liquid sloshing in gravity driven water pool of …
TRANSCRIPT
BARC/2015/E/001B
AR
C/2015/E
/001
LIQUID SLOSHING IN GRAVITY DRIVEN WATER POOL OFADVANCED HEAVY WATER REACTOR:
POOL LIQUID UNDER DESIGN SEISMIC LOAD AND SLOSH CONTROL STUDIES
byM. Eswaran and G.R. Reddy
Structural and Seismic Engineering Section,Reactor Safety Division
BARC/2015/E/001
GOVERNMENT OF INDIAATOMIC ENERGY COMMISSION
BHABHA ATOMIC RESEARCH CENTREMUMBAI, INDIA
2015
BA
RC
/201
5/E
/001
LIQUID SLOSHING IN GRAVITY DRIVEN WATER POOL OFADVANCED HEAVY WATER REACTOR:
POOL LIQUID UNDER DESIGN SEISMIC LOAD AND SLOSH CONTROL STUDIES
byM. Eswaran and G.R. Reddy
Structural and Seismic Engineering Section,Reactor Safety Division
BIBLIOGRAPHIC DESCRIPTION SHEET FOR TECHNICAL REPORT(as per IS : 9400 - 1980)
01 Security classification : Unclassified
02 Distribution : External
03 Report status : New
04 Series : BARC External
05 Report type : Technical Report
06 Report No. : BARC/2015/E/001
07 Part No. or Volume No. :
08 Contract No. :
10 Title and subtitle : Liquid sloshing in gravity driven water pool of Advanced Heavy Water Reactor - Pool liquid under design seismic load and slosh control studies
11 Collation : 84 p., 67 figs., 13 tabs.
13 Project No. :
20 Personal author(s) : M. Eswaran; G.R. Reddy
21 Affiliation of author(s) : Structural and Seismic Engineering Section, Reactor Safety Division, Bhabha Atomic Research Centre, Mumbai
22 Corporate author(s) : Bhabha Atomic Research Centre, Mumbai - 400 085
23 Originating unit : Reactor Safety Division, Bhabha Atomic Research Centre, Mumbai
24 Sponsor(s) Name : Department of Atomic Energy
Type : Government
Contd...
BARC/2015/E/001
BARC/2015/E/001
30 Date of submission : January 2015
31 Publication/Issue date : February 2015
40 Publisher/Distributor : Head, Scientific Information Resource Division, Bhabha Atomic Research Centre, Mumbai
42 Form of distribution : Hard copy
50 Language of text : English
51 Language of summary : English
52 No. of references : 39 refs.
53 Gives data on :
60
70 Keywords/Descriptors : HWLWR TYPE REACTORS; SEISMIC EFFECTS; EARTHQUAKES; FLUID MECHANICS; FOURIER TRANSFORMATION
71 INIS Subject Category : S21
99 Supplementary elements :
Abstract : Sloshing phenomenon is well understood in regular cylindrical and rectangular liquidtanks subjected to earthquake. However, seismic behaviour of water in complex geometry suchas a sectored annular tank, e.g., Gravity Driven Water Pool (GDWP) which is located in AdvancedHeavy Water Reactor (AHWR) need to be investigated in detail in the view of safety significance.Initially, for validation of Computational Fluid Dynamics (CFD) procedure, square and foursectored square tanks are taken. Slosh height and liquid pressure are calculated over time throughtheoretical and experimental procedures. Results from theoretical and experimental approachesare compared with CFD results and found to be in agreement. The present work has two mainobjectives. The first one is to investigate the sloshing behaviour in an un-baffled and baffledthree dimensional single sector of GDWP of AHWR under sinusoidal excitation. Other one is tostudy the sloshing in GDWP water using simulated seismic load along the three orthogonaldirections. This simulated seismic load is generated from design basis floor response spectrumdata (FRS) of AHWR building. For this, the annular tank is modelled along with water andnumerical simulation is carried out. The sinusoidal and earthquake excitations are applied asacceleration force along with gravity. For the earthquake case, acceleration-time history isgenerated compatible to the design FRS of AHWR building. The free surface is captured byVolume of Fluid (VOF) technique and the fluid domain is solved by finite volume method whilethe structural domain is solved by finite element approach. Un-baffled and baffled tankconfigurations are compared to show the reduction in wave height under excitation. Theinteraction between the fluid and pool wall deformation is simulated using a partitioned fluid–structure coupling. In the earthquake case, a user subroutine function is developed to convertFRS in to time history of acceleration in three directions. Wavelet analysis is performed to findthe frequency variation of sloshing with respect to time. Results such as sloshing heights andhydrodynamic pressure considering with and without structure interaction effects have beenpresented
ii
ABSTRACT
Sloshing phenomenon is well understood in regular cylindrical and rectangular liquid
tanks subjected to earthquake. However, seismic behaviour of water in complex geometry
such as a sectored annular tank, e.g., Gravity Driven Water Pool (GDWP) which is located
in Advanced Heavy Water Reactor (AHWR) need to be investigated in detail in the view
of safety significance. Initially, for validation of Computational Fluid Dynamics (CFD)
procedure, square and four sectored square tanks are taken. Slosh height and liquid
pressure are calculated over time through theoretical and experimental procedures. Results
from theoretical and experimental approaches are compared with CFD results and found
to be in agreement.
The present work has two main objectives. The first one is to investigate the sloshing
behaviour in an un-baffled and baffled three dimensional single sector of GDWP of
AHWR under sinusoidal excitation. Other one is to study the sloshing in GDWP water
using simulated seismic load along the three orthogonal directions. This simulated seismic
load is generated from design basis floor response spectrum data (FRS) of AHWR
building. For this, the annular tank is modelled along with water and numerical simulation
is carried out. The sinusoidal and earthquake excitations are applied as acceleration force
along with gravity. For the earthquake case, acceleration-time history is generated
compatible to the design FRS of AHWR building. The free surface is captured by Volume
of Fluid (VOF) technique and the fluid domain is solved by finite volume method while
the structural domain is solved by finite element approach. Un-baffled and baffled tank
configurations are compared to show the reduction in wave height under excitation. The
interaction between the fluid and pool wall deformation is simulated using a partitioned
fluid–structure coupling. In the earthquake case, a user subroutine function is developed to
convert FRS in to time history of acceleration in three directions. Wavelet analysis is
performed to find the frequency variation of sloshing with respect to time. Results such as
sloshing heights and hydrodynamic pressure considering with and without structure
interaction effects have been presented.
iii
CONTENTS
Contents
ABSTRACT ........................................................................................................................................ ii
CONTENTS ....................................................................................................................................... iii
LIST OF FIGURE .............................................................................................................................. v
1. INTRODUCTION ...................................................................................................................... 1
1.1 Background ................................................................................................................................ 1
1.2 Problem Definition ...................................................................................................................... 2
1.3 Objectives .................................................................................................................................. 3
1.4 Report Organisation .................................................................................................................... 4
2 MATHEMATICAL FORMULATIONS FOR FLUID, STRUCTURE AND INTERACTION ..... 5
2.1 Fluid Formulation ....................................................................................................................... 5
2.2 Structural Formulation ................................................................................................................. 6
2.3 Coupling between Fluid Flows and Structural Media ..................................................................... 7
3 NUMERICAL METHODOLOGY ............................................................................................. 9
3.1 Introduction ................................................................................................................................ 9
3.2 Grid Generation and Boundary Conditions .................................................................................... 9
3.3 Volume of Fluid (VOF) Method ................................................................................................. 11
4 SLOSHING ANALYSIS USING SEISMIC DESIGN CODES .................................................. 13
4.1 Introduction .............................................................................................................................. 13
4.2 Equivalent Rectangular Method (ERM): ..................................................................................... 14
4.3 Spring-Mass Damper Model ...................................................................................................... 15
4.4 Result and Discussions .............................................................................................................. 18
5 ESTIMATION OF SLOSH FREQUENCIES USING CFD ....................................................... 24
5.1 Introduction .............................................................................................................................. 24
5.2 Modal Analysis of the GDWP .................................................................................................... 25
5.3 Effect of Excitation Frequency ................................................................................................... 26
6 EXPERIMENTAL AND ANALYTICAL VALIDATION OF CFD SIMULATIONS .................. 29
iv
6.1 Case Study 1: Analytical Validation ............................................................................................ 29
6.2 Case Study 2: Experimental Validation ....................................................................................... 30
7 CFD SIMULATION OF LIQUID SLOSHING CONTROL IN GDWP UNDER SINUSOIDAL
EXCITATION .................................................................................................................................. 34
7.1 Effect of Amplitude of GDWP ................................................................................................... 34
7.2 Structural Analysis .................................................................................................................... 37
7.3 Effect of Baffles in GDWP ........................................................................................................ 37
7.4 Liquid Elevation in Higher Modes of GDWP .............................................................................. 41
8 CFD SIMULATION OF SLOSHING IN GDWP UNDER SEISMIC EXCITATION ................. 44
8.1 Introduction .............................................................................................................................. 44
8.2 Development of Random Waves ................................................................................................. 45
8.3 Wavelet Analysis ....................................................................................................................... 45
8.4 GDWP under Sinusoidal Excitation in Multi-Direction (Case 1): .................................................. 48
8.5 GDWP under Random Excitation in Multi-Direction (Case 2) ...................................................... 49
9 CONCLUSIONS ...................................................................................................................... 57
9.1 General Conclusions from above Study ....................................................................................... 57
9.2 Some more Observations between ERM and CFD ....................................................................... 59
APPENDIX- A.................................................................................................................................. 69
NON-LINEARITY EFFECT DUE TO WAVE BREAKING .......................................................... 69
A-1. Tank with Wave Breaking Device (Passive baffle) ..................................................................... 71
A-2 Discussion and Conclusions ...................................................................................................... 72
v
LIST OF FIGURE
Fig. 1-1 Advanced Heavy Water Reactor Building (Sinha and Kakodkar, 2006) ............................. 2
Fig. 1-2 GDWP located top of the primary containment .................................................................. 3
Fig. 2-1 Feedback loop of fluid structure interaction (Eswaran et al., 2009) .................................... 7
Fig. 3-1 One compartment of GDWP. ............................................................................................. 9
Fig. 3-2 Numerical grid of un-baffled GDWP. ................................................................................ 10
Fig. 3-3 Block structured mesh on fluid and structure portions. ..................................................... 10
Fig. 4-1 Fully curved outer wall GDWP. ......................................................................................... 13
Fig. 4-2 Partially curved outer wall GDWP. ................................................................................... 13
Fig. 4-3 Sectional view of GDWP (updated dimensions) ............................................................... 14
Fig. 4-4 Plan of GDWP .................................................................................................................. 14
Fig. 4-5 Equivalent rectangular model ............................................................................................ 15
Fig. 4-6 Spring –mass model ........................................................................................................... 15
Fig. 4-7 Impulsive pressure on wall ................................................................................................ 23
Fig. 4-8 Convective pressure on wall .............................................................................................. 23
Fig. 5-1 Free surface elevation of liquid under free vibration. ........................................................ 25
Fig. 5-2 Spectrum analysis of signal. .............................................................................................. 25
Fig. 5-3 Dominant mode shapes of structure. ................................................................................ 26
Fig. 5-4 Time history of free surface elevation at 0.1 m excitation amplitude and for different
excitation frequencies. ..................................................................................................................... 28
Fig. 6-1 The sketch of the 2-D rigid rectangular tank. .................................................................... 30
Fig. 6-2 Comparisons of free surface elevation. ............................................................................. 30
Fig. 6-3 Details of the experimental setup. ..................................................................................... 30
Fig. 6-4 Experimental setup. ........................................................................................................... 31
Fig. 6-5 Isometric view of 4-sectored square tank .......................................................................... 31
Fig. 6-6 Comparison of experimental and CFD pressure data at 200 mm from tank bottom
(position 2). ..................................................................................................................................... 31
Fig. 6-7 Actuator with shake table................................................................................................... 32
Fig. 6-8 Water oscillation during the excitation .............................................................................. 32
Fig. 6-9 Water spill out snapshots during base excitation in an 4-sectored rectangular tank .......... 32
vi
Fig. 7-1 Power spectral density for amplitude 0.1 m and 1ω =0.312 Hz. ...................................... 35
Fig. 7-2 Time history of non-dimensional free surface elevation at 1ω =0.312 Hz for different
excitation amplitudes ....................................................................................................................... 35
Fig. 7-3 Phase-plane diagram for amplitude 0.1 m and 1ω =0.312 Hz. ......................................... 35
Fig. 7-4 Pressure point locations. .................................................................................................... 36
Fig. 7-5 Time history of pressure at various locations for 0.1 m amplitude sec and 1ω =0.312 Hz.
