35)

The momentum equation changes as shown by equation

δρ

δ + nabla ∙ (ρU⨂U) = minusnablapprime + nabla ∙ μ

nablaU + (nablaU) + S (36)

Where SM is the sum of body forces μeff is the effective viscosity accounting for

turbulence and prsquo is the modified pressure The k-ε model uses the concept of eddy

viscosity giving the equation for effective viscosity as shown by equation

μ

= μ + μ (37)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

28

μt is the turbulence viscosity is linked to the turbulence kinetic energy and

dissipation by the equation

μ = Cμρ

ε (38)

Where Cμ is a constant

The values for k and ε come from the differential transport equations for the

turbulence kinetic energy and the turbulence dissipation rate

The turbulence kinetic energy equation is given as equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus ρε (39)

The turbulence dissipation rate is given by equation

(ρε)

+ nabla ∙ (ρUε) = nabla ∙ μ +

μ

σε nablaε +

ε

(Cε (P + Pε ) minus Cε ρε) (310)

Where Cε1 Cε2 σk σε are constants

Pk is the turbulence production due to viscous forces and is modeled by the

equation

P = μ nablaU ∙ nablaU + nablaU minus

nabla ∙ U 3μ

nabla ∙ U + ρk (311)

If a buoyancy term is added to the previous equation if the full buoyancy model is

used

The buoyancy term Pkb is modeled as

P = minusμ

ρσρg ∙ nablaρ (312)

322 Wilcox k-ω turbulence model

The k-omega model is similar to k-epsilon instead however it solves for omega mdash

the specific rate of dissipation of kinetic energy It also uses wall functions and

therefore has comparable memory requirements It has more difficulty converging

and is quite sensitive to the initial guess at the solution Hence the k-epsilon model

is often used first to find an initial condition for solving the k-omega model The k-

omega model is useful in many cases where the k-epsilon model is not accurate

such as internal flows flows that exhibit strong curvature separated flows and jets

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

29

This model has an advantage over the k-ε model where it does not involve complex

linear damping functions for near wall calculations at low Reynolds numbers

The k-ω model assumes that the turbulence viscosity is related to the turbulence

kinetic energy k and the turbulent frequency ω by the equation

μ = ρ

ω (313)

The transport equation for k is given by the equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus βprimeρ k ω (314)

The transport equation for ω is shown as equation

(ρω)

δ + nabla ∙ (ρ U ω) = nabla ∙ μ +

μ

μω

nablaω + αω

P + Pω minus β ρ ω (315)

The production rate of turbulence (Pk) is calculated as shown previously in the k-ε

section

The recommended values for the constants in the above equations are

βprime = 009

α = 59

β = 0075

σ = 2

σω = 2

The Reynolds stress tensor ρ u⨂u is calculated by

minusρ u⨂u = μ nablaU + (nablaU) minus

part ρk + μ

nabla ∙ U (316)

323 Shear stress transport model

The disadvantage of the Wilcox model is the strong sensitivity to free-stream

conditions Therefore a blending of the k-ω model near the surface and the k-ε in

the outer region was made by Menter [23] which resulted in the formulation of the

BSL k-ω turbulence model It consists of a transformation of the k-ε model to a k-ω

formulation and subsequently adding the resulting equations The Wilcox model is

multiplied by a blending function F1 and the transformed k-ε by another function 1-

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

30

F1 F1 is a function of wall distance (being the value of one near the surface and zero

outside the boundary layer) Outside the boundary and on the edge of the boundary

layer the standard k-ε model is used [24]

SST model is a combination of the k-epsilon in the free stream and the k-omega

models near the walls It does not use wall functions and tends to be most accurate

when solving the flow near the wall The SST model does not always converge to

the solution quickly so the k-epsilon or k-omega models are often solved first to

give good initial conditions

The k-ω based Shear Stress Transport (SST) model of Menter was applied for

turbulence treatment The transport equations for the SST model are expressed

below where the turbulent kinetic energy lsquokrsquo and turbulent frequency or dissipation

per unit turbulent kinetic energy are computed by using the following relations

For Turbulence Kinetic Energy

kkPukt

kTkk Ntilde+Ntilde+-=Ntilde+

para

para)[()( nsnwb (317)

where Pk is the production limiter

For Specific Dissipation Rate

ww

swnsnbwaww

ww NtildeNtilde-+Ntilde+Ntilde+-=Ntilde+para

para

1)1(2])[()( 21

22 kFSut

T (318)

The first blending function F1 is calculated from

iumlthorn

iumlyacute

uuml

iumlicirc

iumliacute

igrave

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

uacuteuacuteucirc

ugrave

ecircecirceuml

eacute

dividedivideoslash

oumlccedilccedilegrave

aelig=

4

22

21

4500maxmintanh

yCD

k

yy

kF

kw

wrs

w

n

wb (319)

y is the distance to the nearest wall and v is the kinematic viscosity

divideoslash

oumlccedilegrave

aeligNtildeNtilde= -10

2 101

2max ww

rsww kCDk (320)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

31

And Kinematic eddy viscosity

)max( 21

1

SF

kT

wa

an = (321)

where F2 is a blending function which restricts the limiter to the wall boundary and

S is the invariant measure of the strain rate

uacuteuacute

ucirc

ugrave

ecircecirc

euml

eacute

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

dividedivideoslash

oumlccedilccedilegrave

aelig=

2

22

5002maxtanh

w

n

wb yy

kF (322)

33 Brief introduction of simulation process

All aspects of CFD study including modelling meshing solving and post-

processing of the results are briefly introduced in this chapter

331 Pre-processing technique

The speed and accuracy of simulation results depending fundamentally on meshing

and the ability to prepare model geometry correctly

The steps in pre-processing are explained as follows

1) Domain Definitions

The domain definition depends mainly on the physics involved in the problem and

objective of the problem

a) Geometry creation or geometry import

b) Geometry cleanup

i) Removing the parts not necessary for simulation under consideration

ii) Closing the gaps in the geometry

iii) Removing small surfaces curves to make meshing process simple

2)Grid generation

To solve the governing equations the numerical methods like Finite Difference

Finite Volume or Finite Element Method to divide the domain into small parts and

solve the equations on one part at a time are applied

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 38: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

26

Therefore much research has been done to predict the effects of turbulence by

using turbulence models These models account for the effects of turbulence

without the use of a very fine mesh or direct numerical simulation

These turbulence models modify the transport equations by adding averaged and

fluctuating components The transport equations are changed to equations (33) and

(34)

ρ

δ + nabla ∙ (ρU) = 0 (33)

ρ

+ nabla ∙ ρ U ⨂U = minusnablap + nabla ∙ τ minus ρu⨂u + S (34)

The mass equation is not changed but the momentum equation contains extra terms

which are the Reynolds stresses ρ u⨂u and the Reynolds flux ρ u ϕ These

Reynolds stresses need to be modeled by additional equations to obtain closure

Obtaining closure implies that there are a sufficient number of equations to solve for

all the unknowns including the Reynolds stresses and Reynolds fluxes

Various turbulence models provide various ways to obtain closure In this

investigation Shear Stress Transport (SST) model and k-ε model were utilized The

advantage of SST model is that combines the advantages of other turbulence models

(the k-ε Wilcox k-ω) The 3 turbulence models will be discussed briefly

32 Turbulence models

Turbulence models are necessary because we cannot afford big enough computers

to directly capture every scale of motion There are three modelling frameworks

Direct Numerical Simulation (DNS) Reynolds-Averaged Navier-Stokes

Equations(RANS) Large Eddy Simulation(LES)

As RANS method is applied in most productionseveral RANS turbulence models

differ in their usage of wall functions the number of additional variables solved for

and what these variables represent

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

27

Two equation turbulence models

Two-equation turbulence models are widely used in CFD as they give a good

compromise between computational power needed and accuracy The term lsquotwo

equationrsquo refers to the fact that these models solve for the velocity and length scales

using separate transport equations The turbulence velocity scale is obtained by

solving the transport equation The turbulent length scale is estimated from two

properties of the turbulence field namely the turbulent kinetic energy and the

dissipation rate The dissipation rate of the kinetic energy is obtained from its

transport equation The most widely used are k-ε and k-ω two equation models In

the next section the k-ε Wilcox k-ω BSL k-ω and SST models will be briefly

discussed

321 k-ε turbulence model

The k-epsilon model solves for two variables k the turbulent kinetic energy and

epsilon the rate of dissipation of kinetic energy Wall functions are used in this

model so the flow in the buffer region is not simulated The k-epsilon model is very

popular for industrial applications due to its good convergence rate and relatively

low memory requirements It does not very accurately compute flow fields that

exhibit adverse pressure gradients strong curvature to the flow or jet flow It does

perform well for external flow problems around complex geometries

The k-ε model introduces k (m2s2) as the turbulence kinetic energy and ε (m2s3) as

the turbulence eddy dissipation The continuity equation remains the same

ρ

+ nabla ∙ (ρ U) = 0 (35)

