137132221-centrifugal-pumps-pdf.pdf

7/16/2019 137132221-Centrifugal-Pumps-pdf.pdf

http://slidepdf.com/reader/full/137132221-centrifugal-pumps-pdfpdf 1/82

f=

Centrifugal Pumps:Overview of Design, Operation and

Malfunctions

By

D. Craig Sever

And

Charles T. Hatch

Bently Nevada Training Development Group

Bently Nevada Corporation



f=

Copyright (C) 1999 Bently Nevada Corporation. All rights reserved.

The information contained in this document is subject to change without notice.

The following are trademarks of Bently Nevada Corporation in the United States

and other countries:

Actionable Information, Actionable Information to the Right People at the Right

Time, ADRE, Bently Align, Bently Balance, Bently Nevada, CableLoc,ClickLoc, Data Manager, Decision Support, DemoNet, Dynamic Data

Manager, Dynamic Transmitor, Engineer Assist, FieldMonitor, FluidLoc,

flexiTIM, flexiTAM, Helping You Protect and Manage All Your Machinery,

HydroVU, Key ∅∅∅∅, Keyphasor, Machine Condition Manager 2000,

MachineLibrary, Machine Manager, MicroPROX, Move Data, Not People,

Move Information, Not Data, NSV, Preformalign, Process Centered

Maintenance, PROXPAC, Proximitor, REBAM, Seismoprobe,

ServoFluid, Smart Monitor, Snapshot, System 1, System Extenders,

TDXnet, TDIXconnX, Tecknowledgy, TipLoc, TorXimitor, Transient

Data Manager, Trendmaster, TrimLoc, VAM, Velomitor, XLerometer

The Bently Nevada Corporation Orbit Design, Bently Balance and Design, System 1

Enabled and Design, and M2 and Design are all trademarks or registered marks of

Bently Nevada Corporation in the United States and other countries.



f=

Table of Contents

1. INTRODUCTION TO CENTRIFUGAL PUMPS.................................................................................1

THE R OLE OF PUMPS AND THE CONSEQUENCES OF PUMP MALFUNCTION ...................................................1

WHAT IS A PUMP? .......................................................................................................................................2

TYPES OF PUMPS..........................................................................................................................................3

2. DESIGN AND OPERATION OF CENTRIFUGAL PUMPS...............................................................6

TERMINOLOGY OF E NERGY IN LIQUIDS ....................................................................................................... 6

THREE FUNDAMENTAL WAYS CENTRIFUGAL PUMPS ADD E NERGY............................................................ 7

PUMP COMPONENTS AND THEIR PURPOSES .................................................................................................7

Impeller...................................................................................................................................................8

Casing................................................................................................................................................... 13

Multiple Stages .....................................................................................................................................16

Inlet Geometry ......................................................................................................................................16

Seals......................................................................................................................................................17

Sealless Pump Designs .........................................................................................................................19

Wear Rings ...........................................................................................................................................20

Shaft Sleeves......................................................................................................................................... 21Thrust Balancing...................................................................................................................................22

Bearings ................................................................................................................................................ 23

Couplings.............................................................................................................................................. 24

PERFORMANCE, OPERATION, AND TERMINOLOGY ....................................................................................25

Pump Performance Curves: Important Pump Parameters .......... ........... .......... ........... .......... ........... ...... 25

System Curves ...................................................................................................................................... 26

Pump Operation: How Pump and System Curves Relate ..................................................................... 27

The Best Efficiency Point ..................................................................................................................... 27

Specific Speed....................................................................................................................................... 28

Net Positive Suction Head and Suction Specific Speed..... ............ ........... ........... ........... ............ .......... 29

3. MALFUNCTIONS OF CENTRIFUGAL PUMPS..............................................................................32

GENERAL CONCEPTS ................................................................................................................................. 32PUMP MALFUNCTIONS............................................................................................................................... 33

High 1X Vibration due to Unbalance....................................................................................................33

Radial Loads (Misalignment and Sideload)..........................................................................................38

Rub........................................................................................................................................................ 47

Shaft Crack ........................................................................................................................................... 53

Fluid-Induced Instability.......................................................................................................................62

Structural Resonances...........................................................................................................................70