......................................................................................................................................................... 36
Fig. 7-6 Wall horizontal displacement contour at time 1.417 sec ................................................... 36
Fig. 7-7 Outer and inner walls x-displacement versus time ............................................................ 36
Fig. 7-8 VanMises stress contour at time 12.4 sec. ......................................................................... 37
Fig. 7-9 Sectional view of velocity magnitude at time 12.4 sec. ..................................................... 37
Fig. 7-10 Baffles shape and its position in the water pool. ............................................................. 39
Fig. 7-11 Comparison of liquid elevations with no baffle, annular baffle and cap-plate baffle cases
at 1ω =0.312 Hz and amplitude 0.1 m. ............................................................................................ 40
Fig. 7-12 Effect of baffles at right corner of the water pool case at 15.05 sec and 1ω =0.312 Hz. . 40
Fig. 7-13 Comparisons of free surface profile at different time instant for un-baffled and baffled
water pool for 0.1 m amplitude sec and 1ω =0.312 Hz. (Figs. (a)-(d), (e)-(h), (i)-(l) show un-
baffled, annular baffled, cap-plate baffled water pools respectively).............................................. 42
Fig. 7-14 Comparison of free surface elevation of liquid at 0.1 m amplitude and 5 1ω excitation
frequency. ........................................................................................................................................ 43
Fig. 7-15 Comparison of free surface elevation of liquid at 0.1 m amplitude and 10 1ω excitation
frequency. ........................................................................................................................................ 43
Fig. 8-1 Three sectors in GDWP ..................................................................................................... 46
Fig. 8-2 FRS data for AHWR building at 137 m ............................................................................. 47
Fig. 8-3 Acceleration- time history for GDWP of AHWR building at 137 m ................................. 48
Fig. 8-4 Multi directional sinusoidal excitation (Case 1) (a) –(c) Non-dimensional slosh height for
sector 1 to 3 (d) Phase-plane diagram (e)Wavelet diagram for slosh height signal at right corner of
sector 1 ............................................................................................................................................ 49
Fig. 8-5 Non-dimensional slosh height for sector 1 to 3 of the GDWP subjected to design
excitation as given in Fig. 8.3 (Case 2) ........................................................................................... 50
Fig. 8-6 FFT from slosh height at sector 1 of GDWP subjected to design excitation as given in Fig.
vii
8.3 (Case 2) ..................................................................................................................................... 51
Fig. 8-7 FFT from slosh height for sector 2 of GDWP subjected to design excitation as given in
Fig. 8.3 (Case 2) .............................................................................................................................. 51
Fig. 8-8 FFT from slosh height for sector 3 of GDWP subjected to design excitation as given in
Fig. 8.3 (Case 2) .............................................................................................................................. 51
Fig. 8-9 Phase-Plane diagram of sector 1 slosh height .................................................................... 51
Fig. 8-10 Time –frequency curve for sector 1 of GDWP slosh height subjected to design excitation
as given in Fig. 8.3 (Case 2) ............................................................................................................ 52
Fig. 8-11 Time –frequency curve for sector 2 of GDWP slosh height subjected to design excitation
as given in Fig. 8.3 (Case 2) ............................................................................................................ 52
Fig. 8-12 Time –frequency curve for sector 3 of GDWP slosh height subjected to design excitation
as given in Fig. 8.3 (Case 2) ............................................................................................................ 52
Fig. 8-13 Base shear at convective mode for sector 1 through 3 .................................................... 53
Fig. 8-14 Total force (convective + impulsive) at sector wall 1 ..................................................... 54
Fig. 8-15 Total moment (convective + impulsive) at sector wall 1 ................................................. 55
Fig. 8-16 Tank wall displacement versus time ................................................................................ 55
Fig. 8-17 Wall pressure (convective + impulsive) at sector 1 through 3 ......................................... 56
Fig. 9-1 Comparison between CFD and ERM for fully curved outer wall GDWP. ........................ 60
Fig. 9-2 Comparison between CFD and ERM for partially curved outer wall GDWP. .................. 60
Fig. A-1 Slosh height elevation computed from linear and nonlinear equations ............................ 69
Fig. A-2 Snap shot of maximum slosh height at left end ................................................................ 69
Fig. A -3 Snap shot of maximum slosh height at right end ............................................................. 69
Fig. A -4 Tank with wave breaking device (Case 3) ........................................................................ 70
Fig. A -5 Fine mesh arrangement (Case 3) ...................................................................................... 70
Fig. A -6 Snap shot of Non-linear violent flow at 0.99 with baffle arrangement (case 4) ........ 71
viii
LIST OF TABLE
Table 3-1: Material properties ......................................................................................................... 10
Table 4-1: Slosh height estimation through seismic codes for sector 1 of fully curved outer wall
GDWP ............................................................................................................................................. 17
Table 4-2: Slosh height estimation through seismic codes for sector 1 of partially curved outer
wall GDWP ..................................................................................................................................... 18
Table 4-3: X and Y direction values of sector 1 of fully curved outer wall GDWP. ...................... 19
Table 4-4: X and Y direction values of sector 1 of partially curved outer wall GDWP. ................. 21
Table 5-1: Fully curved outer wall GDWP natural frequency in Hz for different sectors ............... 25
Table 5-2: Partially curved outer wall GDWP natural frequency in Hz for different sectors .......... 26
Table 5-3: Numerical cases taken for investigation. ....................................................................... 27
Table 8-1: Sectors frequencies in hertz computed by CFD simulations .......................................... 47
Table 8-2: Numerical case studies ................................................................................................... 47
Table 8-3: Total force and moment in sectors 1, 2 and 3 ................................................................. 54
Table 9-1: X and Y direction values of sector 1 of fully curved wall GDWP. ................................ 61
Table 9-2: X and Y direction values of sector 1 of partially curved wall GDWP. ........................... 62
v
LIST OF ABBREVIATION
ACI American Concrete Institute
AHWR - Advanced Heavy Water Reactor
BIS - Bureau of Indian Standards
CFD - Computational Fluid Dynamics
DFT - Discrete Fourier Transformation
ERM - Equivalent Rectangular Method
FFT - Fast Fourier Transformation
FNPP - Floating Nuclear Power Plant
FRS - Floor Response Spectrum
FSI - Fluid-Structure Interaction
GBS - Gravity-Based Structure
GDWP - Gravity Driven Water Pool
LOCA - Loss of Coolant Accident
NPP - Nuclear Power Plant
ONPP - Offshore Structures of Nuclear Power Plant
STFT - Short Term Fourier Transform
TLD - Tuned Liquid Damper
UDF - User-Defined Function
VOF - Volume of fluid Method
1
1. INTRODUCTION
1.1 Background
For the last few decades, liquid sloshing is an important problem in several areas including
nuclear, aerospace, and seismic engineering. Considering the safety aspects of the liquid
storage structures, liquid sloshing is one of the significant problems needs to be addressed.
Particularly, the load generated on the structure due to liquid sloshing is very important to
ensure structural integrity and liquid spillage due to oscillations. Such an oscillatory
motion of the liquid in its container is termed as sloshing. Under the seismic load, severe
accidents might be possible due to this kind of oscillatory motions. This sloshing can
cause possible leakage (Malhotra, 1997), pollution to the surrounding area, and elephant-
foot buckling due to bucking of the tank (Ibrahim, 2005) wall of NPPs.
To generate the nuclear power safely and continuously, technology development for
enhanced safety is vital for future nuclear power plants. As discussed by Lee at al. (2013),
at the beginning of the 1950s, the USA and USSR began to develop Floating Nuclear
Power Plants (FNPPs). Now days, various countries are focusing their studies towards
floating, Gravity-Based Structure (GBS), and submerged Offshore Structures of Nuclear
Power Plants (ONPP) (Gerwick, 2007, Lee et al., 2011). To enhance the safety of nuclear
power, the conventional nuclear power plant (NPP) can be moved from land to ocean.
However, most of the present working nuclear reactors are land-based. And some of the
future land-based reactors are also being designed with improved safety features by
various countries.
The typical Indian advanced Heavy water reactor (AHWR) as shown in Fig. 1.1 is
land-based with enhanced safety features which have an annular-sectored water pool
called as Gravity Driven Water Pool (GDWP),on its dome region of the primary
containment of reactor building (Sinha and Kakodkar, 2006). This present work is focused
on the liquid sloshing in a GDWP of AHWR. The water in the pool serves as a heat sink
2
of the residual heat removal system and several other passive systems. As reported by
Sinha and Kakodkar (2006), this GDWP is divided into eight compartments as shown in
Fig. 1.2, which are interconnected to each other. Each compartment of water pool contains
an isolation condenser for core decay heat removal during shutdown. Water in the pool is
used to condense the steam flowing through the isolation condenser during reactor
shutdown and also functions as a suppression pool to cool the steam and air mixture
during Loss of Coolant Accident (LOCA). This water pool provides cooling to the fuel in
passive mode during first fifteen minutes of LOCA by high pressure injection from
advanced accumulators and later for three days.
Fig.1.1 AHWR building (Sinha and Kakodkar, 2006)
1.2 Problem Definition
To ensure the safety of the reactor against the seismic load, the GDWP located on the
primary containment needs to be investigated for liquid sloshing under simulated seismic
excitations. And also needs to be studied the control of sloshing using passive bafflesin
GDWP of AHWR. From this study, one can be estimated theslosh height, dynamic
pressure on the wall, water spilled out conditions (i.e., possibility of water spillage),
differe
rectan
slosh
proced
determ
regula
geome
Two c
and ot
the pr
and Eq
1.3 Ob
The ob
(i)
(ii)
ent type o
ngular and
frequency,
dure for the
mine. Many
ar geometri
etries.
configuratio
ther one is p
resent case,
quivalent R
bjectives
bjectives of
To estim
7024, AC
) To eval
conditio
of bafflesar
cylindrical
, slosh ind
e baffled GD
y numerical
ies. Althou
on of GDWP
partially cur
fully and p
Rectangular
f this presen
mate liquid n
CI 350.3 (2
luate slosh
ns, the wate
(a) Isometr
Fig.1.2 GD
rrangement
tanks have
duced press
DWP with
l and experi
ugh only fe
P’s outer w
rved. Furthe
partially cu
Method (ER
nt work are,
natural slos
2001) usingE
heights an
er will be sp
ric view
DWP locate
3
and its e
e a detailed
sure and s
considerab
imental stu
ew studies
wall are cons
er details ar
urved wall G
RM) (ACI 3
hing freque
ERM in GD
nd conditio
pilled out du
ed top of the
effect onliq
d analytical
slosh heigh
le interactio
udies have b
have stud
sidered in th
re discussed
GDWP geo
350.3 (2001
enciesusing
DWP.
ns for wat
uring sloshi
e primary co
quid oscill
l solution p
ht. Howeve
on effects i
been report
died the ba
his work. O
d in the follo
metries ana
1)).
CFD and d
ter spilled
ing).
(b) Pl
ontainment
ation.The
procedure t
er, the ana
s very diffi
ed on slosh
affled asym
One is fully c
owing chapt
alysed using
design code
(i.e., under
lan
simple
to find
alytical
icult to
hing in
mmetric
curved
ters. In
g CFD
s TID-
r what
4
(iii) To calculate water pool wall displacement and stress induced in the wall due to
the liquid load.
(iv) To estimate the slosh height using sinusoidal excitation for first mode and higher
modes along with the combined three dimensional excitations.
(v) To determine the slosh height under seismic load for given design seismic
motion.
(vi) To study the effect of vertical excitation and also combined (Horizontal and
vertical)excitation on the fluid behaviour.
(vii) To compute the liquid sloshing design parameters usingERM and comparing with
CFDresults.
(viii) To study effect of annular and cap-plate baffles to prevent excessive sloshing.
1.4 Report Organisation
In chapter 2, theoretical formulation has been established for fluid and structural domains
to explore the fluid–structure interaction phenomena for a moving liquid water pool.
Chapter 3 explains the numerical approach and the frequency modes of liquid and
structural domains. The estimated slosh frequencies and slosh height using international
seismic codes TID-7024, ACI 350.3 (2001) and Euro 8 are reported in chapter 4.
Estimated slosh frequencies for all the sectors of GDWP using CFD is shown in Chapter
5. Chapter 6 discusses the validation of present CFD work using experimental and
analytical methods for which a sectored rectangular tank has been taken for analysis.
Chapter 7 explains the effect of amplitude in slosh height and effect of higher modes
under sinusoidal excitation while Chapter 8 converses the seismic load on GDWP sectors.
Chapter 9 discusses the slosh control techniques for the fluid in GDWP tanks. Chapter 10
depicts the conclusions from the above completed work.
5
2 MATHEMATICAL FORMULATIONS FOR FLUID, STRUCTURE
AND INTERACTION
In this report,numerical approach is followed to solve the sloshing problem in GDWP. So
that mathematical formulation for the fluid and structural domains and coupling methods
are discussed in this chapter. The formulations for fluid domain is based on the three-
dimensional time-dependent conservation equations of mass and momentum to determine
the sloshing characteristics. For the structure domain, the equation of motion is utilized to
simulate the displacement of the concrete wall for balancing at the fluid and concrete wall
boundaries.
2.1 Fluid Formulation
The forces acting on the fluid in order to conserve momentum must balance the rate of
change of momentum of fluid per unit volume. For the laminar transient, incompressible
flow with constant fluid properties over the computational domain, the mass continuity
and Navier–Stokes (NS) equations are given as follows.
0=⋅∇ ur (2.1)
FuPuutu rrrrr
+∇+−∇=⎟⎠⎞
⎜⎝⎛ ∇⋅+
∂∂ 2μρ (2.2)
where Fr
is the external force vector and μ , ρ are the dynamic viscosity and density
of the fluid respectively. The external force is the sum of the gravitational and applied
forces. The ur and P denote the velocity vector and pressure of the oscillating fluid.
The aforementioned governing equations are discretized by finite volume approach to
replace the partial differential equations with the resulting algebraic equations for the
entire domain. Using the staggered-grid arrangement, grids of velocities are segregated
from grids of scalars and laid directly on the surfaces of the control volumes for estimating
those convective fluxes across cell surfaces. The well-known Semi-Implicit Method for
6
Pressure-Linked Equations Consistent (SIMPLEC) numerical algorithm is employed for
the velocity–pressure coupling. In SIMPLEC, an equation for pressure-correction is
derived from the continuity equation which governs mass conservation. It is an inherently
iterative method. The under-relaxation technique is also implemented to circumvent
divergence during iterations. The velocities and local pressure can be determined until
convergent criteria are satisfied.
2.2 Structural Formulation
In the structure model, the linear elasticity approach is utilized. After the finite element
analysis of the solid wall under the slosh loading condition, the solid wall displacement
caused by the fluid–structure interactions is assumed to be small and linear. Hence to
simulate the motion of solid portions, the governing equations are written as below.
xszxyxxx
s gzyxt
sρ
ττσρ +
∂∂
+∂
∂+
∂∂
=∂
∂2
2
(2.3)
yszyxyyy
s gzxyt
sρ
ττσρ +
∂
∂+
∂
∂+
∂
∂=
∂
∂2
2
(2.4)
zsyzxzzz
s gyxzt
s ρττσρ +∂
∂+
∂∂
+∂
∂=
∂∂
2
2
(2.5)
Here, sρ and S are the structure density and structure displacement respectively. The g
is the gravity due to acceleration. The linear stress– strain relation can be expressed as
( ) 0σεεσ +−= oDrr . Here,
0ε and 0σ are initial strains and stresses respectively. Here, the
symbol D is the elasticity matrix containing the material properties. A finite element
method is used to solve the solid model with the principal of virtual work. Structural
domains are meshed using similar first order quadrilateral elements.