The momentum equation changes as shown by equation

δρ

δ + nabla ∙ (ρU⨂U) = minusnablapprime + nabla ∙ μ

nablaU + (nablaU) + S (36)

Where SM is the sum of body forces μeff is the effective viscosity accounting for

turbulence and prsquo is the modified pressure The k-ε model uses the concept of eddy

viscosity giving the equation for effective viscosity as shown by equation

μ

= μ + μ (37)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

28

μt is the turbulence viscosity is linked to the turbulence kinetic energy and

dissipation by the equation

μ = Cμρ

ε (38)

Where Cμ is a constant

The values for k and ε come from the differential transport equations for the

turbulence kinetic energy and the turbulence dissipation rate

The turbulence kinetic energy equation is given as equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus ρε (39)

The turbulence dissipation rate is given by equation

(ρε)

+ nabla ∙ (ρUε) = nabla ∙ μ +

μ

σε nablaε +

ε

(Cε (P + Pε ) minus Cε ρε) (310)

Where Cε1 Cε2 σk σε are constants

Pk is the turbulence production due to viscous forces and is modeled by the

equation

P = μ nablaU ∙ nablaU + nablaU minus

nabla ∙ U 3μ

nabla ∙ U + ρk (311)

If a buoyancy term is added to the previous equation if the full buoyancy model is

used

The buoyancy term Pkb is modeled as

P = minusμ

ρσρg ∙ nablaρ (312)

322 Wilcox k-ω turbulence model

The k-omega model is similar to k-epsilon instead however it solves for omega mdash

the specific rate of dissipation of kinetic energy It also uses wall functions and

therefore has comparable memory requirements It has more difficulty converging

and is quite sensitive to the initial guess at the solution Hence the k-epsilon model

is often used first to find an initial condition for solving the k-omega model The k-

omega model is useful in many cases where the k-epsilon model is not accurate

such as internal flows flows that exhibit strong curvature separated flows and jets

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

29

This model has an advantage over the k-ε model where it does not involve complex

linear damping functions for near wall calculations at low Reynolds numbers

The k-ω model assumes that the turbulence viscosity is related to the turbulence

kinetic energy k and the turbulent frequency ω by the equation

μ = ρ

ω (313)

The transport equation for k is given by the equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus βprimeρ k ω (314)

The transport equation for ω is shown as equation

(ρω)

δ + nabla ∙ (ρ U ω) = nabla ∙ μ +

μ

μω

nablaω + αω

P + Pω minus β ρ ω (315)

The production rate of turbulence (Pk) is calculated as shown previously in the k-ε

section

The recommended values for the constants in the above equations are

βprime = 009

α = 59

β = 0075

σ = 2

σω = 2

The Reynolds stress tensor ρ u⨂u is calculated by

minusρ u⨂u = μ nablaU + (nablaU) minus

part ρk + μ

nabla ∙ U (316)

323 Shear stress transport model

The disadvantage of the Wilcox model is the strong sensitivity to free-stream

conditions Therefore a blending of the k-ω model near the surface and the k-ε in

the outer region was made by Menter [23] which resulted in the formulation of the

BSL k-ω turbulence model It consists of a transformation of the k-ε model to a k-ω

formulation and subsequently adding the resulting equations The Wilcox model is

multiplied by a blending function F1 and the transformed k-ε by another function 1-

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

30

F1 F1 is a function of wall distance (being the value of one near the surface and zero

outside the boundary layer) Outside the boundary and on the edge of the boundary

layer the standard k-ε model is used [24]

SST model is a combination of the k-epsilon in the free stream and the k-omega

models near the walls It does not use wall functions and tends to be most accurate

when solving the flow near the wall The SST model does not always converge to

the solution quickly so the k-epsilon or k-omega models are often solved first to

give good initial conditions

The k-ω based Shear Stress Transport (SST) model of Menter was applied for

turbulence treatment The transport equations for the SST model are expressed

below where the turbulent kinetic energy lsquokrsquo and turbulent frequency or dissipation

per unit turbulent kinetic energy are computed by using the following relations

For Turbulence Kinetic Energy

kkPukt

kTkk Ntilde+Ntilde+-=Ntilde+

para

para)[()( nsnwb (317)

where Pk is the production limiter

For Specific Dissipation Rate

ww

swnsnbwaww

ww NtildeNtilde-+Ntilde+Ntilde+-=Ntilde+para

para

1)1(2])[()( 21

22 kFSut

T (318)

The first blending function F1 is calculated from

iumlthorn

iumlyacute

uuml

iumlicirc

iumliacute

igrave

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

uacuteuacuteucirc

ugrave

ecircecirceuml

eacute

dividedivideoslash

oumlccedilccedilegrave

aelig=

4

22

21

4500maxmintanh

yCD

k

yy

kF

kw

wrs

w

n

wb (319)

y is the distance to the nearest wall and v is the kinematic viscosity

divideoslash

oumlccedilegrave

aeligNtildeNtilde= -10

2 101

2max ww

rsww kCDk (320)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

31

And Kinematic eddy viscosity

)max( 21

1

SF

kT

wa

an = (321)

where F2 is a blending function which restricts the limiter to the wall boundary and

S is the invariant measure of the strain rate

uacuteuacute

ucirc

ugrave

ecircecirc

euml

eacute

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

dividedivideoslash

oumlccedilccedilegrave

aelig=

2

22

5002maxtanh

w

n

wb yy

kF (322)

33 Brief introduction of simulation process

All aspects of CFD study including modelling meshing solving and post-

processing of the results are briefly introduced in this chapter

331 Pre-processing technique

The speed and accuracy of simulation results depending fundamentally on meshing

and the ability to prepare model geometry correctly

The steps in pre-processing are explained as follows

1) Domain Definitions

The domain definition depends mainly on the physics involved in the problem and

objective of the problem

a) Geometry creation or geometry import

b) Geometry cleanup

i) Removing the parts not necessary for simulation under consideration

ii) Closing the gaps in the geometry

iii) Removing small surfaces curves to make meshing process simple

2)Grid generation

To solve the governing equations the numerical methods like Finite Difference

Finite Volume or Finite Element Method to divide the domain into small parts and

solve the equations on one part at a time are applied

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 39: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

27

Two equation turbulence models

Two-equation turbulence models are widely used in CFD as they give a good

compromise between computational power needed and accuracy The term lsquotwo

equationrsquo refers to the fact that these models solve for the velocity and length scales

using separate transport equations The turbulence velocity scale is obtained by

solving the transport equation The turbulent length scale is estimated from two

properties of the turbulence field namely the turbulent kinetic energy and the

dissipation rate The dissipation rate of the kinetic energy is obtained from its

transport equation The most widely used are k-ε and k-ω two equation models In

the next section the k-ε Wilcox k-ω BSL k-ω and SST models will be briefly

discussed

321 k-ε turbulence model

The k-epsilon model solves for two variables k the turbulent kinetic energy and

epsilon the rate of dissipation of kinetic energy Wall functions are used in this

model so the flow in the buffer region is not simulated The k-epsilon model is very

popular for industrial applications due to its good convergence rate and relatively

low memory requirements It does not very accurately compute flow fields that

exhibit adverse pressure gradients strong curvature to the flow or jet flow It does

perform well for external flow problems around complex geometries

The k-ε model introduces k (m2s2) as the turbulence kinetic energy and ε (m2s3) as

the turbulence eddy dissipation The continuity equation remains the same

ρ

+ nabla ∙ (ρ U) = 0 (35)

The momentum equation changes as shown by equation

δρ

δ + nabla ∙ (ρU⨂U) = minusnablapprime + nabla ∙ μ

nablaU + (nablaU) + S (36)

Where SM is the sum of body forces μeff is the effective viscosity accounting for

turbulence and prsquo is the modified pressure The k-ε model uses the concept of eddy

viscosity giving the equation for effective viscosity as shown by equation

μ

= μ + μ (37)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

28

μt is the turbulence viscosity is linked to the turbulence kinetic energy and

dissipation by the equation

μ = Cμρ

ε (38)

Where Cμ is a constant

The values for k and ε come from the differential transport equations for the

turbulence kinetic energy and the turbulence dissipation rate

The turbulence kinetic energy equation is given as equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus ρε (39)

The turbulence dissipation rate is given by equation

(ρε)

+ nabla ∙ (ρUε) = nabla ∙ μ +

μ

σε nablaε +

ε

(Cε (P + Pε ) minus Cε ρε) (310)

Where Cε1 Cε2 σk σε are constants

Pk is the turbulence production due to viscous forces and is modeled by the

equation

P = μ nablaU ∙ nablaU + nablaU minus

nabla ∙ U 3μ

nabla ∙ U + ρk (311)

If a buoyancy term is added to the previous equation if the full buoyancy model is

used

The buoyancy term Pkb is modeled as

P = minusμ

ρσρg ∙ nablaρ (312)