Cavitation.............................................................................................................................................. 72

Vane Pass Frequencies.......................................................................................................................... 77

4. REFERENCES........................................................................................................................................ 78



f=

1

1. INTRODUCTION TO CENTRIFUGAL PUMPS

THE R OLE OF PUMPS AND THE CONSEQUENCES OF PUMP MALFUNCTION

Vast numbers of processes require liquid to move from one location to another.These processes can be seen in nuclear and non-nuclear power generation, oil pipelines, petrochemical refineries, municipal wastewater and domestic water treatment facilities,

both large and small buildings, on ships and offshore oil platforms, and manufacturing

plants, and the list could go on. In virtually all of these processes, pumps play theessential role of providing the propulsion necessary to move the liquid.

Pumps are a generally robust and reliable class of rotating machinery. However,

pumps are critical machines in many processes because their loss can create serious or

even catastrophic results. Power generation relies on boiler feedpumps, condensate pumps, and water circulation pumps to circulate water through the thermodynamic

process that converts fuel into electrical power. Nuclear power generation would be

impossible without the variety of pumps to circulate water through the primary reactor core loop, secondary power generating loop, and cooling water loop. Power generation

pumps are typically large and custom, one-of-a-kind design. The failure of a powergen

pump can result in significant financial loss due to pump damage, as well as damage toassociated equipment. For example, a large high-pressure boiler evaporating about a

million pounds of water per hour could suffer extensive damage within minutes if

allowed to run dry due to a failed boiler feedpump.Process industries such as petrochemical refineries are also vulnerable to similar

financial consequences. The processing of liquid product employs large numbers of

pumps. A failed pump can shut down an entire process resulting in revenue losses on the

order of tens or even hundreds of thousands of dollars a day. In order to avoid such

losses, many process industries find it necessary to devote large portions of their maintenance budgets to pumps.

Safety is an even greater concern than the financial impact of pump malfunctions.Public and plant personnel can be seriously endangered by accidents stemming from

pump failures in processes that handle radioactive or toxic liquids. For instance,

operating conditions can affect the reliability of pump seals. If a malfunction causesvibration, temperature, or pressure to change radically or to move outside of normal

operating ranges, these seals may leak and expose plant workers and the surrounding

public to the adverse effects of hazardous liquids.

Environmental damage due to pump failure can also be a very serious problem.Hazardous materials released into the environment through leaking pump seals can have

significant environmental impacts. The consequences of such unintended releases are notlimited to environmental damage, there may be heavy financial costs as well.Environmental regulations governing hazardous materials have become very stringent

and environmental regulatory agencies may require the filing of a report, impose large

fines, shut a plant down, or all of these, depending on the nature and amount of liquidreleased.

All of these factors combine to make pumps a class of rotating machinery that

deserve in-depth examination.



f=

2

WHAT IS A PUMP?

Almost everyone is familiar with pumps and their basic function. We readily

recognize that the water pump in an automobile circulates engine coolant through the

radiator and water jacket. However, it is helpful to establish the function of pumps inmore precise terms. Pumps can be compared to the engine of a car. It is well known that

an automobile engine accelerates the mass of the car against the effects of inertia,overcomes frictional resistance caused by air, tires, etc., and overcomes the gravitationalresistance of moving a car up a hill (i.e., elevation changes). Driving against a strong

headwind or up a steep grade gives one a special appreciation for these effects. Likewise,

liquid in a pipe must be accelerated against the inertia of its mass and once accelerated toa desired velocity (or flow rate), energy must be added on a continual basis to keep the

liquid flowing against frictional resistance and elevation changes. In actual practice, the

inertia of flowing liquids are largely ignored because it is of less concern than the other two forms of resistance.

The idea of pumping against an elevation change is not hard to imagine. As with a

car, it is simply the resistance encountered when moving a liquid uphill against earth’s

gravitational pull. However, not all may be as familiar with the concept of frictionalresistance to liquid flow. Just as friction occurs between two blocks of wood that are

rubbed together, friction also occurs between individual molecules of liquid that “rub”

together while flowing down a pipe. The molecules of liquid rub because they are not allmoving at the same velocity. Liquid molecules immediately adjacent to a pipe surface

have zero velocity while molecules in the center of the pipe have maximum velocity.