7
2.3 Coupling between Fluid Flows and Structural Media
The coupling of the fluid and structural response can be attained numerically in different
ways, however in all cases, of course, the conditions of displacement compatibility and
traction equilibrium along the structure–fluid interfaces ( si ) must satisfy the following
conditions.
(i) The fluid and solid wall move concurrently (displacement compatibility).
ssf iondd = (2.6)
(ii) The fluid force (pressure and shear stress) applied on the solid wall is identical to
the wall force exerted to the liquid side (Traction equilibrium).
ssf ionff = (2.7)
where, fd and sd are the displacements, ff and sf are the tractions of the fluid and
solid, respectively, and si is the interface of the fluid and solid domains. These conditions
must be imposed efficiently in the numerical solution.
Fig.2-1Feedback loop of fluid structure interaction (Eswaran et al., 2009)
Input (External Disturbances) like Force.
Output (Dynamic Variables) like Displacement, Stresses.
Structural Response
Fluid Response
Interface Pressure Interface Acceleration
Output (Dynamic Variables) like Velocities and Pressures.
8
In this problem, the solutions are based on partitioned method where separate solutions for
the different domains are obtained. One solution is for fluid and other is for structure from
the independent solvers. At the fluid-structure interface, information for the solution is
shared between the fluid solver and structure solver. The information is exchanged at
interface in a coupled manner. Two way coupling is adopted for calculations (Benra et al.,
2011). Here, portioned strong coupling is used. Interaction (or coupling) between the fluid
and solid response can be viewed as a feedback loop illustrated in Fig. 2.1.
9
3 NUMERICAL METHODOLOGY
3.1 Introduction
This chapter discusses the numerical formulation for liquid sloshing in GDWP. Here, one
sector of GDWP is taken for analysis from eight-sectored water pool (i.e., GDWP) as
shown in Fig. 3.1. Three domains are modelled viz., water pool wall, liquid and air
domains. The dimensions of the water pool are shown in Fig. 3.1. The side wall thickness
is 500 mm and bottom is 1000mm. Height of the water and air is 8 m and 1 m respectively.
Fig.3-1One compartment of GDWP.
3.2 Grid Generation and Boundary Conditions
A grid is an artificial geometric construction that assists governing equations to be solved.
Block structured grid (BSG) has been used to generate the grid, i.e., the flow domain is
split up into a number of topographically simpler domains and each domain is meshed
separately and joined with neighbours. The BSG arrangement for un-baffled and baffled
water pool has been shown respectively in Figs. 3.2 and 3.3. The fluid-structure
interaction is considered by appropriately coupling the nodes that lie in the common
Tank wall
Air
Water
16000
R 27800
R 7000
10000
10
element faces of the two (i.e., fluid and structure) domains. Fig. 3.3 clearly shows that the
fluid and structure domains share the common element face at fluid-structure interface
which may lead to better data transformation between domains. The fluid domain is
divided into 18,000 sub domains and structure wall into 8000.
Fig.3-2Numerical grid of un-baffled GDWP.Fig. 3-3Block structured mesh on fluid and
structure portions.
The NS equations(Eqs. 2.1 and 2.2) are solved in each sub volumes of fluid domains
(liquid and air). Cell centered average value is taken into consideration. In calculations,
the material properties used for Fluid-structure Interaction (FSI) simulations are
summarized in Table 3.1. Top boundary of the air is fixed pressure condition (at
atmosphere condition). Implicit pressure and implicit shear wall condition is applied on
the fluid-structure interfaces. Iteration loops are continued until the corrections are small
enough to satisfy convergence criteria (10-6< R< 10-8).
Table 3-1: Material properties
1 Water Kinematic
Viscosity 1 E-6 m2/sec Density 1000 kg/m3
2 Air Dynamic
Viscosity 1.846E-05 Kg/m
sec Density 1.1614 kg/m3
3 Concrete
wall
Poisson's ratio
0.2 Density 2500 kg/m3
Young Modulus
33E+09 N/m2
11
The choice of method used for the solution of the assembled system of equations can
have a major impact on the overall solution time and solution quality. Here, the algebraic
multigrid is used for solving the system of linear equations. The basic idea of a multigrid
solution is to use a hierarchy of grids, from fine to coarse, with each grid being
particularly effective for smoothing the errors at the characteristic wavelength of the mesh
spacing on that grid. Other iterative solvers are non-optimal in the sense that as the mesh
resolution increases, the convergence rate degrades. (ESI CFD Inc, 2011). The simulation
was carried out on an Intel Xeon, 2.8GHz six core processor workstation and the
simulation ran for approximately 32 CPU hours for 20 seconds sloshing simulation. The
implicit scheme is used for temporal integration and the higher order upwind schemes are
used for the spatial discretization. The free surface elevation (ζ ) has been captured every
0.05 second.
3.3 Volume of Fluid (VOF)Method
For modeling free surface flows, marker and cell (Chen et al., 1997), VOF, level set
method, sigma-transformation (Frandsen, 2004, Chen and Nokes, 2005, Eswaran and
Saha, 2009, 2010, 2011) and meshless method based on smoothed particle hydrodynamics
(Vorobyev et al., 2011) are known methods. Nevertheless, this work adopts VOF method.
The VOF method is developed by Hirt and Nichols (1981) and refined thereafter by
various authors. Since the method is designed for two or more immiscible fluids, a portion
of air is filled above the liquid level for all cases. The air portion is also modeled and
discretized using the 3-D fluid element. In this method, the term, first fluid and second
fluid indicate the air and water domains respectively. It is based on tracking a scalar field
variable f which stands for the distribution of the second fluid in the computational grid. f
specifies the fraction of the volume of each computational cell in the grid occupied by the
second fluid. All cells containing only fluid 2 will take the valuef = 1 and cells completely
filled with fluid 1 is represented by f= 0. Cells containing an interface between air and
water take on a value of fbetween 0 and 1. For a given flow field with the velocity vector
ur and the volume fraction distribution f(and hence the distribution of fluid two) is
12
determined by the passive transport equation:
( ) 0=⋅∇+∂∂ fu
tf r
(3.1)
This equation must be solved jointly with the primary equations of conservation of
mass and momentum, to achieve computational coupling between the velocity field
solution and the liquid distribution. From the f distribution the interface between the two
fluid phases has to be reconstructed at every time step. As depicted in the manual (ESI
CFD Inc, 2011, Glatzel et al., 2008), the VOF method in CFD-ACE+ offers some
additional features like an algorithm to remove the so called flotsam and jetsam caused
due to numerical errors. It is characterized by the generation of tiny isolated droplets of
liquids or gas in the regions of the other medium, especially in regions of high swirl.
13
4 SLOSHING ANALYSIS USING SEISMIC DESIGN CODES
Earthquake prone countries across the world rely on “codes of practise” to mandate that
all structuressatisfy at least a minimum level of safety requirements against future
earthquakes. Indian seismic code IS 1893:1984 had very limited provisions on seismic
design of elevated tanks. In 2002, revised part 1 of IS 1893 has been brought out by the
bureau of Indian Standards (BIS). The ERM approach and use of design codes, slosh
frequencies and slosh height are found and slosh design terms are discussed in detail in
this chapter.
4.1 Introduction
As mentioned in introduction chapter, two GDWP models have taken for analysis. One
is fully curved outer wall and the other one is partially curved wall. Figs. 4.1 and 4.2 are
showing fully curved outer wall and partially curved wall GDWP. The sketch and
dimensions of the partially curved outer wall GDWP are depicted in Fig. 4.3. The ERM
analysis is performed to design against the seismic load.
Fig. 4-1 Fully curved outer wall GDWP. Fig. 4-2Partially curved outer wall GDWP.
Fully curvedouter wall
Inner wallPartially curvedouter wall
14
Fluid in each sector has different dynamic characteristics, however some of the sectors
are symmetric. As shown in sector arrangement Fig. 4.4, the each sector has different
slosh frequencies. The slosh frequencies for sectors, 1 and 5 are same. Also, similarly
sectors 2, 4, 6 and 8 are same and remaining sectors 3 and 7 are same slosh frequencies.
These frequencies are found using ERM approach and listed in Chapter 5 along with
frequencies from CFD approach for comparison.
Fig. 4-3 Sectional view of GDWP (updated
dimensions) Fig. 4-4Plan of GDWP
4.2 Equivalent Rectangular Method (ERM): Since the design code for GDWP (8- sectored annular water pool with outer spherical
wall) geometry is not available directly in seismic design code, as instructed [31], the one
sector s taken alone and ERM analysis is made. The rectangular domain is constructed as
shown in Fig. 4.5. While converting from GDWP fluid domain to equivalent rectangular
domain the following assumptions are taken.
(i) Liquid volume should be equal in GDWP sector and rectangular model.
(ii) Ratio between liquid height (h) and length of the tank (L) is taken almost equal to
the GDWP sector dimensions.
Hydrodynamic forces exerted by liquid on tank wall shall be considered for analysis in
addition to hydrostatic forces. These hydrodynamic forces are evaluated with the help of
spring mass model of rectangular tanks as shown in Fig. 4.6.
Excitation direction
1
23
4
5
67
8
15
Fig. 4-5Equivalent rectangular model Fig. 4-6Spring –mass model
4.3 Spring-Mass Damper Model
For the purposes of incorporating the dynamic effects of sloshing in the pools, it is
convenient to replace the liquid conceptually by an equivalent linear mechanical system.
The equations of motion of oscillating masses and rigid masses are included more easily in
the analysis than are the equations of fluid dynamics. Fig. 4.6 illustrates generalized
spring-masses model for the rectangular tank and the symbols used in the analysis. The
width of the tank are 2L and height of the liquid is denoted as h. The center of the mass of
the liquid is represented as C.G, while, the locations (Hn) of the masses are references to
the C.G. The tank is excited by a small time-varying linear acceleration . Rigidly
attached mass is denoted as m0, while the convective (slosh) masses are showed as m1
through mn .The deflection of the mass is represented as xn which is relative to the tank
walls as a result of the tank motion.
The mathematical equations can be derived from static and dynamic properties of spring-
mass model. These equations and derivations can be found in ACI, Housner (1963) and
Dodge (2000) for simple geometries like rectangular and cylindrical tanks. According to
static properties, the sum of all the masses must be equal to the liquid mass and center of
mass of the model must be same elevation as the liquid. These can be derived as follows,
m0+ m1+ m2+…..+ mn = mliq (4.1)
Z
H = 8m
W=11 mL= 11.5 m
x
yRectangular model
GDWP/ sector
(11.5 x11 x 8) = 1012 m3
1000 m3
16
. m0H0+ m1H1+ m2H2+ ….+ mnHn = 0 at C.G (4.2)
Equation of motion can be derived by inserting the acceleration terms and applying
static properties into the force equation. The equations of motion for each of the spring-
masses is expressed as
0 (4.3)
From the above equations forces acting on the rectangular tank can be estimated. The
slosh height is estimated using the well derived equations as listed here (Eqs. 4.1 to 4.8).
The convective and impulsive masses are,
.)12(
/)12tanh(833 hn
LhnLmm liqn −−
=π
π (4.4)
.)12(
/)12tanh(81
330 ⎥⎦
⎤⎢⎣
⎡−−
−=hn
LhnLmm liq ππ
(4.5)
Height of the convective masses and natural frequency,
.)12(
/)12tanh(8)12(
sinh
12/)12tanh(
)12(2 33 hnLhnL
Lhn
Lhnn
LhH n −−
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
−−−
−−=
ππ
ππ
π (4.6)
⎟⎠⎞
⎜⎝⎛=
LH
Lg
58.1tanh58.1ω (4.7)
The first mode slosh height,
1
)58.1coth(527.0
2
max,1
−==
Lg
LhLd
h
n
θω
where, n =1, (4.8)
Based on the above formula, the slosh height and other parameters have been calculated
and tabulated for sector 1 of fully curved outer wall GDWP and partially curved outer wall
GDWP in Tables 4.1 and 4.2 respectively.However the detailed computations for both
models are shown in Tables 4.3 and 4.4.
17
Table 4-1: Slosh height estimation through seismic codes for sector 1 of fully curved outer
wall GDWP
Length direction Width direction unit
Length of pool 2L 11.5 11 m
Width of pool W 11 11.5 m
Height of pool 10 10 m
Height of water 8 8 m
Density of water 1000 1000 kg/m3
Mass of water 1.01E+06 1.01E+06 kg
Volume of water 1012 1012 m3
Convective acceleration 0.13 0.145 g
Impulsive acceleration 1.11 1.88 g
Convective acceleration 1.275 1.422 m/sec2
Displacement (A) 0.485 0.515 m
Convective
frequency
ω 1.622 1.662 rad/sec
f 0.258 0.265 Hz
Slosh height
TID-7024 0.778 0.856 m
Slosh height
ACI 350.3 (2001) 0.748 0.798 m
18
Table 4-2: Slosh height estimation through seismic codes for sector 1 of partially curved
outer wall GDWP
Length direction Width direction unit
Length of pool 2L 13.5 12.5 m
Width of pool W 12.5 13.5 m
Height of pool 8 8 m
Height of water 6 6 m
Density of water 1000 1000 kg/m3
Mass of water 1.01E+06 1.01E+06 kg
Volume of water 1012.5 1012.5 m3
Convective acceleration 0.13 0.145 g
Impulsive acceleration 1.11 1.88 g
Convective acceleration 1.275 1.422 m/sec2
Displacement (A) 0.627 0.632 m
Convective
frequency
ω 1.427 1.501 rad/sec
f 0.227 0.239 Hz
Slosh height
TID-7024 0.893 0.953 m
Slosh height
ACI 350.3 (2001) 0.878 0.906 m
4.4 Result and Discussions
The fully curved wall GDWP slosh height and other calculations are shown in Table 4.2,
while slosh calculations are shown for partially curved wall GDWP in Table 4.3.The
approved model of GDWP’s outer wall is fully curved (Fig. 4.5). During internal review
meeting, a partially curved wall GDWP model (Fig. 4.6) is proposed (at AHWR Review
Meeting 201, BARC). To address the both cases, two cases are studied and tabulated in
Tables 4.2 and 4.3.