322 Wilcox k-ω turbulence model

The k-omega model is similar to k-epsilon instead however it solves for omega mdash

the specific rate of dissipation of kinetic energy It also uses wall functions and

therefore has comparable memory requirements It has more difficulty converging

and is quite sensitive to the initial guess at the solution Hence the k-epsilon model

is often used first to find an initial condition for solving the k-omega model The k-

omega model is useful in many cases where the k-epsilon model is not accurate

such as internal flows flows that exhibit strong curvature separated flows and jets

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

29

This model has an advantage over the k-ε model where it does not involve complex

linear damping functions for near wall calculations at low Reynolds numbers

The k-ω model assumes that the turbulence viscosity is related to the turbulence

kinetic energy k and the turbulent frequency ω by the equation

μ = ρ

ω (313)

The transport equation for k is given by the equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus βprimeρ k ω (314)

The transport equation for ω is shown as equation

(ρω)

δ + nabla ∙ (ρ U ω) = nabla ∙ μ +

μ

μω

nablaω + αω

P + Pω minus β ρ ω (315)

The production rate of turbulence (Pk) is calculated as shown previously in the k-ε

section

The recommended values for the constants in the above equations are

βprime = 009

α = 59

β = 0075

σ = 2

σω = 2

The Reynolds stress tensor ρ u⨂u is calculated by

minusρ u⨂u = μ nablaU + (nablaU) minus

part ρk + μ

nabla ∙ U (316)

323 Shear stress transport model

The disadvantage of the Wilcox model is the strong sensitivity to free-stream

conditions Therefore a blending of the k-ω model near the surface and the k-ε in

the outer region was made by Menter [23] which resulted in the formulation of the

BSL k-ω turbulence model It consists of a transformation of the k-ε model to a k-ω

formulation and subsequently adding the resulting equations The Wilcox model is

multiplied by a blending function F1 and the transformed k-ε by another function 1-

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

30

F1 F1 is a function of wall distance (being the value of one near the surface and zero

outside the boundary layer) Outside the boundary and on the edge of the boundary

layer the standard k-ε model is used [24]

SST model is a combination of the k-epsilon in the free stream and the k-omega

models near the walls It does not use wall functions and tends to be most accurate

when solving the flow near the wall The SST model does not always converge to

the solution quickly so the k-epsilon or k-omega models are often solved first to

give good initial conditions

The k-ω based Shear Stress Transport (SST) model of Menter was applied for

turbulence treatment The transport equations for the SST model are expressed

below where the turbulent kinetic energy lsquokrsquo and turbulent frequency or dissipation

per unit turbulent kinetic energy are computed by using the following relations

For Turbulence Kinetic Energy

kkPukt

kTkk Ntilde+Ntilde+-=Ntilde+

para

para)[()( nsnwb (317)

where Pk is the production limiter

For Specific Dissipation Rate

ww

swnsnbwaww

ww NtildeNtilde-+Ntilde+Ntilde+-=Ntilde+para

para

1)1(2])[()( 21

22 kFSut

T (318)

The first blending function F1 is calculated from

iumlthorn

iumlyacute

uuml

iumlicirc

iumliacute

igrave

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

uacuteuacuteucirc

ugrave

ecircecirceuml

eacute

dividedivideoslash

oumlccedilccedilegrave

aelig=

4

22

21

4500maxmintanh

yCD

k

yy

kF

kw

wrs

w

n

wb (319)

y is the distance to the nearest wall and v is the kinematic viscosity

divideoslash

oumlccedilegrave

aeligNtildeNtilde= -10

2 101

2max ww

rsww kCDk (320)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

31

And Kinematic eddy viscosity

)max( 21

1

SF

kT

wa

an = (321)

where F2 is a blending function which restricts the limiter to the wall boundary and

S is the invariant measure of the strain rate

uacuteuacute

ucirc

ugrave

ecircecirc

euml

eacute

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

dividedivideoslash

oumlccedilccedilegrave

aelig=

2

22

5002maxtanh

w

n

wb yy

kF (322)

33 Brief introduction of simulation process

All aspects of CFD study including modelling meshing solving and post-

processing of the results are briefly introduced in this chapter

331 Pre-processing technique

The speed and accuracy of simulation results depending fundamentally on meshing

and the ability to prepare model geometry correctly

The steps in pre-processing are explained as follows

1) Domain Definitions

The domain definition depends mainly on the physics involved in the problem and

objective of the problem

a) Geometry creation or geometry import

b) Geometry cleanup

i) Removing the parts not necessary for simulation under consideration

ii) Closing the gaps in the geometry

iii) Removing small surfaces curves to make meshing process simple

2)Grid generation

To solve the governing equations the numerical methods like Finite Difference

Finite Volume or Finite Element Method to divide the domain into small parts and

solve the equations on one part at a time are applied

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 40: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

28

μt is the turbulence viscosity is linked to the turbulence kinetic energy and

dissipation by the equation

μ = Cμρ

ε (38)

Where Cμ is a constant

The values for k and ε come from the differential transport equations for the

turbulence kinetic energy and the turbulence dissipation rate

The turbulence kinetic energy equation is given as equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus ρε (39)

The turbulence dissipation rate is given by equation

(ρε)

+ nabla ∙ (ρUε) = nabla ∙ μ +

μ

σε nablaε +

ε

(Cε (P + Pε ) minus Cε ρε) (310)

Where Cε1 Cε2 σk σε are constants

Pk is the turbulence production due to viscous forces and is modeled by the

equation

P = μ nablaU ∙ nablaU + nablaU minus

nabla ∙ U 3μ

nabla ∙ U + ρk (311)

If a buoyancy term is added to the previous equation if the full buoyancy model is

used

The buoyancy term Pkb is modeled as

P = minusμ

ρσρg ∙ nablaρ (312)

322 Wilcox k-ω turbulence model

The k-omega model is similar to k-epsilon instead however it solves for omega mdash

the specific rate of dissipation of kinetic energy It also uses wall functions and

therefore has comparable memory requirements It has more difficulty converging

and is quite sensitive to the initial guess at the solution Hence the k-epsilon model

is often used first to find an initial condition for solving the k-omega model The k-

omega model is useful in many cases where the k-epsilon model is not accurate

such as internal flows flows that exhibit strong curvature separated flows and jets

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

29

This model has an advantage over the k-ε model where it does not involve complex

linear damping functions for near wall calculations at low Reynolds numbers

The k-ω model assumes that the turbulence viscosity is related to the turbulence

kinetic energy k and the turbulent frequency ω by the equation

μ = ρ

ω (313)

The transport equation for k is given by the equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus βprimeρ k ω (314)

The transport equation for ω is shown as equation

(ρω)

δ + nabla ∙ (ρ U ω) = nabla ∙ μ +

μ

μω

nablaω + αω

P + Pω minus β ρ ω (315)

The production rate of turbulence (Pk) is calculated as shown previously in the k-ε

section

The recommended values for the constants in the above equations are

βprime = 009

α = 59

β = 0075

σ = 2

σω = 2

The Reynolds stress tensor ρ u⨂u is calculated by

minusρ u⨂u = μ nablaU + (nablaU) minus

part ρk + μ

nabla ∙ U (316)

323 Shear stress transport model

The disadvantage of the Wilcox model is the strong sensitivity to free-stream

conditions Therefore a blending of the k-ω model near the surface and the k-ε in

the outer region was made by Menter [23] which resulted in the formulation of the

BSL k-ω turbulence model It consists of a transformation of the k-ε model to a k-ω

formulation and subsequently adding the resulting equations The Wilcox model is

multiplied by a blending function F1 and the transformed k-ε by another function 1-

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

30

F1 F1 is a function of wall distance (being the value of one near the surface and zero

outside the boundary layer) Outside the boundary and on the edge of the boundary

layer the standard k-ε model is used [24]

SST model is a combination of the k-epsilon in the free stream and the k-omega

models near the walls It does not use wall functions and tends to be most accurate

when solving the flow near the wall The SST model does not always converge to

the solution quickly so the k-epsilon or k-omega models are often solved first to

give good initial conditions

The k-ω based Shear Stress Transport (SST) model of Menter was applied for

turbulence treatment The transport equations for the SST model are expressed

below where the turbulent kinetic energy lsquokrsquo and turbulent frequency or dissipation

per unit turbulent kinetic energy are computed by using the following relations

For Turbulence Kinetic Energy

kkPukt

kTkk Ntilde+Ntilde+-=Ntilde+

para

para)[()( nsnwb (317)

where Pk is the production limiter

For Specific Dissipation Rate

ww

swnsnbwaww

ww NtildeNtilde-+Ntilde+Ntilde+-=Ntilde+para

para

1)1(2])[()( 21

22 kFSut

T (318)

The first blending function F1 is calculated from

iumlthorn

iumlyacute

uuml

iumlicirc

iumliacute

igrave

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

uacuteuacuteucirc

ugrave

ecircecirceuml

eacute

dividedivideoslash

oumlccedilccedilegrave

aelig=

4

22

21

4500maxmintanh

yCD

k

yy

kF

kw

wrs

w

n

wb (319)

y is the distance to the nearest wall and v is the kinematic viscosity

divideoslash

oumlccedilegrave

aeligNtildeNtilde= -10

2 101

2max ww

rsww kCDk (320)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

31

And Kinematic eddy viscosity

)max( 21

1

SF

kT

wa

an = (321)

where F2 is a blending function which restricts the limiter to the wall boundary and