This can be seen by observing flow in a river where the flow is slowest at the edges and becomes swifter toward the center. It follows then that there must be a variation in speed,

or gradient , between the molecules closest to the stationary pipe surface and those in the

center of the flow. This gradient means that adjacent liquid molecules have slightly

different speeds causing them to rub against each other and produce friction. This

friction combined with gravity creates significant resistance that a pump must overcomeif a liquid is to flow.

A closer examination of the concept of friction in liquids allows us to recognize thatthe magnitude of pipe friction loss depends on several factors. Rougher pipe surfaces

create more drag on a liquid than smoother surfaces and hence more friction. Smaller

diameter pipes have less cross-sectional flow area than larger pipes which yields greater resistance to flow. In addition, certain properties of the liquid itself are contributing

factors. Emptying a can of motor oil versus a glass of water illustrates how the higher

viscosity oil molecules cling to each other more than the water molecules. This “cling” is

largely due to the cohesion between molecules. The greater the cohesion, the greater theamount of energy required to make a liquid flow. This translates directly into greater

frictional resistance for higher viscosity liquids. The actual flow of liquids in pipes isquite complex and these are just a few of the factors that affect frictional resistance.However, this simple explanation gives us sufficient understanding of the task that pumps

must perform. Just as a car engine provides energy to keep the car moving against

friction and gravity, so too a pump provides energy to keep a liquid moving againstelevation and frictional resistance.



f=

3

Energy added

by Pump to

overcome

Friction and

Elevation

Energy lost

to Friction

Energy lost

to Friction

Graph of Energy Change in Liquid Flowing

through Piping System and Pump

Pump supplies an abrupt

increase in Energy

Energy Grade Line

Figure 1.1 Energy Grade Line (EGL) shows how the energy of a liquid

changes as it flows through a piping system, pump, and change in elevation.

Energy losses are represented by a decreasing EGL while energy gains are

represented by an increasing EGL. Energy is required to make a liquid flow

against the effects of frictional and gravitional resistance, the purpose of a

pump is to provide the energy necessary to overcome these resistances.

The relationship

between energy lost to

flow resistance andthe energy gained

from a pump can be

shown graphically in adiagram called an

energy gradeline

diagram (Figure 1.1).Of course, pumps

do not accomplish the

task of supplying

energy by themselves.Pumps are actually

energy converters;

they take rotative shaft

energy from a driver and convert it to

increased energy inthe pumped liquid.

The goal is to pump as

efficiently and cost

effectively as possible,using the minimal

amount of mechanical

energy per unit of energy added.

TYPES OF PUMPS

Pumps fall into two broad categories depending on how they add energy to the pumped liquid. The first category is known as displacement pumps and these utilize

plungers, pistons, diaphragms, screws, gears, or other similar means, to exert a force

directly on the liquid. Except for screw and gear type pumps, displacement pumps use acyclical process that imparts the energy in pulses. The second category is referred to as

kinetic pumps because they add energy by passing the liquid through an impeller which

“speeds up” the liquid thereby increasing its kinetic energy. In other words, kinetic

pumps do not push on the liquid quite as directly as displacement pumps. They addenergy using a different principle that will be explained in more detail in following

sections. In contrast to displacement pumps, kinetic pumps add energy in a smoother and

more continuous process.Kinetic pumps are sometimes referred to as centrifugal pumps. Most of the pumps

used in power generation and process industries are centrifugal pumps. While all

centrifugal pumps use an impeller to add kinetic energy, there are many different designsdepending on the specific application. Centrifugal pumps may be single-stage (one



f=

4

impeller) or multi-stage (more than one impeller) and may rotate about a horizontal or

vertical axis. Single-stage centrifugal pumps may have their impeller over-hung

(supported at one end only) or have their impeller supported on both ends between bearings. In addition to the impeller, the design of the pump case also varies widely

depending on application. These are just a few of the design differences among

centrifugal pumps.The pumps shown in Figure 1.2 and 1.3 illustrate a typical single-stage end-suction

volute pump. It is only one of many variations among centrifugal pump designs, albeit a

common one. Figure 1.2 illustrates most of the components that are common to all(radial and mixed flow) centrifugal pumps.