19
Table 4-3: X and Y direction values of sector 1 of fully curved outer wall GDWP.
Sl. No
Term
Longitudinal
(x)direction
Lateral (y) direction
Unit Comments
1 Impulsive mass (mi) 6.88E+05 7.06E+05 kg Sum of Impulse and convective mass is slightly higher than total mass of fluid. 2 Convective mass (mc) 3.75E+05 3.60E+05 kg
3
Height of the impulsive mass above the bottom of the tank wall (hi) (without considering base pressure)
3.00 3.00 m See the Fig. 4.7
4
Height of the convective mass above the bottom of the tank wall (hc)(without considering base pressure)
5.09 5.15 m See the Fig. 4.6
5
Height of the impulsive mass above the bottom of the tank wall (hi*)(with considering base pressure)
4.88 4.73 m See the Fig. 4.7
6
Height of the convective mass above the bottom of the tank wall (hc*)(with considering base pressure)
5.91 5.87 m See the Fig. 4.6
7 Wall deflection (d) due to load
0.0192 0.0180 m Considered as fixed at three edges and free at top
8 Impulsive frequency (fi)
3.600 3.715 hz ACI 350.3 (2001)
9 Convective frequency (fc)
0.243 0.255 hz ACI 350.3 (2001)
10
Seismic co-efficient (Ah)
FRS at 137 m height of AHWR building. Impulsive (Ah)i 1.11g 1.88g
Convective (Ah)c 0.13g 0.145g
20
11
Total shear force (V) at bottom of the wall
13765.85 23402.02 KN Lateral base shear 23.2 % of total seismic weight in x direction while same in 24 % in y direction
Impulsive Vi 13757.55 23396.42 KN
Convective Vc 477.86 512.11 KN
12
Total bending moment at bottom of the wall (M)
50.72 85.79 MN SRSS rule as followed in all international code except Eurocode 8 (1988).
Impulsive Mi 50.66 131.71 MN Convective Mc 2.43 3.52 MN
13 Over turning moment at bottom of base slab. (M*)
78.59 131.76 MN-m Housner(1963)
14 Impulsive time period (Ti)
0.28 0.27 sec ACI 350.3 (2001) and NZS 3106 (1986)
15 Convective time period (Tc)
4.11 3.92 sec ACI 350.3 (2001) Housner (1963)
16 Slosh height (hs) 0.748 0.798 m Free board is 1 m (Importance factor 1 )
17
Impulsive pressure on wall (y=0) (Piw)
63.89 106.15 KN/m2 ACI 350.3 Housner(1963) See the Fig. 4.7
Impulsive pressure on top of base (y=0) (Pib)
30.74 51.83 KN/m2
18
Convective pressure on wall (y=0) (Pcw)
1.34 1.29 KN/m2 ACI 350.3 Housner(1963) See the Fig. 4.7
Convective pressure on wall (y=h) (Pcwt) 6.11 6.52 KN/m2
Convective pressure on top of base (y=0) (Pcb)
1.34 1.29 KN/m2
19
Pressure due to wall inertia (Pww)
13.61 23.05 KN/m2 ACI 350.3 ; Housner(1963) Pressure due to
vertical excitation (Pv) 41.20 41.20 KN/m2
20 Maximum hydrodynamic pressure (P)
87.78 135.62 KN/m2 Hydro static pressure is 78.8KN/m2.
21
Table 4-4: X and Y direction values of sector 1 of partially curved outer wall GDWP.
Sl. No
Term
Longitudinal
(x)direction
Lateral (y) direction
Unit Comments
1 Impulsive mass (mi) 4.99E+05 5.32E+05 kg
Sum of Impulse and convective mass is slightly higher than total mass of fluid.
2 Convective mass (mc) 5.33E+05 5.06E+05 kg
3
Height of the impulsive mass above the bottom of the tank wall (hi) (without considering base pressure)
2.25 2.25 m See the Fig. 4.7
4
Height of the convective mass above the bottom of the tank wall (hc)(without considering base pressure)
3.41 3.47 m See the Fig. 4.8
5
Height of the impulsive mass above the bottom of the tank wall (hi*)(with considering base pressure)
5.34 4.96 m See the Fig. 4.7
6
Height of the convective mass above the bottom of the tank wall (hc*)(with
considering base pressure)
5.67 5.31 m See the Fig. 4.8
7 Wall deflection (d) due to
load 0.0068 0.0060 m Considered as fixed at
three edges and free at top
8 Impulsive frequency (fi) 6.063 6.433 hz ACI 350.3 (2001)
9 Convective frequency (fc) 0.224 0.239 hz ACI 350.3 (2001)
10
Seismic co-efficient (Ah) FRS at 137 m height of AHWR building.
Impulsive (Ah)i 1.11g 2.16g
Convective (Ah)c 0.13g 0.145g
11
Total shear force (V) at bottom of the wall
11224.95 22294.37 KN
Lateral base shear 23.2 % of total seismic weight in x direction while same in 24 % in y direction
Impulsive Vi 11204.35 22282.76 KN
Convective Vc 679.80 719.36 KN
22
12
Total bending moment at bottom of the wall (M)
32.51 63.96 MN SRSS rule as followed in all international code except Eurocode 8(1988).
Impulsive Mi 32.42 116.76 MN
Convective Mc 2.32 4.54 MN
13 Over turning moment at
bottom of base slab. (M*) 60.57 116.85 MN-m Housner (1963)
14 Impulsive time period (Ti) 0.16 0.16 sec ACI 350.3 (2001) and NZS 3106 (1986)
15 Convective time period
(Tc) 4.46 4.18 sec
ACI 350.3 (2001) Housner (1963)
16 Slosh height (hs) 0.878 0.906 m Free board is 1 m (Importance factor 2 )
17
Impulsive pressure on wall (y=0) (Piw)
54.33 104.29 KN/m2
ACI 350.3 Housner (1963) See the Fig. 4.7
Impulsive pressure on top of base (y=0) (Pib)
20.73 41.95 KN/m2
18
Convective pressure on wall (y=0) (Pcw)
3.32 3.10 KN/m2
ACI 350.3 Housner (1963) See the Fig. 4.7
Convective pressure on wall (y=h) (Pcwt) 7.17 7.41 KN/m2
Convective pressure on top of base (y=0) (Pcb)
3.32 3.10 KN/m2
19
Pressure due to wall inertia (Pww)
13.61 26.49 KN/m2 ACI 350.3 ; Housner (1963)
Pressure due to vertical excitation (Pv)
30.90 30.90 KN/m2
20 Maximum hydrodynamic
pressure (P) 74.71 134.42 KN/m2 Hydro static pressure
(58.86KN/m2).
23
Fig.4-7 Impulsive pressure on wall Fig. 4-8 Convective pressure on wall
Formulas used here are taken from IS 1893 (Part 1):2002, ACI 350.3, Housner (1963) and
NZS 3106 (1986). The deflection of the wall due to this pressure, it can be considered to
be fixed at three edges and free at top. Deflection of wall can be obtained by performing
analysis of wall or by classical analysis using theory of plates. However, here, simple
approach is used as suggested in international codes. Sum of Impulse and convective mass
is slightly higher than total mass of fluid. But however up to 3% is allowed.
24
5 ESTIMATION OF SLOSH FREQUENCIES USING CFD
The estimation of modal frequency through numerical investigation is discussed in the
chapter and followed by the test cases are made for the complete numerical investigation.
5.1 Introduction
The total fluid dynamic pressure for a flexible tank partially filled with liquid undergoing
a seismic motion consists of three components. The first pressure component is called the
impulsive pressure which varies synchronously with input tank base motion and the tank
wall is assumed to be rigid, moving together with the tank base. The second component is
caused by the fluid sloshing motion. This pressure is generally referred to as the
convective pressure or non-impulsive pressure, while the third component is induced by
the relative motion of the flexible tank wall with respect to the tank base (Chang et al.,
1989). This fluid-structure interaction effect results in the dynamic characteristics of the
tank-liquid system to be notably different from that of a rigid tank has led to the inclusion
of a third hydrodynamic component to quantify the dynamic response of flexible tanks
namely the ‘flexible-impulsive’ component. Methods for determining the contribution of
the flexible-impulsive component to the total response such as base shear, overturning
moments, wall stresses of tanks under seismic excitations have been proposed by various
researchers (Haroun and Housner, 1981, Tedesco et al., 1989).
In the present study, the problems are restricted to the water tank under surge motion.
During analysis, the complete flow regime is assumed to be laminar. Some localized
turbulence effects may be caused at the sheared interface during sloshing which should not
affect the fluid-wall interactions and global fluid behaviour. This type of gravity force
dominant flow problems are mainly considered as inviscid. It had been proved that the
free surface oscillations of low viscosity fluids in partly-filled tank persist over long
durations (Kandasamy et al., 2010).
25
5.2 Modal Analysis of the GDWP
The sinusoidal excitation is applied on the fluid domain in terms of acceleration force via
user subroutine functions. As discussed earlier, GDWP is divided into eight
compartments. Among the eight, one compartment has taken for analysis. The sketch and
dimensions of the water pool are depicted in Fig. 3.1. The modal analysis is required to
find the inherent dynamic properties of the any domain in terms of its natural frequencies.
In the beginning, the first mode natural frequency of water pool has been calculated by
free vibration i.e., free surface elevation (ζ ) of liquid shown in Fig. 5.1.
Fig. 5-1Free surface elevation of liquid
under free vibration. Fig. 5-2 Spectrum analysis of signal.
Table 5-1: Fully curved outer wall GDWP natural frequency in Hz for different sectors
Mode
number Sectors 1 and 5 Sectors 2,4, 6 and 8 Sectors 3 and 7
Analytical CFD Analytical CFD Analytical CFD
1 0.257 0.312 0.212 0.292 0.264 0.309
2 0.368 0.393 0.313 0.412 0.377 0.421
3 0.451 - 0.384 - 0.461 -
4 0.521 - 0.443 - 0.533 -
0 2 4 6 8 10 12 14-1500
-1000
-500
0
500
1000
1500
Free
sur
face
ele
vatio
n (m
m)
Time (Sec)
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.00
50
100
150
200 ω1=0.312 Hz
Frequency (Hz)
Pow
er
26
Table 5-2: Partially curved outer wall GDWP natural frequency in Hz for different sectors
Mode
number Sectors 1 and 5 Sectors 2,4, 6 and 8 Sectors 3 and 7
Analytical CFD Analytical CFD Analytical CFD
1 0.226 0.242 0.201 0.231 0.228 0.247
2 0.338 0.342 0.310 0.332 0.407 0.390
3 0.416 - 0.383 - 0.499 -
4 0.48 - 0.442 - 0.576 -
The first mode natural frequency is observed as 0.312 Hz by spectrum analysis using
Fast Fourier Transformation (FFT) (Fig. 5.2). The FFT is a faster version of the Discrete
Fourier Transform (DFT). The FFT utilizes some clever algorithms in much less time than
DFT. The FFT is extremely important in the area of frequency (spectrum) analysis since it
takes a discrete signal in the time domain and converts that signal into its discrete
frequency domain representation. Tables 5.1 and 5.2 show the comparison of slosh modes
computed through analytical and CFD approaches for fully and partially curved outer wall
GDWP respectively. The dominant modes of the structure i.e., the first four mode shapes
are illustrated in Fig. 5.3. The material properties are listed in Table 3.1.The case studies
for numerical simulation were summarized in Table 5.3.
Mode 1 (11.21 Hz) Mode 2 (17.61 Hz) Mode 3 (21.16 Hz) Mode 4 (27.89 Hz) Fig. 5-3 Dominant mode shapes of structure.
5.3 Effect of Excitation Frequency
Fig. 5.4 shows the numerical results for the liquid heights at the left, right and center point
in a three-dimensional GDWP subject to harmonic motions under 0.1 m amplitude for
different frequencies.
27
Table 5-3: Numerical cases taken for investigation.
Sl. No Excitation
Sectors consider
ed in GDWP
tank
Baffle Excitation Amplitude
(m)
Excitation
direction
Excitation Frequency
(Hz)*
Condition
1
Sinusoidal excitation
1 (Fully curved outer wall
GDWP)
No baffle
0.01
x
1ω
Fixed wall
2 0.02 1ω
3 0.06 1ω
4
0.1
1ω 5 0.5 1ω
6 0.8 1ω
7 1.2 1ω
8 0.1 1ω Flexible wall
9 Annul
ar 0.1
1ω Flexible wall
10 5 1ω
11 10 1ω
12 Cap-Plate
0.1 1ω Flexible
wall 13 5 1ω
14 10 1ω
15 Sinusoidal excitation
3 (partially curved outer wall
GDWP)
No baffle
0.03 xyz 1ω Flexible wall
16 Earthquake excitation
3 (partially curved outer wall
GDWP)
No baffle
Max. floor response acceleration
xyz Design FRS data
Flexible wall
*First mode natural frequency of liquid is 1ω =0.312 Hz.
When the excitation frequencies are close to the natural frequency as shown in Figs. 5.4
(a) and (c), the beat phenomena are noticeable (Eswaran et al., 2009). It can be observed
from Fig. 5.4 (d) that when the excitation frequency is far-off from the natural frequency,
i.e., 1.92 rad/sec, the liquid heights are very small and frequency is equal to excitation
28
frequency. When the frequency is almost near to the first mode natural frequency, i.e. Fig.
5.4 (b), the amplitude grows monotonically with time. There is a slight difference in liquid
elevation between right and left corner of the water pool.
Fig.5-4 Time history of free surface elevation at 0.1 m excitation amplitude and
for different excitation frequencies.
0 2 4 6 8 10 12 14 16 18 20 22 24-2
-1
0
1
2
Right Left Center
Time (Sec)
-2
-1
0
1
2
Right Left Center
Free
sur
face
Ele
vatio
n (m
)
-2
-1
0
1
2
Right-2
-1
0
1
2
ωx/ω1 = 0.5
ωx/ω1 = 0.8
ωx/ω1 = 0.99
ωx/ω1 = 1.2
(d)
(c)
(b)
(a)
Right Left Center
29
6 EXPERIMENTAL AND ANALYTICAL VALIDATION OF CFD
SIMULATIONS
In this chapter, slosh height is computed through experimental and analytical methods
and compared with CFD procedure for validation.