S is the invariant measure of the strain rate

uacuteuacute

ucirc

ugrave

ecircecirc

euml

eacute

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

dividedivideoslash

oumlccedilccedilegrave

aelig=

2

22

5002maxtanh

w

n

wb yy

kF (322)

33 Brief introduction of simulation process

All aspects of CFD study including modelling meshing solving and post-

processing of the results are briefly introduced in this chapter

331 Pre-processing technique

The speed and accuracy of simulation results depending fundamentally on meshing

and the ability to prepare model geometry correctly

The steps in pre-processing are explained as follows

1) Domain Definitions

The domain definition depends mainly on the physics involved in the problem and

objective of the problem

a) Geometry creation or geometry import

b) Geometry cleanup

i) Removing the parts not necessary for simulation under consideration

ii) Closing the gaps in the geometry

iii) Removing small surfaces curves to make meshing process simple

2)Grid generation

To solve the governing equations the numerical methods like Finite Difference

Finite Volume or Finite Element Method to divide the domain into small parts and

solve the equations on one part at a time are applied

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 41: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

29

This model has an advantage over the k-ε model where it does not involve complex

linear damping functions for near wall calculations at low Reynolds numbers

The k-ω model assumes that the turbulence viscosity is related to the turbulence

kinetic energy k and the turbulent frequency ω by the equation

μ = ρ

ω (313)

The transport equation for k is given by the equation

(ρ )

+ nabla ∙ (ρUk) = nabla ∙ μ +

μ

σ nablak + P + P minus βprimeρ k ω (314)

The transport equation for ω is shown as equation

(ρω)

δ + nabla ∙ (ρ U ω) = nabla ∙ μ +

μ

μω

nablaω + αω

P + Pω minus β ρ ω (315)

The production rate of turbulence (Pk) is calculated as shown previously in the k-ε

section

The recommended values for the constants in the above equations are

βprime = 009

α = 59

β = 0075

σ = 2

σω = 2

The Reynolds stress tensor ρ u⨂u is calculated by

minusρ u⨂u = μ nablaU + (nablaU) minus

part ρk + μ

nabla ∙ U (316)

323 Shear stress transport model

The disadvantage of the Wilcox model is the strong sensitivity to free-stream

conditions Therefore a blending of the k-ω model near the surface and the k-ε in

the outer region was made by Menter [23] which resulted in the formulation of the

BSL k-ω turbulence model It consists of a transformation of the k-ε model to a k-ω

formulation and subsequently adding the resulting equations The Wilcox model is

multiplied by a blending function F1 and the transformed k-ε by another function 1-

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

30

F1 F1 is a function of wall distance (being the value of one near the surface and zero

outside the boundary layer) Outside the boundary and on the edge of the boundary

layer the standard k-ε model is used [24]

SST model is a combination of the k-epsilon in the free stream and the k-omega

models near the walls It does not use wall functions and tends to be most accurate

when solving the flow near the wall The SST model does not always converge to

the solution quickly so the k-epsilon or k-omega models are often solved first to

give good initial conditions

The k-ω based Shear Stress Transport (SST) model of Menter was applied for

turbulence treatment The transport equations for the SST model are expressed

below where the turbulent kinetic energy lsquokrsquo and turbulent frequency or dissipation

per unit turbulent kinetic energy are computed by using the following relations

For Turbulence Kinetic Energy

kkPukt

kTkk Ntilde+Ntilde+-=Ntilde+

para

para)[()( nsnwb (317)

where Pk is the production limiter

For Specific Dissipation Rate

ww

swnsnbwaww

ww NtildeNtilde-+Ntilde+Ntilde+-=Ntilde+para

para

1)1(2])[()( 21

22 kFSut

T (318)

The first blending function F1 is calculated from

iumlthorn

iumlyacute

uuml

iumlicirc

iumliacute

igrave

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

uacuteuacuteucirc

ugrave

ecircecirceuml

eacute

dividedivideoslash

oumlccedilccedilegrave

aelig=

4

22

21

4500maxmintanh

yCD

k

yy

kF

kw

wrs

w

n

wb (319)

y is the distance to the nearest wall and v is the kinematic viscosity

divideoslash

oumlccedilegrave

aeligNtildeNtilde= -10

2 101

2max ww

rsww kCDk (320)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

31

And Kinematic eddy viscosity

)max( 21

1

SF

kT

wa

an = (321)

where F2 is a blending function which restricts the limiter to the wall boundary and

S is the invariant measure of the strain rate

uacuteuacute

ucirc

ugrave

ecircecirc

euml

eacute

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

dividedivideoslash

oumlccedilccedilegrave

aelig=

2

22

5002maxtanh

w

n

wb yy

kF (322)

33 Brief introduction of simulation process

All aspects of CFD study including modelling meshing solving and post-

processing of the results are briefly introduced in this chapter

331 Pre-processing technique

The speed and accuracy of simulation results depending fundamentally on meshing

and the ability to prepare model geometry correctly

The steps in pre-processing are explained as follows

1) Domain Definitions

The domain definition depends mainly on the physics involved in the problem and

objective of the problem

a) Geometry creation or geometry import

b) Geometry cleanup

i) Removing the parts not necessary for simulation under consideration

ii) Closing the gaps in the geometry

iii) Removing small surfaces curves to make meshing process simple

2)Grid generation

To solve the governing equations the numerical methods like Finite Difference

Finite Volume or Finite Element Method to divide the domain into small parts and

solve the equations on one part at a time are applied

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 42: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

30

F1 F1 is a function of wall distance (being the value of one near the surface and zero

outside the boundary layer) Outside the boundary and on the edge of the boundary

layer the standard k-ε model is used [24]

SST model is a combination of the k-epsilon in the free stream and the k-omega

models near the walls It does not use wall functions and tends to be most accurate

when solving the flow near the wall The SST model does not always converge to

the solution quickly so the k-epsilon or k-omega models are often solved first to

give good initial conditions

The k-ω based Shear Stress Transport (SST) model of Menter was applied for

turbulence treatment The transport equations for the SST model are expressed

below where the turbulent kinetic energy lsquokrsquo and turbulent frequency or dissipation

per unit turbulent kinetic energy are computed by using the following relations

For Turbulence Kinetic Energy

kkPukt

kTkk Ntilde+Ntilde+-=Ntilde+

para

para)[()( nsnwb (317)

where Pk is the production limiter

For Specific Dissipation Rate

ww

swnsnbwaww

ww NtildeNtilde-+Ntilde+Ntilde+-=Ntilde+para

para

1)1(2])[()( 21

22 kFSut

T (318)

The first blending function F1 is calculated from

iumlthorn

iumlyacute

uuml

iumlicirc

iumliacute

igrave

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

uacuteuacuteucirc

ugrave

ecircecirceuml

eacute

dividedivideoslash

oumlccedilccedilegrave

aelig=

4

22

21

4500maxmintanh

yCD

k

yy

kF

kw

wrs

w

n

wb (319)

y is the distance to the nearest wall and v is the kinematic viscosity

divideoslash

oumlccedilegrave

aeligNtildeNtilde= -10

2 101

2max ww

rsww kCDk (320)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

31

And Kinematic eddy viscosity

)max( 21

1

SF

kT

wa

an = (321)

where F2 is a blending function which restricts the limiter to the wall boundary and

S is the invariant measure of the strain rate

uacuteuacute

ucirc

ugrave

ecircecirc

euml

eacute

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

dividedivideoslash

oumlccedilccedilegrave

aelig=

2

22

5002maxtanh

w

n

wb yy

kF (322)

33 Brief introduction of simulation process

All aspects of CFD study including modelling meshing solving and post-

processing of the results are briefly introduced in this chapter

331 Pre-processing technique

The speed and accuracy of simulation results depending fundamentally on meshing

and the ability to prepare model geometry correctly

The steps in pre-processing are explained as follows

1) Domain Definitions

The domain definition depends mainly on the physics involved in the problem and

objective of the problem

a) Geometry creation or geometry import

b) Geometry cleanup

i) Removing the parts not necessary for simulation under consideration

ii) Closing the gaps in the geometry

iii) Removing small surfaces curves to make meshing process simple

2)Grid generation

To solve the governing equations the numerical methods like Finite Difference

Finite Volume or Finite Element Method to divide the domain into small parts and

solve the equations on one part at a time are applied

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 43: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

31

And Kinematic eddy viscosity

)max( 21

1

SF

kT

wa

an = (321)

where F2 is a blending function which restricts the limiter to the wall boundary and