Casing

Impeller Coupling

Suction

Discharge

Bearings (2)

Shaft

Mean

Flow Line

Wear Rings (4)

Seal Lubrication

Port

Shaft

Sleeve

Sealing Area

Suction

Eye

Figure 1.2 Cross-section of a typical end-suction centrifugal pump with single-suction, over-hung

impeller. Section on left is taken through pump shaft, section on right is taken through impeller and volute

along mean flow line.

Figure 1.3 Centrifugal pump used to pump water. Suction enters from left in both photos, discharges

through top. Whereas the pump illustrated in Figure 2 is supported by its own mounting feet, the pump

shown above mounts directly to the driver (electric motor) housing with no additional support.



f=

5

This paper focuses on centrifugal pumps because they are the most widely used pump

design in the world. In addition, centrifugal pumps represent a significant portion of the

capital, operating and maintenance costs of the power generation and process industries.This paper will discuss the root causes underlying some of the more common

centrifugal pump malfunctions, how to recognize their characteristic symptoms and how

to correct them. In order to understand pump malfunctions, it is first necessary tounderstand how they are designed and operated.



f=

6

2. DESIGN AND OPERATION OF CENTRIFUGAL PUMPS

TERMINOLOGY OF ENERGY IN LIQUIDS

We have established that the fundamental purpose of pumps is to add energy to liquidso that it can flow against the effects of frictional and gravitational resistance. Beforecontinuing with the explanation of how pumps accomplish this task, it is first necessary

to understand the terms used to commonly describe this type of energy. Those who work

with pumps refer to the energy added by a pump as head (H). Head is measured in unitsof feet or meters. Head can take on three forms with each form being measured by a

different means. The first form is static pressure head, or simply pressure head ( H P ).

Pressure head is the energy measured with a pressure gauge. The second form is

elevation head ( H E ). Elevation head is the potential energy that a liquid has by virtue of its relative vertical position in a system. Thus, the higher a liquid is, the greater its

elevation head. The third form is velocity head ( H V ). Velocity head is the kinetic energy

of a liquid due to its velocity. Velocity head is commonly measured with a pitot tube.The total energy in a liquid consists of the sum of these three forms of energy.

The total energy, or head, of a liquid can be distributed in any proportion among the

three forms. The total energy may exist completely in one form to the exclusion of theother two, or it may exist as 30% pressure head, 30% elevation head, and 40% velocity

head, or it may exist in any other combination as long as the sum of the three forms

equals 100% of the total head. For example, the water at the bottom of a swimming poolwill have no elevation head (compared to water at the pool’s surface) and will have no

velocity head (assuming there is no circulation in the pool). However, it will have energy

in the form of pressure head and this is exactly the pressure felt on one’s ears while

diving to the bottom of a deep swimming pool. Conversely, water at the top of the pool

will have potential energy because of its elevation head but it will have no pressure head.The lack of pressure head is sensed by the absence of pressure on one’s ears immediately

below the water surface. Water situated at levels between the top and bottom of the poolwill have some combination of pressure and elevation head depending on depth.

We can convert the pressure head to velocity head by opening an imaginary valve

located at the bottom of the pool. Water will flow through the valve and we canintuitively understand that higher pressure in the pool will correlate directly with a higher

velocity through the valve.

If we calculate the total head in the high elevation water at the top of the pool, and the

total head in the high pressure water at the bottom of the pool, and the total head in thehigh velocity water flowing out of the valve it will be the same in all three cases. The

fact that the total head converts among its three forms without increasing or decreasing(assuming no energy losses or gains by external means) is known as Bernoulli’s law.Bernoulli’s law is an expression of the fundamental principle of the conservation of

energy.

We must understand that head can exist in one of three forms and that it can

convert between them because centrifugal pumps operate by first adding velocity

head and then converting some portion of it into pressure head.



f=

7

THREE FUNDAMENTAL WAYS CENTRIFUGAL PUMPS ADD ENERGY

All centrifugal pumps use an impeller to add velocity head to a liquid. However, not

all impellers accomplish this in exactly the same manner. Depending on the method

used, impeller designs are grouped into three general types. The difference between themis the direction in which each type forces the high velocity liquid to flow.