6.1 Case Study 1: Analytical Validation
In this section, a 2D model partially filled tank has been taken and the liquid elevation has
been captured under sinusoidal excitation by numerical as well as analytical relation. The
2-D rigid tank which is 570 mm long and 300 mm high is excited with )sin( tA ω as shown
in Fig. 6.1. The water depth is 150 mm and excited amplitude is 5 mm. The lowest natural
frequency 1ω for this case is 6.0578 rad/sec. The natural frequency is calculated from
Equation 6.2. Liquid free surface elevation has been calculated from the following third
order analytical relations (Faltinsen et al., 2000) and compared with the present numerical
simulation results for frequency ratio 0.583.
⎟⎟⎟⎟⎟
⎠
⎞
⎜⎜⎜⎜⎜
⎝
⎛
⎟⎟⎟⎟⎟
⎠
⎞
⎜⎜⎜⎜⎜
⎝
⎛
−−−
+
⎟⎟⎠
⎞⎜⎜⎝
⎛
−−
−−
++
+= )2cos(
)cos()4(
3
2
1)2cos(*
)4(2
33
8
1
8
1
)cos()cos(),(
222
22
22422
2
22
22
224
4
224
4
224
2
xk
tkgt
kgkgkg
gaxktAtx n
nnnn
nnnnn
nnn
nn
n
nn
n
nn
nnn
ωωωω
ωωωω
ωωωω
ωω
ωω
ωωζ (6.1)
where, the linear sloshing frequencies
)tanh( snnn hkgk=ω and )2tanh(22 snnn hkkg=ω . (6.2)
The initial conditions are )cos(),(0
xkanx xt=
=ζ and 0),(
0=
=tzxφ , where A is the
amplitude of the initial wave profile, bnk n /π= is the wave profile for n = 0,1,2... , hs is
still water level and x is the horizontal distance from the left wall. This analytical result is
30
compared with the present numerical approach. From Fig. 6.2, it is observed that the free
surface elevation of analytical and present numerical coincides with each other.
Fig.6-1 The sketch of the 2-D rigid rectangular
tank. Fig.6-2 Comparisons of free surface elevation.
Fig.6-3 Details of the experimental setup.
6.2 Case Study 2: Experimental Validation
This case study shows the comparison of experimental and numerical results. For this
purpose, a model square tank with sectored arrangement was built and experiments were
conducted. The experiments were performed on a shake table (1.2 m x 1.0 m) coupled
with a servo-controlled hydraulic actuator of 250 KN capacity. The test setup is specially
designed for the sloshing experiments as shown in Fig. 6.3. The perspective view of the
setup is shown in Fig. 6.4 with the actuator coupling arrangement.
570
Probe
20
All dimensions are in mm
Water
Air
300xy
Time (Sec)
Free
surfa
ceel
evat
ion
(m)
0 2 4 6 8 10-0.015
-0.01
-0.005
0
0.005
0.01
0.015
AnalyticalPresent Numerical
Platform
Liquid A sin (ω t)
DAS
10020
0
500250
1000
A
arrang
all dir
install
Water
(a) S
Fig.6
10
20
30
40
50
60
70
80
Pres
sure
(Pa)
square tan
gement as s
rections. Pr
led at 100 m
r fill level in
Fig.6-4
Square tank
6-6 Compari
0 1 20
00
00
00
00
00
00
00
00
nk has 1 m
shown in Fi
ressure var
mm (positio
n the tank is
4 Experimen
at excitatio
1ω .
ison of expe
3 4 5Time (Sec
m length an
ig. 6.5 for th
riations are
on 1) and 2
s maintained
ntal setup.
on frequency
erimental an
(
6 7 8
Numerical Experimenta
c)
31
nd 0.5 m h
he experim
e sensed by
200 mm (po
d as 250 mm
y 0.57 (
nd CFD pre
(position 2)
9 10
al
Pres
sure
(Pa)
height alon
ments in orde
y two flush
osition 2) f
m.
Fig.6-5 Is
(b) Sectored
fr
essure data a
).
0 1 20
100
200
300
400
500
600
700
800
Pres
sure
(Pa)
ng with rem
er to allow
h type pre
from the bo
sometric vie
square ta
d square tan
requency 0.
at 200 mm f
2 3 4 5
Time (
movable 4
liquid moti
ssure trans
ottom of the
ew of 4-sect
ank
nk with exci
99 1ω .
from tank b
6 7 8
Sec)
Numerical Experimen
Baffle
Wate
-sector
ions in
sducers
e tank.
tored
tation
ottom
9 10
ntal
er
32
Fig.6-7Actuator with shake table Fig.6-8 Water oscillation during the
excitation
Fig.6-9Water spill out snapshots during base excitation in an 4-sectored rectangular tank
33
For this case the liquid natural frequency has been calculated as 0.71 Hz through
analytical relation and sine sweep experiments. The comparison of experimental and
numerical time history of pressure at position 2 for square tank and 4- sectored square tank
cases are shown in Fig. 6.6 (a) and (b) respectively. For these cases, the excitation
frequency ratio is taken as 0.57 and 0.99 of the first mode square tank. The comparisons of
numerical with experiment results are shown and found the CFD results are good in
agreement with the experiment. Figs. 6.7 through 6.9 show the snap shots of the
experimental setup to display the tank and actuator arrangement and water spill outs.
In this chapter, experimental and analytical studies are also performed to validate present
numerical results. For experimental validation, a simple square and four-sectored squared
tanks was taken. The pressure variations were captured at different locations under the
surge motions of the tank and found the CFD results are good in agreement with the
experiment.
34
7 CFD SIMULATION OF LIQUID SLOSHING CONTROL IN
GDWP UNDER SINUSOIDAL EXCITATION
In this chapter, the detailed sloshing studies are carried out for studying the effect of
amplitude on liquid sloshing and structural analysis to compute the wall displacement and
induced stress. Un-baffled and baffled tank configurations are compared to show the
reduction in wave height under sinusoidal excitation. For this, annular baffle and cap-plate
baffles are taken for analysis. The slosh height is compared between un-baffled and
baffled configurations under design acceleration.
7.1 Effect of Amplitude of GDWP
Fig. 7.2 depicts the effect of excitation amplitude (A) under its first mode natural
frequency. For this purpose, the non-dimensional free surface elevation is captured at the
right corner of the water pool for 20 seconds. If the excitation amplitude is increased, the
fluid response becomes large. Fig. 7.2 is plotted with the assumption of with and without
fluid-structure interaction conditions. In the case of without FSI, the boundaries are
considered as rigid wall. It is also found in the Fig. 7.3 that FSI consideration has slight
more free surface elevation than the without FSI. It is caused due to the interaction of the
fluid domain with structure produces the relative pressure component. However, there is a
large gap between the first mode frequency of the structure and fluid portions. Deviations
are not high as the excitation frequency is low and faraway from the structural first mode
frequency. It is also observed that the amplification of the fluid motion is relatively larger
at lower amplitude while at the higher amplitude; the amplification of free surface
elevation is less than the lower amplitude case. Fig. 7.1 shows the power spectral density
of liquid elevation wave at 0.1 m excitation amplitude. Closer to natural frequency, a
single dominant frequency is absorbed. The phase-plane diagram is plotted in Fig. 7.3,
shows that non-linearity exists in the flow.
35
Fig.7-1Power spectral density for amplitude 0.1 m and 1ω =0.312 Hz.
Fig.7-2Time history of non-dimensional free surface elevation at 1ω =0.312 Hz for different excitation amplitudes
Fig.7-3Phase-plane diagram for amplitude 0.1 m and 1ω =0.312 Hz.
Pressure waves are captured in different locations of the water pool and the locations
(i.e., A through E) are depicted in Fig. 7.4. Positions A through C are 1 m below from the
liquid free surface and positions D and E are in 5 m and 8 m from the free surface
respectively. The time histories of pressure at the different places of the water pool for un-
baffled water pool are plotted in Fig. 7.5 for 0.1 m amplitude and 1ω =0.312 Hz. It can be
seen from the Fig. 7.5 that when the water pool is excited, the impulse pressures occur
because of the relatively large amplitude of the external excitation. If liquid oscillation is
not controlled efficiently, sloshing of liquids in storage water pools may lead to large
dynamic stress to cause structural failure. On the other hand, if the baffle exists in the
water pool, the dynamic pressure will be minimal. The horizontal displacement histories
0 2 4 6 8 10 12 14 16 18 20-40
-20
0
20
40
Amplitute = 0.01 m
Time (Sec)
-40
-20
0
20
40
Amplitute = 0.02 m
ξ / A -40
-20
0
20
40
(d)
(c)
(b)
(a)
Amplitute = 0.06 m
-40
-20
0
20
40
Amplitute = 0.1 m (FSI) Amplitute = 0.1 m (without FSI)
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
0.000
0.005
0.010
0.015
0.020
0.025
Frequency (Hz)
Pow
er
ζ/A
(∂ζ/
∂t)/(
A*ω
)
-15 -10 -5 0 5 10 15 20-1500
-1000
-500
0
500
1000
1500
36
of the container inner and outer walls are drawn in Figs. 7.6 and 7.7. The Fig. 7.1 shows
that the steady state values are reached from around 12 sec for 0.1 m excitation amplitude.
The displacement is captured at 12.04 sec. The displacement frequency is almost equal to
wave frequency. And it can be seen that the horizontal displacement is symmetric in both
side walls as shown in Fig. 7.6.
Fig.7-4Pressure point locations. Fig.7-5Time history of pressure at various locations for 0.1 m amplitude sec and 1ω=0.312 Hz.
Fig.7-6Wall horizontal displacement contour at time 1.417 sec
Fig.7-7Outer and inner walls x-displacement versus time
0 2 4 6 8 10 12 14 160
20
40
60
80
100
120 Position A Position B Position C Position D Position E
Pres
sure
(Kpa
)
Time (Sec)
0 5 10 15 20
-0.0010
-0.0005
0.0000
0.0005
0.0010 Displacement of outer wall Displacement of inner wall
Wal
l dis
plac
emen
t in
x di
rect
ion
(m)
Time (Sec)
37
7.2 Structural Analysis
Fig. 7.8 shows the VonMises stress on water pool wall at 12.4 sec. The maximum stress
created on the side wall is observed as 7 X 106 N/m2. Sloshing occurs primarily at the
liquid surface and oscillates, exerting forces on the tank structure walls. The liquid free
surface profile has a positive gradient when it moves towards right side. As soon as the
free surface elevation reaches its peak at the right wall, fluid vertical velocity will become
zero. Then, due to its own gravity and external applied forces move the liquid free surface
to down. Now, the direction of the fluid velocity switches from right to left and the
magnitudes of these velocities continue to increase until they reach their maximum. This
cycle will continue until the free surface stops its oscillations which may happen due to
the removal of external excitation to the system. Fig. 7.9 shows the velocity magnitude
from right to left at the time of 12.4 sec. During the surge motion of the water pool, a
single directional standing wave is moving upward and downward direction inside the
water pool.
Fig. 7-8VanMises stress contour at time 12.4 sec.
Fig. 7-9 Sectional view of velocity
magnitude at time 12.4 sec.
7.3 Effect of Baffles in GDWP
Tanks of asymmetric shapes and tanks with baffles, give rise to complications in fluid-
structure interaction, which is not amenable to analytical solution. The studies of liquid
sloshing in a tank with baffles are still very necessary (Eswaran and Saha, 2011; Xue et
al., 2012). Baffle is a passive device which reduces sloshing effects by dissipating kinetic
38
energy due to the production of vortices into the fluid. The linear sloshing in a circular
cylindrical tank with rigid baffles is being investigated by many authors in the context of
spacecraft and ocean applications. The shapes and positions need to be designed with the
use of either numerical model or experimental approaches. Nonetheless, the damping
mechanisms of baffle are still not fully understood. The effects of baffle on the free and
forced vibration of liquid containers were studied by Gedikli and Erguven (1999), Biswal
et al., (2006). To the author’s knowledge, there is a very limited set of analytically
oriented approaches to the sloshing problem in baffled tanks.
Here, two types of baffles are taken for analysis. First one is an annular baffle as
depicted in Fig. 7.10 (a) and (b). Few authors worked on this annular baffle for their own
geometries mainly two dimensional. This article is focused on annular baffle for a three
dimensional annular cylindrical water pool. Biswal et al. (2006) found that the baffle has
significant effect on the non-linear slosh amplitude of liquid when placed close to the free
surface of liquid. The effect is almost negligible when the baffle is moved very close to
the bottom of the tank. Past investigations also convey that the performance of the annular
baffle is better when it is near to the liquid free surface (Eswaran et al., 2009). Second one
is cap-plate baffle or shroud as shown in Fig. 7.10 (c) and (d) which is fixed at center of
the water pool. (More details about baffle for thermal stratification can be found in
Vijayan, 2010). Under reactor shutdown conditions, natural convection process starts due
to the strong heat source at the IC wall. Long-time effect of this natural convection
process leads to warm fluid layers floating on the top of gradually colder layers. This
results in a thermally stratified pool having steep temperature gradient along the vertical
plane. Over a period of time, the substantial part of this pool gets thermally stratified
except for the region close to the heat source where there is horizontal temperature
gradient as well (Gupta et al., 2009). Cap-plate model has been proposed to satisfy the
thermal stratification inside the water pool, since this water pool is mainly designed to
perform as a suppression pool to cool the steam and air mixture during LOCA in the
reactor vessel.
39
(a) Annular baffle. (b) Annular baffle in the water pool.
(c) Cap-plate baffle. (d) Cap-plate baffle (or shroud) in the water
pool.
Fig.7-10Baffles shape and its position in the water pool.
Liquid sloshing is violent near free surface and the liquid motion at the bottom of the
tank is almost zero. Here, one could assume that the mounting of the cap-plate does not
disturb the liquid sloshing as it is placed at bottom of the tank. The present work also
estimates and compares the annular and cap-plate baffle performance against the liquid
sloshing under the regular excitation. Fig. 7.11 illustrates that the comparison of liquid
elevations for un-baffle, annular baffle and cap-plate baffle cases. As expected, both the
baffle cases reduce the liquid oscillations as well. It is found that cap-plate baffle is more
effective in reducing the sloshing oscillations and sloshing pressure.