S is the invariant measure of the strain rate

uacuteuacute

ucirc

ugrave

ecircecirc

euml

eacute

iumlthorn

iumlyacuteuuml

iumlicirc

iumliacuteigrave

dividedivideoslash

oumlccedilccedilegrave

aelig=

2

22

5002maxtanh

w

n

wb yy

kF (322)

33 Brief introduction of simulation process

All aspects of CFD study including modelling meshing solving and post-

processing of the results are briefly introduced in this chapter

331 Pre-processing technique

The speed and accuracy of simulation results depending fundamentally on meshing

and the ability to prepare model geometry correctly

The steps in pre-processing are explained as follows

1) Domain Definitions

The domain definition depends mainly on the physics involved in the problem and

objective of the problem

a) Geometry creation or geometry import

b) Geometry cleanup

i) Removing the parts not necessary for simulation under consideration

ii) Closing the gaps in the geometry

iii) Removing small surfaces curves to make meshing process simple

2)Grid generation

To solve the governing equations the numerical methods like Finite Difference

Finite Volume or Finite Element Method to divide the domain into small parts and

solve the equations on one part at a time are applied

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 44: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

32

3) Labeling (Putting Tags)

Applying boundary conditions in pre-processing tool is the process of putting

appropriate tags (labels) on specific boundaries There are typically two types of

boundary conditions

a) Surface boundary conditions inlet outlet wall etc

b)Domain(regions) boundary conditions to specify the solid region or fluid

region

The data provided on boundaries (eg velocity at inlet pressure at outlet etc) is

going to be used during solution of the governing equations

332 Solving the simulation

Once the models have the various variables and boundary conditions specified the

following step is to solve for the solutions using a solver This study used the CFX-

Solver 130 software as the solver

There are several techniques that can be used to solve the governing equations that

were described earlier CFX uses the Finite Volume Method (FVM) approach to

solve the required equations

333 Post processing

The post processing proceed after the required simulation is completed Post

Processing refers to processing the result of the simulation by a number of ways

Generating a visual representation of various flow variables over the

geometry

Generating animations of flow vectors

Analyzing flow variable distributions

Processing of results for output onto a chart

The ANSYS software provides CFX-Post as a Post processor function

These steps combined with the specific model case will be further discussed in the

next chapter

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 45: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

33

Chapter 4 Description of Model Cases

This chapter gives methodology details of both cases a scaled rectangular sump

model and a full size sump model with a mixed flow pump installed Meanwhile

the arrangement of PIV measurements is introduced

41 Creating the geometry

The three dimensional models of the geometry for the sump fluid domain were

made using the commercial code Unigraphics NX 60 according to the available

design

411 Geometry of the scaled sump model

The numerical simulations of the case 1 were performed both in one phase model

and two-phase model Furthermore a serial of sump models with devices installed

for free surface vortex were studied

Fig 41 Geometry of scaled sump model

Table 41 Designed specifications of scaled sump model

Volume flow rate (m3h) 882

Inlet bell outside diameter (D mm) 2708

Pipe inner diameter (d mm) 122

Inlet bell submergence (S mm) 510

Distance between the inlet bell and floor (C mm) 100

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 46: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

34

Fig 41 shows the shape of sump model The width of intake channel is 500mm and

the center of suction bell is located at 201mm 250mm and 100mm from rear wall

side wall and bottom respectively The model (scale ratio 172 to prototype) was

applying the Froude Number similarity principle and all the dimensions are

confirmed to chapter 2

eg S=D(10+23Fr) according to ANSIHI

While Fr=Vb(gD)05

Solving we got recommend minimum S=434mm <510mm

As illustrated in the Schematic diagram of Fig 42 and Table 42 the location and

dimensions of anti free surface vortex device (curtain wall or square bar) are

confirmed

(a) Sump model with curtain wall

(b) Sump model with square bar

Fig 42 Schematic diagram of sump with anti free surface vortex devices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 47: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

35

Table 42 Parameters of anti free surface vortex devices

Case Distance to sump back wall (mm)

Submergence of the bottom (mm)

Dimensions below water surface (mm)

Curtain wall 1

760 100 100times60times500

Curtain wall 2

760 200 200times60times500

Curtain wall 3

760 300 300times60times500

Square bar 1

760 200 100times50times500

Square bar 2

760 300 100times50times500

412 Geometry of the full size pump sump model

1) Model of the sump layout

Fig 43 shows the 3D overall sump model with the mixed flow pump installed

Fig 43 Overall 3D drawing of the pump sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 48: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

36

As illustrated in Fig 44 (a) the total flow direction length of the model is 30m

with a long channel of 26m in width 18m in length and a baffle about 86m

distance from the intake bell center The width of the intake channel is 4m and the

center of the intake bell is located at 18m 2m and 07m from rear wall side wall

and bottom respectively All the dimensions referred the recommended values of

Chapter 2

(a) Top view

(b) Side view

Fig 44 2D drawing of the sump sketch

2) Model of the mixed flow pump

Generally pump is a mechanic equipment which is required to lift liquid from low

level to high level or to flow liquid from low pressure area to high pressure area

Principally pump converts mechanic energy of motor into fluid flow energy Pumps

are commonly rated by power flow rate and total head The head can be simplified

as the number of meters the pump can raise or lower a column of water at

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 49: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

37

atmospheric pressure From an initial design point of view engineers often use a

quantity termed the specific speed to identify the most suitable pump type for a

particular combination of flow rate and head

The blade profile of the mixed flow pump is created in BladeGen Fig 45 shows

the 3D geometry and meridional geometry of the mixed flow pump and Table 43

provides the design specifications of the mixed flow pump

(a) 3D geometry

(b) Meridional geometry

Fig 45 Shape of the mixed flow pump

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 50: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

38

Table 43 Design specifications of the mixed flow pump

Volume flow rate (m3h) 21700

Rotational speed (rpm) 423

Total head (m) 23

Number of impeller blade 5

Number of diffuser vane 9

Impeller inlet diameter (m) 1096

3) Sump model with the submerged AVD

Fig 46 shows the sump model with AVD installed The AVD is designed to

suppress the submerged vortex that occurs near the side wall and bottom

Fig 46 Sump model with the submerged AVD

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 51: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

39

Fig 47 Submerged AVD dimensions

The trident shaped AVD installed under the pump intake is made up of three wall

fillets (800mm height) and one central splitter (200mm height) The design of this

AVD referred to Appendix A of the standard ANSIHI 98

42 Mesh generation

Mesh generation refers to the creation of a numerical domain over a certain

geometry in which the computer can use to solve the required equations After the

completion of the geometry a number of computer based software may be used to

generate the numerical domain In this study ICEM CFD was used to generate the

mesh of the sump fluid domain while for the mixed flow pump grid generation

TurboGrid is used along with BladeGen

ICEM CFD allows for the generation of several types of meshes including

tetrahedral hexahedral or even hybrid meshes Tetrahedral meshes are generally

less time consuming to build whereas hexahedral meshes provide more accurate

results that the mesh quality is high

In addition to this ICEM CFD allows for the generation of structured or

unstructured meshes Structured meshes have cells that are regular in shape and

each grid point is uniquely identified by indices and coordinates Unstructured

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 52: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

40

meshes contain cells that are not necessarily regular in shape and have grid points in

no particular ordering The main advantage of unstructured meshes is the flexibility

it provides in the generation of a computational grid in complex geometries

The computational grid of all the geometries in this investigation was made of

unstructured hexahedral volumes to ensure accurate results

The flow domain was divided into a number of smaller regions for two reasons

Firstly it improved mesh quality and secondly it enabled named selections to be

created The mesh quality was noticeably improved by separating the flow domain

in regions free from complex geometrical shapes This is because a hexahedral

mesh was easier to be applied for uniform volumes Named selections were utilized

to specify boundary conditions and to facilitate results viewing These domains

were meshed separately and imported to CFX-Pre forming one computational

domain

(a) Overall mesh generation for single phase model

(b) Overall mesh generation for two-phase model

Fig 48 Overall mesh generation for scaled sump models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 53: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

41

As the overall mesh shows the scaled sump model is divided into several parts to

employ hexa-hedral mesh The total nodes for the single phase model is about 165

millions and for the two-phase model is about 228 millions

Fig 49 Overall sump fluid domain mesh for the pump sump model

a) Tetrahedral mesh of Bell mouth b) Hexahedral mesh of the mixed flow pump

Fig 410 Grid details of the pump sump model

Fig 49 shows the overall mesh of the sump domain as the complex fluid domain

was divided into 5 parts the girds were generated according to the geometry

separately The grid details of bell mouth and the mixed flow pump are shown in

Fig 410

The mesh information of various parts is summarized in Table 44

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 54: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