Volume = 7.3 m3
Thickness = 200 mmHeight = 250 mm
Volume = 36 m3
Thickness = 200 mmHeight = 5 m
40
Fig.7-11Comparison of liquid elevations with no baffle, annular baffle and cap-plate baffle cases at 1ω =0.312 Hz and amplitude 0.1 m.
Fig.7-12Effect of baffles at right corner of the water pool case at 15.05 sec and 1ω=0.312 Hz.
To elucidate the performance of baffles, at near right corner the liquid height is
captured and shown in Fig. 7.12. The liquid height is captured near the right corner of
water pool at 14.85 sec under the excitation frequency ( 1ω ) of 0.312 Hz. The liquid height
deviation for no baffle case is found at excitation amplitude between 0.01 m and 0.1 m is
0 5 10 15 20-2
-1
0
1
2
No baffle Annular baffle Cap-plate baffle
ζ (m
)
Time (Sec)
0.01 0.02 0.06 0.17.5
8.0
8.5
9.0
9.5
10.0
Line equivalent to design acceleration
Line of mean water level
Line of the top of the tank
Max
imum
Liq
uid
Hei
ght (
m)
Excitation Amplitude (m)
No baffle with FSI No baffle without FSI Annular Baffle with FSI Cap-plate Baffle with FSI
41
around 1.23 m. At the same time, this value for annular baffle and cap - plate baffle is
around 0.563 m and 0.182 m respectively. Moreover, it is found from the numerical
investigation that the liquid from the GDWP will spill out around 0.06 m excitation
amplitude ( ≈0.023 g acceleration) under liquid first mode frequency. The response
spectrum for the structure will give us design acceleration corresponding to first mode
frequency. Here, design acceleration is 0.16g at 0.312 Hz and corresponds to 0.028 m
equivalent harmonic amplitude. This line is shown as vertical in Fig. 7.12 to mark free
surface elevations for all cases. Baffle reduces the liquid slosh height 0.7 m to 0.3 m at
design acceleration as shown in Fig. 7.12. The Fig. 7.13 is drawn for qualitative
comparison between no baffle, annular and cap-plate baffle case. Here, snap shots of
liquid water pool (under regular excitation of 0.1 m amplitude) for different time step has
been shown.
7.4 Liquid Elevation in Higher Modes of GDWP
To study the effect of higher modes, both the annular and cap-plate baffled water pools
studied previously are analysed in this section. When the tanks are subjected to motions at
higher than first mode, the fluid in the tank will tend to undergo sloshing motions under
near to the same tank frequency. At the beginning of the disturbance, the fluid dynamic
pressure is dominated by the impulsive pressure. After few seconds, sloshing pressure or
convective pressure becomes the dominant component pressure. The small oscillations on
the pressure curve are the impulsive pressures. Figs. 7.14 and 7.15 show the free surface
elevation at 5 and 10 times of 1ω excitation frequency respectively for annular and cap-
plate baffle cases. Due to the strong impact forces at beginning, i.e., 0 - 2 seconds, liquid
rise is more, it reaches the steady state around 10 second on wards.
42
(a) Time at 0 sec (e) Time at 0 sec (i) Time at 0 sec
(b) Time at 14.6 sec (f) Time at 14.6 sec (j) Time at 14.6 sec
(c) Time at 15.6 sec (g) Time at 15.6 sec (k) Time at 15.6 sec
(d) Time at 16.6 sec (h) Time at 16.6 sec (l) Time at 16.6 sec
Fig.7-13Comparisons of free surface profile at different time instant for un-baffled and baffled water pool for 0.1 m amplitude sec and 1ω =0.312 Hz. (Figs. (a)-(d), (e)-(h), (i)-(l) show un-
baffled, annular baffled, cap-plate baffled water pools respectively).
43
Fig.7-14Comparison of free surface elevation of liquid at 0.1 m amplitude and 5
1ω excitation frequency.
Fig.7-15Comparison of free surface elevation of liquid at 0.1 m amplitude and
10 1ω excitation frequency.
0 2 4 6 8 10 12 14 16
-1.2
-0.8
-0.4
0.0
0.4
0.8
1.2
ζ (m
)
Time (Sec)
Annular Baffle Cap-plate baffle
0 2 4 6 8 10 12 14 16
-2.0
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
2.0
ζ (m
)
Time (Sec)
Annular Baffle Cap-plate Baffle
44
8 CFD SIMULATION OF SLOSHING INGDWP UNDER SEISMIC
EXCITATION
In this chapter, a random wave is created using time histories compatible to design floor
response spectrum (FRS). For that, the floor response spectrum (FRS) at 137m height of
reactor building is used to generate the acceleration time history. A User-Defined Function
(UDF) subroutine has been developed to apply the random acceleration as a volume force.
The slosh height and forces on tank wall have been calculated for different excitation
directions. The combined effects in longitudinal and lateral directions are studied.
8.1 Introduction The sloshing studies are usually performed to ensure the safety of plants and to avoid
consequences of any seismic induced accidents.The mechanical mass–spring model
(Chapter 4) based on linear theory is generally employed by design standards to predict
the free surface displacement as well as other seismic design parameters of the liquid
storage tanks. However, nonlinear effects are always present and they occasionally
dominate the sloshing response. These nonlinear slosh effects arise mostly as a
consequence of large wave amplitudes (Chapter 7). Large amplitude waves may appear
when the great earthquakes are accompanied withpretty long period (3 to 10 sec)
components of seismic wave which coincide with the primary natural period of the
contained liquid (Goudarzia and Sabbagh-Yazdi, 2012). In this chapter, the floor response
spectrum (FRS) at 137m height of AHWR building is used to generate the acceleration
time history. This simulated earthquake accelearation data is used to study the sloshing
beheavior in GDWP.
The need to include nonlinearity in the hydrodynamics of the tank–liquid system arises
whenever high amplitude sloshing waves form on the liquid surface, leading to a nonlinear
influence of sloshing wave on the dynamic response of a tank. To estimate the non-linear
sloshing, the fluid momentum equations are solved as discussed in chapters 2 and 3. To
45
estimate the frequencies and slosh height in each sector of GDWP under seismic
excitation through CFD, three sectors are modelled among eight-sectored water pool as
depicted in Fig. 8.1. In each sector, three domains are modelled viz., water pool wall,
liquid and air domains. The sketch of the water pool are depicted as in Fig. 8.1.
8.2 Development of Random Waves
To study the response of GDWP under seismic load a time history is generated from FRS
along three orthogonal directions separately. FRS for AHWR building at 137 m is shown
in Fig. 8.2. The 5% broaden spectrum is generated from FRS. Acceleration time history is
generated using SIMQKE code [33] in three directions separately. These graphs are shown
in Fig. 8.3. A user subroutine function is developed to call the random acceleration data
and applied on the all fluid in terms of gravity force. Implicit pressure and implicit shear
stress conditions have been applied on the fluid solid interfaces. Air at top is at fixed
pressure condition (at atmosphere condition). The free surface elevation has been captured
every 0.005 sec.
8.3 Wavelet Analysis
In the past, the wavelet transform has been used to detect the frequencies at different
regions. In the field of fluid mechanics the wavelet analysis has been used to detect the
multi stable flow regions. In this problem, wavelet tool is used to find the sloshing
frequencies information. The Fourier transforms provide the spectral coefficients which
are independent of time i.e. they can give the amplitude-frequency information and donot
have any information about frequency with respect to time. Thus, it is useful only for a
stationary signal where the amplitude-frequency does not change with time. But, in real
life cases the signals are time dependent and also non-stationary. In such cases a scan
analysis using the Short Term Fourier Transform (STFT) is used but it has its limitations
like it can give information only about the amplitude and frequency, but not anything
about the time and frequency relation. The limitations of STFT are overcome by the
46
wavelet transform which gives a better idea about the time-frequency information about
the signals. The wavelet transform is a linear convolution of a given one dimensional
signal which is to be analysed and the mother wavelet (t). Mathematically a wavelet
transform is as shown below:
( ) dts
bttps
bsW ⎟⎠⎞
⎜⎝⎛ −
∫= *1),( ψ (8.1)
where, W(s, b) is the wavelet coefficient, the asterisk sign denotes the complex conjugate,
‘b’ is the translation parameter and ‘s’ is the scale parameter.There is a number of mother
wavelet which is used in practise but only some of the mother wavelets such as Mexican
hat wavelet, Gabor wavelet and Morlet wavelet are used in the field of fluid dynamics.
Wavelet has been used which is given by,
2/)/(4
12
)( γωωπψ tti oo eet −−= (8.2)
Where 2 2⁄ and is the number of wave in the wavelets. In practise the
value of varies from 5 to 12 and generally it is taken as 6. A frequency resolution of 12
is chosen when frequency of resolution of a signal is more important than time resolution.
(a) Plan (b) Isometric view
Fig. 8-1 Three sectors in GDWP
Sector 1
Sector 2
Sector3
Excitation direction
47
Table 8-1:Sectors frequencies in hertz computed by CFD simulations
Mode
number Sectors 1 and 5 Sectors 2,4, 6 and 8 Sectors 3 and 7
1 0.262 0.231 0.247
2 0.332 0.332 0.394
3 0.4738 0.460 0.447
4 0.532 0.542 0.548
Table 8-2: Numerical case studies
Case Excitation
Sectors
considered in
GDWP tank
Excitation
direction
Excitation
Amplitude
(m)
Excitation
Frequency
(Hz)
Condition
1 Sinusoidal
Excitation 3 xyz 0.03 First mode
Flexible
wall
2 Random
excitation 3 xyz 137m FRS data
Flexible
wall
(a) Longitudinal (x) (b) Vertical (y) (c) Lateral (z)
Fig.8-2FRS data for AHWR building at 137 m
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00 18.00 20.000.00
1.00
2.00
3.00
4.00
5.00
6.00
7.00
8.00
9.00
10.00
11.00
12.00
13.00
14.00
15.00
Floor Response Spectra at GDWP SLAB LEVELEL. 137.00 m (Node 19) in X-Direction
ζ=1% ζ=2% ζ=4%
S a /g
Frequency (Hz)
0.00 10.00 20.00 30.00 40.00 50.000.000.501.001.502.002.503.003.504.004.505.005.506.006.507.007.508.008.50
Floor Response Spectra at GDWP SLAB LEVELEL. 137.00 m (Node 19) in Y-Direction
ζ=1% ζ=2% ζ=4%
Sa /
g
Frequency (Hz)
0.00 5.00 10.00 15.00 20.00 25.00 30.00 35.000.00
2.00
4.00
6.00
8.00
10.00
12.00
14.00
Floor Response Spectra at GDWP SLAB LEVELEL. 137.00 m (Node 19) in Z - Direction
ζ=1% ζ=2% ζ=4%
S a /g
Frequency (Hz)
48
(a) Longitudinal (x) (b) Vertical (y) (c) Lateral (z)
Fig. 8-3 Acceleration- time history for GDWP of AHWR building at 137 m
8.4 GDWP under Sinusoidal Excitation in Multi-Direction (Case 1):
Table 8.1 shows the sectors frequencies in hertz computed by CFD simulations. The first
mode frequency is taken for sinusoidal excitation study. Table 8.2 shows the details of
case studies considered to understand the sloshing behaviour in GDWP under harmonic
and random excitations. First, GDWP is excited with harmonic loading having frequency
equal first mode natural frequency in three directions (x, y, z). Non-dimensional slosh
height is obtained and shown in Figs. 8.4 (a) through (c). The phase-plane diagram for
regular excitation is presented in Fig 8.4 (d). As it is the horizontal sinusoidal excitation
case, single frequency is observed with respect to time as depicted in Fig 8.4 (e) which is
obtained from wavelet analysis.
‐0.8
‐0.6
‐0.4
‐0.2
0
0.2
0.4
0.6
0.8
0 5 10 15 20 25
Acceleration (g)
Time (Sec)
‐0.6
‐0.4
‐0.2
0
0.2
0.4
0.6
0 5 10 15 20 25
Acceleration (g)
Time (sec)
‐0.8
‐0.6
‐0.4
‐0.2
0
0.2
0.4
0.6
0.8
0 5 10 15 20 25
Acceleration (g)
Time (Sec)
49
Fig. 8-4 Multi directional sinusoidal excitation (Case 1) (a) –(c) Non-dimensional slosh
height for sector 1 to 3 (d) Phase-plane diagram (e)Wavelet diagram for slosh height signal at right corner of sector 1
8.5GDWP under Random Excitation in Multi-Direction (Case 2)
Slosh height is computed in three directional excitation, viz., two horizontal and one
vertical directions. Non-dimensional slosh height for sector 1- 3 is captured and shown in
-2 0 2 4 6 8 10 12 14 16 18 20 22 24 26-12
-8
-4
0
4
8
12
(a)
Sector 1 Right sector 1 center
Non
-dim
ensi
onal
slo
sh h
eigh
t
Time(Sec)-2 0 2 4 6 8 10 12 14 16 18 20 22 24 26
-12
-8
-4
0
4
8
12
(d)(c)
(b) Sector 2 Right sector 2 center
Non
-dim
ensi
onal
slo
sh h
eigh
t
Time(Sec)
-2 0 2 4 6 8 10 12 14 16 18 20 22 24 26-12
-8
-4
0
4
8
12
Sector 3 Right sector 3 center
Non
-dim
ensi
onal
slo
sh h
eigh
t
Time(Sec)-15 -10 -5 0 5 10 15
-20
-15
-10
-5
0
5
10
15
20
dζ/d
t / (A
ω2 )
ζ/A
0 5 10 15 20 250
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Time (Sec)
Freq
uenc
y (H
z)
(e)
50
Fig. 8.5. It is found from Fig. 8.5,slosh height at sector 1 is more than other two sectors
(i.e., sectors 2 & 3). The FFT is computed from the steady state signal. FFT of slosh
height varying with time for sector 1 through 3 is shown in Figs. 8.6 to 8.8. Dominant
frequency is near to first mode frequency (i.e., 0.312 hz). During random excitation, the
frequency for the each sector is varies with respect to the length of the free surface. Fig 8.9
shows the phase plane diagram for sector1. The phase plane diagram is visualize the
presences of nonlinearity in the curve.
Fig. 8-5 Non-dimensional slosh height for sector 1 to 3 of the GDWP subjected to
design excitation as given in Fig. 8.3 (Case 2)
0 5 10 15 20 25 30 35 40-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5 Inner Right Inner Center
Time (Sec)
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
Inner Right Inner Center
Non-
dim
ensi
onal
Slo
sh h
eigh
t
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
Sector 2
Sector 3
Sector 1
Inner Right Inner Center
51
Fig. 8-6 FFT from slosh height at sector 1 of GDWPsubjected to design excitation as
given in Fig. 8.3 (Case 2)
Fig. 8-7 FFT from slosh height for sector 2 of GDWP subjected to design excitation as
given in Fig. 8.3 (Case 2)
Fig. 8-8 FFT from slosh height for sector 3
of GDWP subjected to design excitation as
given in Fig. 8.3 (Case 2)
Fig. 8-9 Phase-Plane diagram of sector 1
slosh height
The slosh height signals are analysed to get the frequency information along time using
wavelet analysis. Figs. 8.10 through 8.12 show the wavelet analysis in sector1, sector 2
and sector 3 respectively during random excitation. Due to this random excitation, the
sector 1 through 3 have the multi-frequencies data. These frequencies are mainly first few
sloshing frequencies of GDWP in three directions. Slosh height is found more at the tank
wall corners of direction excitation.
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.00.000
0.002
0.004
0.006
0.008
Frequency (Hz)
Ampl
itude
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.00.00
0.01
0.02
0.03
0.04
Frequency (Hz)
Ampl
itude
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.00.00
0.01
0.02
0.03
0.04
Frequency (Hz)
Ampl
itude
-3 -2 -1 0 1 2 3-1.0
-0.8
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
(dζ
/ d t)
/ (A
/ ω
1)
ζ / A
52
Fig. 8-10 Time –frequency curve for sector 1 of GDWP slosh height subjected to design excitation as given in Fig. 8.3 (Case 2)
Fig.8-11 Time –frequency curve for sector 2 of GDWP slosh heightsubjected to design
excitation as given in Fig. 8.3 (Case 2)
Fig.8-12 Time –frequency curve for sector 3 of GDWP slosh heightsubjected to design excitation as given in Fig. 8.3 (Case 2)
0 5 10 15 20 25 30 350
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Time (Sec)
Freq
uenc
y (H
z)
0 5 10 15 20 25 30 350
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Time (Sec)
Freq
uenc
y (H
z)
0 5 10 15 20 25 30 350
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Time (Sec)
Freq
uenc
y (H
z)
53
Fig. 8-13 Base shear at convective mode for sector 1 through 3
Base shear is the estimate of the maximum expected lateral force that will occur due to
seismic ground motion at the base of the structure. An accurate prediction of the base
shear and moment is essential to ensure the safety of the tanks against the shell bucking
and uplift. Base shear at sector 1, 2 and 3 are shown in Figs. 8. 13 (a) through (c). The sum
of convective and impulsive forces (pressure and shear force) on tank wall is shown in
Fig. 8.14. The average total force in longitudinal and lateral directions are 1.02 MN and
2.25 MN respectively. Pressure moment and viscous moment also calculated at the bottom
of the tank and depicted in Fig. 8.15. Maximum force and moment values for sector 1
through 3 are also shown in Table 8.3. Due to the liquid load during its oscillation, the
(a)
(b)
(c)
54
wall deflection is also varying with respect to time. Wall deflection due to liquid sloshing
for different time has been depicted in Fig. 8.11. Finally, Fig. 8.17 shows the liquid
pressure near wall at free surface and base. The dynamic pressure is computed as 2.7
KN/m2 and 7 KN/m2 respectively.
Fig. 8-14 Total force (convective + impulsive) at sector wall 1
Table 8-3: Total force and moment in sectors 1, 2 and 3
Condition Sector 1 Sector 2 Sector 3
Total force at GDWP in MN
(convective + impulsive)
Maximum 13.2 13.4 15.7
Minimum 7 6.69 4.2
Average 9.91 9.9 9.8
Total momentin bottom of
GDWPMN-m
(convective + impulsive)
Maximum 205.2 207.8 241
Minimum 65 64 54
Average 146 146 146
(a)
(b)
55
Fig. 8-15 Total moment (convective + impulsive) at sector wall 1
Fig.8-16Tank wall displacement versus time
0 5 10 15 20 250
1x107
2x107
3x107
4x107
5x107
6x107
7x107
8x107
My
Mom
ent a
t tan
k bo
ttom
due
to s
ecto
r1 fo
rce
Time(Sec)
‐4
‐2
0
2
4
6
8
10
0 0.0005 0.001 0.0015 0.002
Vertical len
gth in m
Inner wall displacement in m
0 Sec 0.05 Sec 0.25 sec
56
Fig. 8-17 Wall pressure (convective + impulsive) at sector 1 through 3
The convective frequency, slosh height, dynamic wall pressure near free surface, and
impulsive and convective (total) pressure on wall near base are compared between CFD
and ERM approaches for fully curved outer wall GDWP and partially curved outer wall
GDWP separately in chapter 10.
0 5 10 15 20 25 30 3530
40
50
60
70
80
Sector 1 Sector 2 Sector 3
Wal
l pre
ssur
e ne
ar b
ase
KN/m
2
Time (Sec)
0 5 10 15 20 25 30 350
2
4
6
8
(b)
(a)
Maximum wave amplitute is 17 KN/m2
Maximum wave amplitute is 2.7 KN/m2
Sector 1 Sector 2 Sector 3
Wal
l pre
ssur
e ne
ar fr
ee s
urfa
ce K
N/m
2
57
9 CONCLUSIONS
This work focused on the numerical investigation of liquid sloshing in GDWP of AHWR. The
sloshing behaviour in an un-baffled and baffled GDWP of AHWR are studied under sinusoidal
excitation. Effect of baffles in the annular water pool is studied for different excitation and
amplitude. If liquid oscillation is not controlled efficiently, sloshing of liquids in storage water
pools may lead to water spilled out from water pool or large dynamic stress to cause structural
failure. Hence, the study of sloshing and measures to suppress it are well justified with two types
of baffles for this kind of annular water pools. To estimate the slosh height and other design
parameters sloshing in GDWP is studied using simulated seismic load along the three orthogonal
directions.
From the above numerical investigation, the following observations are made.
Maximum slosh height under design seismic load is found around 0.62 m for fully curved
outer wall GDWP (existing model) and 0.72 m partially curved outer wall GDWP. It is
also observed that GDWP has sufficient free board.
X and Y directions of convective frequencies for GDWP are 0.265 hz and 0.315 hz for
fully curved wall and 0.231hz and 0.262hzfor partially curved wall of GDWP. It is also
noted that the convective frequencies are varies with respect to the direction of excitation.
Seismic design codes (through ERM approach) for tanks and CFD results are compared.
It is found that ERM has around 25% difference in slosh height from CFD results.
Cap-plate baffle is more effective than annular in reducing the sloshing oscillations.
9.1 General Conclusions from above Study
1) The liquid free surface elevation has been captured for different excitations and different
amplitudes. Since the liquid first mode frequency is very less than the structure first mode
58
frequency, the fluid-structure interaction effect is found to be negligible. As a result, the fluid
pressure on the free surface is dominated by the sloshing pressure or convective pressure and
the relative or fluid-structure interaction pressures are negligibly small for the case of first
mode excitation.
2) When the water pools are subjected to higher modes (i.e., higher than first mode excitation),
the fluid in the water pool will tend to undergo sloshing motions. In the higher modes the
liquid elevation is lower than the first mode frequency. At the beginning of the disturbance,
the fluid dynamic pressure is dominated by the impulsive pressure. After few seconds,
sloshing pressure or the convective pressure becomes the dominant component of pressure.
3) Experimental and analytical studies are also performed to validate present numerical results.
For experimental validation, a simple square and four-sectored squared tanks was taken. The
pressure variations were captured at different locations under the surge motions of the tank
and found the CFD results are good in agreement with the experiment.
4) It is found from the numerical investigation that the liquid will spill out around 0.06 m
excitation amplitude ( ≈0.023g acceleration) from the fully curved wall GDWP under first
mode sloshing frequency. However, design acceleration is 0.16g at 0.312Hz and corresponds
to 0.028 m equivalent harmonic amplitude. At this amplitude, liquid slosh height is found 0.7
m.
5) Further, the tank with different baffles was studied. As expected, annular baffle and cap-plate
baffle cases were reducing the liquid oscillations as well. However, cap-plate baffle was more
effective in reducing the sloshing oscillations for this kind of water pool geometry. This baffle
was reduced the liquid slosh height from 0.7 m to 0.3 m under design acceleration.
6) During random excitation, frequency for the each sector is varies with respect to the length of
the free surface. At the end of the seismic force, free vibration shows the natural frequency of
the tank. So that, maximum slosh height was found at the end of the seismic force. It is also
observed that more slosh height found at tank wall corners in the direction of excitation.
59
9.2Some more Observations between ERMand CFD
GDWP is an annular 8-sectored tank as depicted in Fig. 1.2. The outer wall of GDWP is
spherical in shape while the inner one is in cylindrical shape. Since seismic codes are not
available for these kind of sectored geometries, here CFD and ERM approaches are used to
estimate the sloshing behaviour in GDWP. For the effective comparison, two GDWP models
have been chosen as presented in Figs. 4.1 and 4.2. They are as follows,
(i) Fully curved outer wall GDWP (existing model)
(ii) Partially curved outer wall GDWP (proposed model).
The partially curved wall model is proposed for thermal hydraulic issues (in AHWR review
meeting 201). Slosh height and dynamic pressure is obtained through CFD for both GDWP
models. CFD results are compared with corresponding ERM as shown in Tables 9.1 and 9.2,
while making an equivalent rectangle the following assumptions are considered.
(i) Liquid volume should be equal in GDWP sector and rectangular model.
(ii) Ratio between liquid height (h) and length of the tank (L) is taken almost equal in
both cases.
60
Fig.9-1Comparison between CFD and ERM for fully curved outer wall GDWP.
Fig.9-2Comparison between CFD and ERM for partially curved outer wall GDWP.
0.258 0.265
0.7780.856
0.312 0.315
0.6150.692
00.10.20.30.40.50.60.70.80.9
x y x y
Convective frequency in Hz Slosh height in m
Fully curved outer wall GDWP
ERM CFD
0.227 0.239
0.893 0.953
0.262 0.27
0.721 0.782
0
0.2
0.4
0.6
0.8
1
1.2
x y x y
Convective frequency in Hz Slosh height in m
Partially curved outer wall GDWP
ERM CFD
61
Tabl
e 9-
1:X
and
Y d
irec
tion
val
ues
of s
ecto
r 1
of f
ully
cur
ved
wal
l GD
WP.
D
irec
tion
Equ
ival
ent
rect
angl
e m
etho
d
(ER
M)
CF
D
Dif
fere
nce
in %
& C
omm
ents
|(CF
D-E
RM
)|/C
FD
Geo
met
ry
(Equ
ival
ent r
ecta
ngle
vs
full
y cu
rved
wal
l)
Slo
pe w
all w
ill s
uppr
ess
the
slos
hing
.
Liq
uid
volu
me
in m
3
1012
10
12
Con
vect
ive
freq
uenc
y in
Hz
x 0.
258
0.31
2 17
.3
y 0.
265
0.31
5 15
.8
Slo
sh h
eigh
t in
m
x 0.
778
0.61
5 26
.5
y 0.
856
0.69
2 19
.1
Dyn
amic
wal
l pre
ssur
e
near
fre
e su
rfac
e in
KN
/m2
(Con
vect
ive)
x 6.
11
4.2
45.4
y 6.
52
4.25
53
.4
Tot
al p
ress
ure
on w
all i
n
KN
/m2
(Im
puls
ive
+ C
onve
ctiv
e)
x 63
.89+
6.11
=70
.0
58.2
20
.2
y 10
6.15
+6.
52
=16
6.67
54
.8
204
8 m
62
Tabl
e 9-
2: X
and
Y d
irec
tion
val
ues
of s
ecto
r 1
of p
arti
ally
cur
ved
wal
l GD
WP
.
D
irec
tion
Equ
ival
ent
rect
angl
e m
etho
d
(ER
M)
CF
D
Dif
fere
nce
in %
& C
omm
ents
|(CF
D-E
RM
)|/C
FD
Geo
met
ry
(Equ
ival
ent r
ecta
ngle
vs
part
ially
cur
ved
wal
l)
Slo
pe r
egio
n is
loca
ted
near
the
free
sur
face
. It
wil
l sup
pres
s th
e
slos
hing
.
Liq
uid
volu
me
in m
3
1012
10
12
Con
vect
ive
freq
uenc
y in
Hz
x 0.
227
0.26
2 13
.3
y 0.
239
0.27
11
.4
Slo
sh h
eigh
t in
m
x 0.
893
0.72
1 23
.8
y 0.
953
0.78
2 18
.36
Dyn
amic
wal
l pre
ssur
e
near
fre
e su
rfac
e in
KN
/m2
(Con
vect
ive)
x 3.
32
2.7
5.6
y 3.
1 2.
5 51
.9
Tot
al p
ress
ure
on w
all i
n
KN
/m2
(Im
puls
ive
+ C
onve
ctiv
e)
x 54
.33+
7.17
=61
.5
74
16.8
y 10
4.29
+7.
41
=11
1.7
69
61.8
6m
4 m
6m
63
From the Fig. 9.1 and 9.2 and Tables 9.1 and 9.2 show the difference between the
ERM and CFD slosh analysis results for fully curved outer wall GDWP and partially
curved outer wall GDWP respectively. For comparison between ERM and CFD, the
convective frequency, slosh height, pressure are taken. While comparing the fully and
partially curved wall GDWP cases, the partially curved wall has very close to the CFD
results. Since length of the free surface is become very less in the case of fully curved
outer wall, the rectangular assumption leads slightly more error in these slosh results.