42

Table 44 Mesh information for the pump sump model

Number of nodes Mesh type

Impeller domain 800000 Hexa-hedral

Diffuser domain 2370000 Hexa-hedral

Total original model 5040000

Hybrid Mesh model with AVD 6290000

43 Numerical approach

The numerical analysis of three-dimensional steady-state turbulent flow based

on the Reynolds-averaged Navier-Stokes equations has been performed to get good

convergence result to predict the flow condition The physics of the simulation

domain was defined in CFX-Pre the preprocessing module of ANSYS CFX All

simulations were performed using ANSYS CFX with HP MPI Distributed Parallel

model

A free surface is an interface between a liquid and a gas in which the gas can only

apply a pressure on the liquid Free surfaces are generally excellent approximations

when the ratio of liquid to gas densities is large Most CFD codes include the

Volume of Fluid (VOF) technique which was originally developed by Hirt and

Nichols [25]

The VOF method consists of three ingredients a scheme to locate the surface an

algorithm to track the surface as a sharp interface moving through a computational

grid and a means of applying boundary conditions at the surface The basic concept

of the VOF method is the definition of a non-dimensional scalar quantity value

which represents the fraction of the mesh cell volume occupied by the continuous

phase eg the liquid phase

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 55: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

43

(a) Boundaries for single phase model

(b) Boundaries for two-phase model

Fig 411 Boundaries of the scaled sump

For the single phase model the fluid (water) was considered incompressible and

the free surface of water was specified as a free slip wall For the two-phase model

both the phases (air and water) were distinctly defined by giving initial volume

fraction as either one or zero Initial free surface was the separation of the sump

water domain and air domain The top of the air domain is assigned as opening

condition with absolute pressure equal to atmospheric pressure Buoyant option

with density difference fluid buoyant model is assigned to all domains Buoyant

reference density of air is assigned to all domains VOF is used to model the

interaction between the two phases It is a simple multiphase model that is well

suited to simulate flows of several fluids on numerical grids capable to resolve the

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 56: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

44

interface between the mixturersquos phases The two phases are assumed to share the

same velocity pressure and temperature fields The initial condition for the volume

fraction of phase is given as Fig 411 (b) SST (Shear stress transport) turbulence

model was selected for both cases The total pressure was prescribed at the inflow

boundary whereas the mass flowrate was specified at the outlet section for single

phase model and normal speed was specified as outlet boundary for the two-phase

model

For the full size pump sump model the sump and stator fluid domain are stationary

components while pump rotor is the rotating component As for the boundary

conditions the total pressure was prescribed at the inflow boundary whereas the

mass flowrate was specified at the outlet section of the pump stage All

computations have been carried out using water at 25 ˚C as a working fluid The

flow regime at both inlet and outlet boundaries was specified as subsonic All

the outer walls of the flow region and the internal walls (pump columns below

free surface) were specified as the boundary type wall with flow condition as -

no slip The free surface of fluid was specified as a free slip wall The ANSYS

CFX-Solver module of ANSYS CFX was used to obtain the solution of the CFD

problem The solver control parameters were specified in the form of solution

scheme and convergence criteria High Resolution was specified for the solution

while for convergence the residual target for RMS values was specified as 1times10-5

For the cavitation phenomenon study the working fluid is changed to two phases

(water and water vapor at 25) Rayleigh Plesset model is selected for the

cavitation model and the required parameters Saturation Pressure is set to the value

of 3574Pa the mean nucleation site diameter is specified as the default value of

20e-6m To solve the cavitation problem the converged solution of a model with

cavitation model turned off should be first performed The initial condition in fluid

setting should use water volume fraction of 1 and water vapor volume fraction of 0

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 57: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

45

44 Experimental setup

Particle Image Velocimetry (PIV) is a non-intrusive laser optical measurement

technique for research and diagnostics into flow turbulence micro fluidics spray

atomization and combustion processes etc It is supposed to obtain

instantaneous velocity measurements and related properties in fluids The fluid

is seeded with tracer particles which for sufficiently small particles are assumed to

faithfully follow the flow dynamics The fluid with entrained particles is illuminated

so that particles are visible

The total size of the pump chamber is 800mm in height 500mm in width and

4000mm in length The sump model experiment parameters were according to the

scaled model geometry Fig 412 shows experimental arrangement including PIV

system for the scaled sump model Poly vinyl chloride tracer particles with an

average diameter of 100μm were seeded into the water Diode-pumped solid-state

continuous laser with max 4W was used to illuminate the particles The motion of

the particles was captured by a high-speed camera (1024times1024 pixel max fps

10000) During the experiment the camera options and laser output were adjusted

to obtain suitable records CACTUS 33 was used for post-processing of record data

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 58: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

46

(a) Sump test model

(b) PIV system arrangement

Fig 412 Experimental setup for the scaled sump model

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 59: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

47

Chapter 5 Results and Discussion

In this section two model cases results are discussed The main objective of first

case is the prediction of flow field in the sump both by numerical simulation and

PIV results Comparisons in the same geometry configuration between single one

phase (water) model and two-phase (air water) model are made

In the Case 2 hydraulic performances flow characteristics and the effect of an

anti-vortex device (AVD) for the submerged vortex has been studied Furthermore

cavitation phenomenon is analyzed

51 Results of the scaled sump model

511 Numerical simulation vortex check

Comparisons of numerical simulation between single phase model and two-phase

model are predicted here In this section all left side figures are the single phase

model results while right side figures are two-phase model results

1) Free surface vortex check

Fig 51 Streamline pattern of free surface

In Fig 51 streamline pattern in single phase model shows there are symmetrical

free surface vortex structures around pump intake besides the main vortex vortices

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 60: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

48

formation near the side wall are observed While in the two-phase model only one

main vortex structure is observed near the pump intake

2) Submerged vortex check

Fig 52 shows the calculation results of vortex core region for both single phase

model and two-phase model The vertex core region in this study is revealed by iso-

surface of swirling strength which represents the strength of the local swirling

motion Vortex structures occurring from bottom and side wall can be observed

Fig 52 Vortex core region of the sump model

(a) Side wall section

(b) Bottom section

Fig 53 Velocity vectors on different sections

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 61: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

49

Velocity vectors are given in Fig 53 in both cases vortex formation and location

on the side wall section and bottom section are predicted

512 PIV results analysis

During the PIV experiment the motion of particles on different light sheets was

captured by a high speed camera Before importing to the post-processing program

CACTUS 33 the 24 bit-real-color raw images were converted to 8 bit-gray-scale

pictures

1) Vortex check

Fig 54 shows averaged velocity vectors on each cross-sectional plane Free surface

vortex and submerged vortex on back wall bottom and side wall are observed

Actually with the velocity vector video created by CACTUS the location and

formation disappearance of vortices are predicted quite good and especially for the

bottom and back wall section the transient phenomenon is significant

2) Flow conditions check

The flow conditions around the intake and in the pipe are also investigated as shown

in Fig 55 As to the velocity vectors outside of intake center there is no obvious

vortex structure both from the side view and back view The velocity in pipe is

relatively high To capture the motion of particles in pipe 1000 fps of camera is

selected for the side view section while 3000 fps for the bottom view section The

tangential velocity of pipe flow is predicted in the form of rotational motion in the

bottom view section

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 62: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

50

(a) velocity vector near the free surface

(b) velocity vector near the side wall

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 63: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

51

(c) velocity vector near the bottom

(d) velocity vector near the back wall

Fig 54 Velocity vectors of PIV results

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 64: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

52

Fig 55 Velocity vectors on different sections (PIV)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 65: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

53

513 Sump with anti free surface vortex devices

1) Free surface vortex check

Fig 56 shows the free surface vortex check results of sump model with various

curtain wall or square bar (specified in Chapter 4) The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed as Fig 56(a) (b) (c) The

square bar installed in the channel failed to suppress the free surface vortex

(a) Curtain wall 1 (b) Curtain wall 2 (c) Curtain wall 3

(d) Square bar 1 (e) Square bar 1

Fig 56 Velocity vectors on free surface of different cases

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 66: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

54

2) Swirl angle analysis

Swirl angle is used to predict the intensity of flow rotation As specified in Chapter

2 for swirl angle CFD calculation the swirl check circle is located at the section of

4d height with diameter of 025d 05d and 075d respectively Table 51 shows the

swirl angle results both in CFD calculation and experiment For experiment the

rotation of swirl meter is recorded over 10 minutes Table 52 is the summary of

swirl angle CFD calculation results and Fig 57 shows tangential velocity profile of

different models As referred to the maximum value of circumferential velocity the

sump with curtain wall 2 and sump with square bar 2 show higher fluctuation

which is good agreement with the swirl angle calculation results

Table 51 Swirl angle of model test results

CFD results Experimental results

Average tangential

velocity(ms)

average axial velocity

(ms)

Swirl angle(deg)

Swirl meter Rpm(CCC)

average axial velocity

(ms)

Swirl angle(deg)

00366 21 1 168 21 42

Rotational direction C(clockwise) CC( counter clockwise)

Table 52 Summary of swirl angle CFD calculation results

CFD Average tangential

velocity (ms) Average axial velocity (ms)

Swirl angle(deg)

Sump model 0036641331 21 1

Curtain wall 1 0070152 21 19

Curtain wall 2 0179534 21 49

Curtain wall 3 0007058 21 019

Bar wall 1 0042852 21 117

Bar wall 2 0110428001 21 3

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 67: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