However, the partially curved wall has close relation between ERM and CFD except the
pressure in y direction.
Acknowledgement
Authors express their sincere gratitude and graceful acknowledgement to our colleagues
Shri. Ravi Kiran, Shri. PN Dubey, Dr. Parulekar and Mr. Piyanshu Goyal, (Scientific
Officers, Reactor Safety Division) and we sincerely thankDr. RK Singh (AD, RDDG &
Head, RSD) forhis time, careful work and valuable comments and suggestions on the
report.
64
REFERENCES
[1] ACI 350.3, (2001), “Seismic design of liquid containing concrete structures”, American
Concrete Institute, Farmington Hill, MI, USA.
[2] Bathe KJ. (1996), Finite element procedures. Prentice Hall.
[3] Benra, F.K., Dohmen, H.J., Pei, J., Schuster, S., and Wan, B. (2011), “A comparison of one-
way and two-way coupling methods for numerical analysis of fluid-structure interactions”,
Journal of Applied Mathematics, Article ID 853560.
[4] Biswal, K.C., Bhattacharyya, S.K., Sinha, P.K. (2006), “Non-linear sloshing in partially
liquid filled containers with baffles”, International Journal for Numerical Methods in
Engineering, 68, 317–337.
[5] Chang, Y.W., Gvildys, J., Ma, D.C., Singer, R., Rodwell, E., Sakurai, A. (1989), “Numerical
simulation of seismic sloshing of LMR reactors”, Nuclear Engineering and Design, 113,
435-454.
[6] Chen, B.F., Nokes, R. (2005), “Time-independent finite difference analysis of 2D and
nonlinear viscous liquid sloshing in a rectangular tank”, Journal of Computational Physics,
209, 47–81.
[7] Chen, S., Johnson, D.B., Raad P.E., Fadda, D. (1997), “The surface marker and micro cell
method”, International Journal for Numerical Methods in Fluids, 25, 749-778.
[8] Chopre AK, (2001), “Dynamics of structures -Therory and applications to earthquake
engineering”, third reprint, Prentice-Hall of India Private Limited, NewDelhi.
[9] ESI CFD Inc., CFD-ACE+, V2011.0, Modules Manual, Part 2, ESI-Group.
65
[10] ESI US R&D, CFD-ACE (U)® V2011, User’s Manual. Huntsville (AL, USA): ESI-CFD
Inc.; 2011. (Web site: www.esi-cfd.com).
[11] Eswaran M., and Saha UK. (2011), “Sloshing of liquids in partially filled tanks – A review
of experimental investigations”, International Journal of Ocean Systems Engineering, 1(2),
131-155.
[12] Eswaran, M., Saha, U.K. (2009), “Low steeping waves simulation in a vertical excited
container using sigma transformation”, ASME Proceedings of the 28th International
Conference on Ocean, Offshore and Arctic Engineering (OMAE 2009, pp number:
OMAE2009-80248, Hawaii, USA.
[13] Eswaran, M., Saha, U.K. (2010), “Wave simulation in an excited cylindrical tank using
sigma transformation”, ASME Proceedings of the International Mechanical Engineering
Congress & Exposition, Vancouver, BC, Canada, pp number IMECE 2010-39752.
[14] Eswaran, M., Singh, A., Saha, U.K. (2010), “Experimental measurement of the surface
velocity field in an externally induced sloshing tank”, Journal Proceedings of the Institution
of Mechanical Engineers, Part M, Journal of Engineering for the Maritime Environment,
225(2), 133-148.
[15] Eswaran, M., Saha, U.K., Maity, D. (2009), “Effect of baffles on a partially filled cubic
tank: Numerical simulation and experimental validation”, Computers and Structures, 87(3-
4), 198-205.
[16] Faltinsen, O.M., Rognebakke, O.F., Lukovsky, I.A., Timokha, A.N. (2000),
“Multidimensional modal analysis of nonlinear sloshing in a rectangular tank with finite
water depth”, Journal of Fluid Mechanics, 407, 201–234.
66
[17] Frandsen, J.B. (2004), “Sloshing motions in excited tanks”. Journal of Computational
Physics 196, 53–87.
[18] Gedikli, A., Erguven, M.E. (1999), “Seismic analysis of a liquid storage tank with a baffle”,
Journal of Sound and Vibration, 223(1), 141–155.
[19] Glatzel, T., Litterst, C., Cupelli, C., Lindemann, T., Moosmann, C., Niekrawietz, R., Streule,
W., Zengerle, R., Koltay, P. (2008), “Computational fluid dynamics (CFD) software tools
for microfluidic applications – A case study”, Computers & Fluids 37, 218–235.
[20] Goudarzia M.A and Sabbagh-Yazdi S.R, (2012), “Investigation of nonlinear sloshing effects
in seismically excited tanks”, Soil Dynamics and Earthquake Engineering, 43, 355–365.
[21] Gupta, A., Eswaran, V., Munshi, P., Maheshwari, N.K., Vijayan, P.K. (2009), “Thermal
stratification studies in a side heated water pool for advanced heavy water reactor
applications”, Heat Mass Transfer 45, 275–285.
[22] Haroun, M.A., Housner, G.W. (1981), “Seismic design of liquid storage tanks”, ASCE, 107,
191-207.
[23] Hirt, C.W., Nichols, B.D. (1981), “Volume of fluid (VOF) method for the dynamics of free
boundaries”, Journal of Computational Physics 39(1), 201–25.
[24] Housner GW (1963) Dynamic analysis of fluids in containers subjected to acceleration.
Nuclear reactors and earthquakes, Report no. TID 7024, US Atomic Energy Commision,
Washington D.C.,USA.
[25] Ibrahim, R.A. (2005), “Liquid sloshing dynamics: Theory & applications”, Cambridge
University Press, New York.
67
[26] Kandasamy T., Rakheja S., Ahmed, A.K.W. (2010), “An analysis of baffles designs for
limiting fluid slosh in partly filled tank trucks”, The Open Transportation Journal 4, 23-32.
[27] Kolukula SS and Chellapandi P, (2013), “Nonlinear Finite Element Analysis of Sloshing”,
Advances in Numerical Analysis, Vol. 2013, Article ID 571528.
[28] Lee, K., Lee K.H., Lee J.I., Jeong Y.H., Lee P.S. (2013), “A new design concept for
offshore nuclear power plants with enhanced safety features”, Nuclear Engineering and
Design, 254, 129-141.
[29] Malhotra, P.K. (1997), “Seismic response of soil supported unanchored liquid storage
tanks”, Journal of Structural Engineering, 123(4), 440-450.
[30] Malhotra, PK, (2004), “ Seismic analysis of FM approved suction tanks”, Draft copy, FM
Global, USA
[31] Meeting on sloshing issues of GDWP with Director RD&DG on 24/May/2013.
[32] NZS 3106, (1986), “Code of practise for concrete structures for the storage of liquids”,
standards association of New Zealand, Wellington.
[33] Shimada S, Noda S and Yoshida T., (1988), Nonlinear sloshing analysis of liquid storage
tanks subjected to releatively long-period motions”, Proceedings of ninth world conference
on earthquake engineering, Vol-VI, Augest 2-9, Japan.
[34] Simqke, (1976), “Simulated earthquake motions compatible with prescribed response
spectra”, Report No 2, Department of civil Engineering, Massachusetts Institute of
Technology.
[35] Sinha, R.K., Kakodkar A. (2006), “Design and development of the AHWR- the Indian
thorium fuelled innovative nuclear reactor”, Nuclear Engineering and Design 236, 683-700.
68
[36] Tedesco, J.W., Landis, D.W., Kostem, C.N. (1989), “Seismic Analysis of cylindrical liquid
storage tanks”, Computers and Structures, 32: 5, 1165–1174.
[37] Vijayan, P.K. (2010), “Reducing thermal stratification and boiling in pools with immersed
heat exchangers”, 3rd Meeting, 4-5 November 2010, IAEA Headquarters, Vienna, Austria
(www.iaea.org/INPRO/CPs/AWCR/3rd_Meeting/Indian-3-1.pdf ).
[38] Vorobyev, A., Kriventsev, V., Maschek, W. (2011), “Simulation of central sloshing
experiments with smoothed particle hydrodynamics (SPH) method”, Nuclear Engineering
and Design, 241, 3086– 3096.
[39] Xue, M.A., Zheng, J., Lin, P. (2012), “Numerical simulation of sloshing phenomena in cubic
tank with multiple baffles”, Journal of Applied Mathematics, Article ID 245702.
69
APPENDIX- A
NON-LINEARITY EFFECT DUE TO WAVE BREAKING
The nonlinear nature of sloshing is the greatest hindrance in solving such a problem analytically
and even computationally. Sloshing flows are sturdily nonlinear in most cases. Wave breaking,
impact, and splashes can be observed in violent sloshing flows. In general, the nonlinearity of
sloshing flow becomes dominant at any of the following situations.
(i) Tank under angular excitations (ex. Ship cargo)
(ii) Tank with wave breaking device/mechanism. (TLD, passive baffle)
(iii) Tank with lower filling conditions (Shallow depth)
(iv) External unusual loads (Thermal loads, Suppression of high velocity steam)
The slosh-induced impact pressure is affected by wave breaking phenomena, therefore the
observation and understanding of physical phenomena on these flows are important to develop
the proper models. This phenomena dissipate part of the slosh energy induced by earthquakes. It
is used as vibration control mechanism in Tuned Liquid Damper (TLD). The fundamental
frequency of the fluid in the TLD should be close to the natural frequency of the structure if the
TLD is to dissipate energy efficiently. The study on dissipation induced by a free surface flow is
arduous, especially in the presence of a wave breaking flow. The local wave breaking is not
likely to disturb the global motion of sloshing flow (i.e., the local splashes and wave breaking
may be ignored when global fluid motion is concerned).
To study the wave braking, a dominant nonlinearity condition needs to be made. As
discussed in above paragraph, all the situations can be omitted except second one for seismic
studies (i.e. Tank with wave breaking device/mechanism). Here, three case studies are shown to
demonstrate the nonlinearity effects as indicated below. Simple rectangular tank is taken for this
study. Tank dimensions are 10.8 m length and water fill height is 8.4 m. Water is used as tank
fluid. For, simulation, 0.1 m is used as an excitation amplitude. Since the non-linearity need to be
captured from simulation, around 40,000 elements have been created in the computational
domain. VOF technique is used to track the interface between liquids.
70
Fig. A-1 Slosh height elevation computed from linear and nonlinear equations
(a) case1 (b) case2 (c) case3
Fig. A-2 Snap shot of maximum slosh height at left end (a- linear violent flow; b- non-linear normal flow at 0.5 ; c- Non-linear violent
flow at 0.99 )
(a) case1 (b) case2 (c) case3
Fig.A-3 Snap shot of maximum slosh height at right end (a- linear violent flow; b- non-linear normal flow at 0.5 ; c- Non-linear violent
flow at 0.99 )
‐3
‐2
‐1
0
1
2
3
0 5 10 15 20 25 30
slos
h he
igh
in m
Time in Sec
Linear violent flow (case 1) Non-linear violent flow (case 3)
Non-linear normal flow (case 2)
To see t
(i) Cas
freq
(ii) Cas
freq
(iii) Cas
freq
(iv) Cas
Fig. A-
flows. T
the snap
A-1. Ta
Tank w
Fig. A5
baffle s
Fig.A-
the slosh be
se 1 – Linea
quency)
se 2 – Non-
quency)
se 3- Non-li
quency)
se 4 -Tank w
-1 shows th
There is no
p shots of li
ank with Wa
with passive
5 respective
ize is 1 m le
-4 Tank with
ehaviour du
ar equations
-linear equa
inear equati
with wave b
hat slosh h
much non-
inear and no
ave Breakin
baffle com
ely. To capt
ength and 0
h wave brea
e to linear a
s and high e
ations and lo
ions and hig
breaking de
height captu
linearity is
on-linear eq
ng Device (P
mputational d
ture the no
0.08 m thick
aking devic
71
and nonline
excitation fr
ow excitatio
gh excitatio
evice (Passiv
ured from l
found in th
quations slos
Passive baff
domain and
n-linearity
k. Baffles ar
e (Case 3)
ar equation
requency (A
on frequency
n frequency
ve baffle)
linear and
his simple st
sh profile.
fle)
d grid arran
a very fine
e fixed at 1
Fig.A-5 F
s are solved
Around 100
y (Around 5
y (Around 1
non-linear
tructures. Fi
ngement are
e mesh is cr
.6 m from li
Fine mesh a
d with four c
% of natura
50 % of natu
100 % of na
equation fr
igs. A-2 and
shown in F
reated near
iquid free su
arrangemen
cases.
al
ural
atural
rom violent
d A-3 show
Fig. A4 and
r walls. The
urface.
nt (Case 3)
t
w
d
e
72
The passive baffles are used to dissipate the sloshing motion energy by breaking a main sloshing
flow into several weaker sub-streams to induce the non-linearity. Fig.A-6 shows the snapshots of
non-linear free surface profiles. This wave breaking causes due to the presence of the passive
baffles.
Fig. A -6Snap shot of Non-linear violent flow at 0.99 with baffle arrangement (case 4)
A-2 Discussion andConclusions
Sloshing flows are sturdily non-linear in most cases. Wave breaking, impact, and splashes can be
observed in violent sloshing flows. However, non-linearity will occur due to any one or
combined, situations such as tank under angular excitations, tank with wave breaking
device/mechanism, tank with lower filling conditions, external unusual loads, etc. The local
wave breaking is not likely to disturb the global motion of sloshing flow. So that, the local
splashes and wave breaking may be ignored when global fluid motion is concerned. During the
non-linear violent slosh flow at 0.99 , some local wave breaking is observed, but however, it
is insignificant, compare to global motions of liquid. It is also observed that the slosh height from
linear and non-linear formulations is around 2% difference near trough and crest of the wave.
Tank with baffles case, a high non-linear slosh flow is found near corner of the tank. Baffle
dissipates sloshing energy by breaking a main sloshing flow into several weaker sub-streams.