55

(a) Original sump model (b) Sump model with curtain wall 1

(c) Sump model with curtain wall 2 (d) Sump model with curtain wall 3

(e) Sump model with square bar 1 (f) Sump model with square bar 2

Fig 57 Tangential velocity profile of different models

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 68: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

56

52 Results of the mixed flow pump sump model

521 Performance analysis of the mixed flow pump sump model

For the performance presentation the pump shaft power head and efficiency were

defined as

η =

(51)

P = Q times ρ times Δh (52)

P = Torque times rps (53)

Δh =

ρtimes (54)

where Δh is the change in total pressure between the inlet (h ) and out (h ) Q is

the given fluid flow rate g is the acceleration due to gravity and ρ is the density of

the fluid

The hydraulic performance analysis is performed by changing the flow rate range

from 50 to 140 of the design value while the rotor speed is kept constant A 10

growth of the design flow rate is conducted between the adjacent varying flow rate

Table 53 Mixed flow pump performance results

Flow ate ( m3h )

Shaft power ( kW )

Head ( m )

Efficiency

11000 132257 2869 6483 13500 144004 2852 7264 15000 150171 2803 7608 17000 151227 2757 8420 19000 152489 2566 8685 21700 152772 2323 8964 24000 149153 2032 8883 26500 141402 1635 8324 28500 135409 1315 7519 30500 122874 974 6570

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 69: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

57

Fig 58 Performance curve of the mixed flow pump with the sump

Fig 58 shows the hydraulic performance of flow rate versus head shaft power and

pump efficiency of the pump sump model according to the data of Table 53 The

performance curves show that as the flow rate increases the head of pump decreases

gradually the shaft power and efficiency increase first and then decrease The shaft

power increases up to 15278kW at the flow rate of design The best efficiency

point (BEP) is at the design flow rate with the corresponding efficiency of 896

Volume flow rate(m3h)

5000 10000 15000 20000 25000 30000 35000

Hea

d(m

)

0

10

20

30

40

Sh

aft

Po

we

r(kW

)

1200

1400

1600

1800

2000

Eff

icie

ncy

()

0

20

40

60

80

100

HeadShaft Power

Efficiency

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 70: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

58

522 Flow characteristics analysis

The flow condition around the intake structure is quite complex In this section the

post results of streamlines vortex core and vorticity distributions are analyzed to

predict the flow characteristics Comparisons of these data in sump model without

and with AVD are shown in the following section

1) Flow pattern around the pump intake

(a) Streamline near the side wall (right AVD installed)

(b) Streamline near the rear wall (right AVD installed )

Fig 59 Streamline pattern around the intake structure

Fig 59 shows the comparison of streamline pattern around the pump intake

structure with and without the AVD at the design flow rate Near the side wall and

the rear wall in the vicinity of the intake vortex formation is observed and with the

AVD installed the vortex is suppressed

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 71: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

59

2) Vortex core region in the model

Fig 510 is the calculation results of vortex core region by the pump sump model

both without and with the AVD when the flow rate is set to 21700m3h

(a) model without AVD (b) model with AVD

Fig 510 Vortex core region of the pump sump model

3) Vorticity distributions in the sump model

To check the vortex intensity around the intake structure quantitatively Fig 511

and Fig 512 shows the vorticity distribution comparison of the sump without and

with the AVD installed both in the flow direction and the channel width direction

The horizontal axis coordinate x=00 means the center of the intake bell mouth and

H is the distance of the cross sectional plan from the bottom wall where H=07m is

the same height of the bell mouth inlet

Fig 511 and Fig 512 show at the radius region of the bell mouth inlet wall

maximum value of vorticity is obviously higher than those at other heights ie

H=07m x=plusmn1m which means flow with high vorticity near the bell mouth tip Fig

511 shows the vorticity distribution in the flow direction at each height Below the

height of 07m vorticity almost disappears except near the center of the bell mouth

Fig 512 also shows that below the height of 07m there are only symmetrical

vorticity values near the center of the bell mouth In the comparison of the vorticity

distribution with and without the AVD though the vorticity of the region below the

bell mouth is small relatively lower values of the maximum vorticity are obtained

by installation of the AVD This implies that the possible vortex occurrence around

the pump intake is reduced

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 72: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

60

Fig 511 Vorticity distribution in the flow direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 73: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

61

Fig 512 Vorticity distribution in channel width direction

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 74: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

62

523 Cavitation phenomenon analysis

As an undesirable phenomena caused by the adverse flow condition cavitation is

the process of the formation of vapor bubbles within a liquid where flow dynamics

cause the low pressure With the rapid formation growth and collapse of the

bubbles cavitation manifests in the form of pump performance decrease vibration

additional noise increase and even the equipment damage Fig 513 predicts the

inception of cavitation in the mixed flow pump

Fig 513 Cavitation region of the mixed flow pump

Pump cavitation occurs when the local absolute pressure of the flow falls below the

vapor pressure of the liquid The net positive suction head (NPSH) is used to predict

the cavitation performance defined as

NPSH=(P0minus Pν)ρg (55)

where P0 is the total pressure of the impeller inlet suction Pν is the vapor pressure

g is the acceleration due to gravity and ρ is the density of the fluid

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 75: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

63

Fig 514 Cavitation performance curves of the mixed flow pump

Every pump has a critical cavitation spot the required net positive suction head

(NPSHR) which is defined as the minimum necessary NPSH to avoid cavitation

usually a corresponding 3 or 5 head drop Fig 514 shows the cavitation

performance characteristics at different flow rate under the same rotational speed

where Q0 is the design flow rate When the inlet total pressure high enough there is

no cavitation and the head remains constant As the inlet pressure decreases the

NPSH approaches the NPSHR due to which cavitation occurs and expands

Comparing the three curves the NPSHR for the flow rate 078Q0 100Q0 and

141Q0 is 114m 72m and 110m as the curves show respectively which means the

cavitation performance of the Q0 is the best The curves show that under the design

working condition ie flow rate of Q0 the NPSH is 134m there is a considerable

extent over the NPSHR=72m which means the pump at the BEP condition has met

the requirement of the cavitation performance

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 76: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

64

Fig 515 Vapor volume fraction distribution on blade surface

To clarify cavitation phenomenon in flow passages of the mixed flow pump the

vapor volume fraction distribution at design flow rate condition versus different

NPSH ( marked as a b c in the curve ) is plotted in Fig 515

Cavitation appears on the blade suction surface where the leading edge meets the tip

as NPSH decreases cavitation zone expands Although NPSH=134m (point c) is in

the NPSH safety extent there are slight cavitation bubbles on the suction surface

whose influence to the pump performance is negligible However further reduction

below the NPSHR=72m (point b) like when NPSH=61m (point a) under the

severe cavitation condition the cavitation region takes about 50 area of the

suction surface ie serious flow passage blockage which can result in a major

deterioration in the pump performance Cavitation on the pressure surface shows the

similar situation however the inception of cavitation is observed quite late due to

the higher pressure distribution

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 77: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

65

Chapter 6 Conclusions

In this study two different sump models were studied a scaled sump model was

adopted to carry out numerical simulation and PIV experiment a mixed flow pump

sump model was applied to conduct the analysis of pump hydraulic performance

and flow conditions of the sump model The results show

a) For the scaled sump model although there is a little difference between the

single phase and two-phase simulation results the free surface vortex and

submerged vortex formation and location are predicted PIV experiment

was conducted to investigate all the flow conditions around the sump intake

structure and in the pipe The CFD simulations of flow condition show

good agreement with the PIV results The effectiveness of curtain wall

installed to inhibit the free surface vortex is confirmed in the CFD results

while the square bar installed in the channel failed to suppress the free

surface vortex

b) Based on the numerical calculation the efficiency of the sump pump the

effectiveness of an AVD and the cavitation phenomenon were predicted as

follows

In the hydraulic performance analysis the mixed flow pump with the sump

shows high efficiency of 896 at the BEP condition And it is confirmed

that the pump operating under the BEP condition has met the requirement

of cavitation performance

According to the flow characteristics analysis with the AVD installed the

suppression to vortex structure and the reduction of the maximum vorticity

values at each height are verified which confirm the effectiveness of AVD

to inhibit the submerged vortex Cavitation first occurs on the blade suction

surface near the leading edge and the shroud As NPSH decreases the

caviting region is spread out over the blade and the serious blockage in flow

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 78: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

66

passage caused by cavitation results in a sharp reduction of the pump head

which directly reduces the pump performance

c) As the CFD calculation of swirl angle was adopted in this research work

more investigations with experiments should be conducted With the

comparison analysis the effectiveness of AVDs for both free surface vortex

and submerged vortex is confirmed For improve the AVD suggestion

more investigations for various parameters should be studied by experiment

(scaled model test PIV ) and CFD in the future

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 79: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

67

Acknowledgement

I would like to express my gratitude to all those who helped me during the writing

of this thesis

My deepest gratitude goes first and foremost to my supervisor Prof Dr Young-Ho

Lee for his constant encouragement and guidance Without his patience and tutorial

it is no way that I can carry out this research work

Secondly I extremely appreciate Prof Dr Jae-Hyun JEONG (Chairperson review

panel) and Prof Dr Hyung-Ho JEONG (Co-chairperson review panel) for their

valuable suggestions and helpful comments

Thirdly I am also indebted to my lab mates both Korean and foreign for their

knowledge and assistance in many areas

Finally sincere thanks would go to my beloved family for their loving

considerations and great confidence in me all through these years

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 80: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

68

References

[1] Shyam N S Kshirsagar J T 2008 Numerical prediction of air entrainment in

pump intakes Proc of the 24th int pump users Symp pp 29-33

[2] Matsui J Kamemoto K and Okamura T 2006 CFD Benchmrak and a Model

Experiment on the Flow in Pump Sump Proc of 23th IAHR Symp Oct

Yokohama 110-16

[3] Okamura T Kyoji K and Matsui J (2007) CFD Prediction and Model

Experiment on Suction Vortices in Pump Sump Proceedings of the 9th Asian

International Conference on Fluid Machinery Jeju Korea

[4] Tanweer S Desmukh amp VK Gahlot 2011 Numerical study of flow behavior in

a multiple intake pump sump International Journal of Advanced Engineering

Technology Vol2 118-128

[5] Rajendran V P Constantinescu S G and Patel V C 1998 Experiments on Flow

in Model Water-pump Intake Sump to Validate a Numerical Model Procof ASME

Fluids Engineering Division Summer Meeting FEDSM98-5098

[6] Iwano R Shibata T Nagahara T and Okamura T 2002 Numerical Prediction

Method of a Submerged Vortex and Its Application to the Flow in Pump Sumps

with and without a Baffle Plate Proc of the 9th International Symposium on

Transport Phenomena and Dynamics of Rotating Machinery 1-6

[7] Choi J W Choi Y D Kim C G and Lee Y H 2010 Flow uniformity in a multi-

intake pump sump model Journal of Mechanical Science and Technology 24(7)

1389-1400

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 81: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

69

[8] Oh H W Jan 2010 Application of Computational Fluid Dynamics to Practical

Design and Performance Analysis of Turbomachinery Book Computational Fluid

Dynamics ISBN 978-953-7619-59-6 pp 420 INTECH Croatia

[9] Kim J H ANH H J Kim K Y 2010 High-efficiency design of a mixed flow

pump Science China Vol53 24~27

[10] van Os M J Op de woerd J G H Jonker J B 1997 A parametric study of the

cavitation inception behavior of a mixed-flow pump impeller using a three-

dimensional potential flow model Proc of ASME Fluids Engineering Division

Summer Meeting FEDSM97-3374

[11] Dupont P Okamura T 2003 Cavitating Flow Calculations in Industry

International Journal of Rotating Machinery 9(3) 163ndash170

[12] Li J Liu L J Li G J and Feng Z P 2007 Numerical prediction of cavitation

flows in a centrifugal pump impeller Journal of Engineering Thermophysics 28(6)

948-950 in Chinese

[13] An Y J Shin B R 2011 Numerical investigation of suction vortices behavior

in centrifugal pump Journal of Mechanical Science and Technology 25 (3)

767~772

[14] Kim C G Choi Y D and Lee Y H 2012 A study on the effectiveness of an anti

vortex device in the sump model by experiment and CFDEarth and environment

science

[15] Tanweer S Desmukh amp VK Gahlot July 2010 Simulation of flow through a

pump sump and its valiation IJRRAS 4 (1)

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 82: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

70

[16] Mohammed K A 2010 Cavitation in Centrifugal Pumps Diyala Journal of

Engineering Sciences pp170~180

[17] Chang S P Wang Y S 2012 Cavitation performance research of mixed-flow

pump based on CFD Journal of Drainage and Irrigation Machinery Engineering

30(2)171-180 in Chinese

[18] Hydraulic Institute 2012 American National Standard for Rotodynamic

Pumps for Pump Intake Design (ANSIHI 98-2012)

[19] Padmanabhan M and GE Hecker 1984 ldquoScale effects in pump sump

modelsrdquo J Hydr Div ASCE 110(11) Pp 1540-1556

[20] Shin C S E M Greitzer WK Cheng CS Tan and CL Shippee 1986

ldquoCirculation

measurements and vortical structure in an inlet-vortex fieldrdquo Journal of Fluid

Mechanics Vol 162 Pp 463-487

[21] Iwano R 1993 Onset condition of vortex-induced gas entrainment at free

surface Proc of Japan Society of Mechanical Engineers No930-9 pp594-596

[22] Shibata T Iwano R Nagahara T and Okamura T 2000 A numerical method

for predicting the cavitation inception of a submerged vortex in a pump sump Proc

of 20th IAHR symp

[23] Menter F R 1994 Two equation eddy viscosity Turbulence for Engineering

Applications AIAA-Journal Vol 32(8) pp1598-1605

[24] Ansys inc 2009 Two Equation Turbulence Models ANSYS CFX- Solver

Theory Guide

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt

Page 83: Master's Thesis Study of Flow Field in Rectangular Sump Models …repository.kmou.ac.kr/bitstream/2014.oak/8473/1/... · 2019-11-21 · Chapter 1 Introduction ... stations depend

INSIDabcdef_MS_0001MS_0001IN

SID

abcdef_

MS

_0001M

S_0001

71

[25] Hirt C W Nichols B D 1981 Volume of Fluid (VOF) Method for the

dynamics of free boundariesJournal of computational physics 39 201~225

  • List of Tablesiii
  • List of Figuresiv
  • Abstractvi
  • Nomenclatureviii
  • Abbreviationsix
  • Chapter 1 Introduction
    • 11 Background
    • 12 Previous study
    • 13 Methodology of study
      • Chapter 2 Basic theory of sump model
        • 21 Pump intake design
          • 211 Inlet bell diameter design
          • 212 Recommended dimensions for a rectangular sump
            • 22 Model test
              • 221 The scale effects
              • 222 Principle of similarity
              • 223 Preliminary model test
              • 224 Test criteria
              • 225 Remedial measures for problem intakes
                  • Chapter 3 Methodology
                    • 31 Numerical modeling
                    • 32 Turbulence models
                      • 321 k-ε turbulence model
                      • 322 Wilcox k-ω turbulence model
                      • 323 Shear stress transport model
                        • 33 Brief introduction of simulation process
                          • 331 Pre-processing technique
                          • 332 Solving the simulation
                          • 333 Post processing
                              • Chapter 4 Description of Model Cases
                                • 41 Creating the geometry
                                  • 411 Geometry of the scaled sump model
                                  • 412 Geometry of the full size pump sump model
                                    • 42 Mesh generation
                                    • 43 Numerical approach
                                    • 44 Experimental setup
                                      • Chapter 5 Results and Discussion
                                        • 51 Results of the scaled sump model
                                          • 511 Numerical simulation vortex check
                                          • 512 PIV results analysis
                                          • 513 Sump with anti free surface vortex devices
                                            • 52 Results of the mixed flow pump sump model
                                              • 521 Performance analysis of the mixed flow pump sump model
                                              • 522 Flow characteristics analysis
                                              • 523 Cavitation phenomenon analysis
                                                  • Chapter 6 Conclusions
                                                  • Acknowledgement
                                                  • References
                                                    • ltstartpagegt1List of TablesiiiList of FiguresivAbstractviNomenclatureviiiAbbreviationsixChapter 1 Introduction 1 11 Background 1 12 Previous study 1 13 Methodology of study 3Chapter 2 Basic theory of sump model 4 21 Pump intake design 5 211 Inlet bell diameter design 5 212 Recommended dimensions for a rectangular sump 7 22 Model test 10 221 The scale effects 11 222 Principle of similarity 12 223 Preliminary model test 13 224 Test criteria 19 225 Remedial measures for problem intakes 20Chapter 3 Methodology 25 31 Numerical modeling 25 32 Turbulence models 26 321 k-ех turbulence model 27 322 Wilcox k-еdeg turbulence model 28 323 Shear stress transport model 29 33 Brief introduction of simulation process 31 331 Pre-processing technique 31 332 Solving the simulation 32 333 Post processing 32Chapter 4 Description of Model Cases 33 41 Creating the geometry 33 411 Geometry of the scaled sump model 33 412 Geometry of the full size pump sump model 35 42 Mesh generation 39 43 Numerical approach 42 44 Experimental setup 45Chapter 5 Results and Discussion 47 51 Results of the scaled sump model 47 511 Numerical simulation vortex check 47 512 PIV results analysis 49 513 Sump with anti free surface vortex devices 53 52 Results of the mixed flow pump sump model 56 521 Performance analysis of the mixed flow pump sump model 56 522 Flow characteristics analysis 58 523 Cavitation phenomenon analysis 62Chapter 6 Conclusions 60Acknowledgement 67References 68ltbodygt