1. Radial flow impellers increase liquid velocity in a direction perpendicular (or radial) to the pump axis (Figure 2.1a).

2. Mixed flow impellers increase liquid velocity in a direction that is a mixture of

perpendicular and parallel flow with respect to the pump axis (Figure 2.1b).

3. Axial flow impellers increase liquid velocity in a direction parallel (or axial) tothe pump axis (Figure 2.1c).

Each type of impeller provides a certain combination of performance features.

Hence, each type is best suited to meet the needs of particular applications.

PUMP COMPONENTS AND THEIR PURPOSES

Centrifugal pump designs range from small and simple to large and intricate.

However, no matter how complex or simple the overall machine, there are parts common

to all designs that provide the same function.

The following is a discussion of these common pump components. It is intended thatthis will provide a context for the latter discussion of pump malfunctions.

Impeller

High Velocity

Out

Low

Velocity

In

Figure 2.1a Radial Flow Impeller

directs flow radially outward from

pump axis.

Low

Velocity

In

Impeller

High Velocity

Out

Low

Velocity

In

Figure 2.1b Mixed Flow Impeller

directs flow both radially and

axially to pump axis.

High Velocity

Out

Low

Velocity

In

Impeller

Figure 2.1c Axial Flow Impeller

directs flow axially along the pump

axis.



f=

8

Impellers

The impeller, or more specifically, the impeller vanes, are that part of the pumpwhere the rotative shaft energy from the driver is converted into kinetic energy in the

pumped liquid. Radial and axial flow impellers perform this conversion using different

mechanisms while mixed flow impellers combine the two methods.

Radial Flow Impellers

A radial flow impeller is essentially a rotating disk with several evenly spaced radialvanes protruding on one side (Figure 2.2). Liquid is guided into the “eye” or center of

the impeller via the suction passage of the pump casing where it is then caught by the

leading edges of the vanes. (The vanes are usually curved backward against the direction

of rotation, this will be explained shortly.) Once caught by the vanes, centrifugal forcedrives the liquid toward the periphery of the impeller (hence the name centrifugal pump).

The liquid accelerates as it travels outward. One way to understand the change in

velocity is to think of it as having 2 vector components. One component, U , is equal to

radial distance (r ) times angular velocity (ω ). The other component, V T , is the velocity of

the flow tangential to the vanes and is related to the velocity of liquid flowing through the

vane passages (Figure 2.3).

Rotation

Figure 2.2 Flow of liquid through radial flow impeller. Pumped liquid enters center of impeller

where it is caught by vanes and driven outward by centrifugal force. Total velocity of the liquid

increases as it moves further out toward the periphery of the impeller.



f=

9

The velocity at inlet and outlet is determined by summing the respective U and V T

components (vector components are summed graphically by placing the tail of one to the

head of the other). A visual comparison of the length of the total velocity vectors at inletand outlet shows that total velocity V is greatest at the periphery of the impeller. The

increase in U from inlet to outlet accounts for the gain in total velocity V.

From this description, we see that the net increase in velocity head is the difference between V 2 and V 1. V 1 cannot be counted as energy gained because it was already

present in the liquid prior to entering the impeller [8].

While this description represents an overly simplified and highly idealized approach,

it is a useful model for understanding how radial flow impellers boost velocity head andit provides a basis from which actual pump performance can be calculated. There are

energy losses (for example, fluid and mechanical friction) which cause the actual

performance to be less than ideal. Manufacturers include these losses when estimatingthe actual performance. Even with the best estimation techniques, the actual performance

must always be determined by testing.

V 1

U 2 V T2

V 2

V T1

+

r 2

r 1ω ωω ω

= r 2 ω ωω ω

U 1 = r 1ω ωω ω

Figure 2.3 Vector Components (U and V T ) of Total Velocity (V ) at Impeller Inlet

and Outlet. The vector components sum to the make up the total velocity. The

growth in U from inlet to outlet explains the difference between V 2 and V 1. The

backward curve of the vanes determines to what degree an increase in V T2 reduces V 2.

Since the direction of V T2 is tangent to the vane, the net effect of increasing backward

vane curve is an overall reduction in total velocity, V 2, as V T2 increases. Note also

that U 2 depends on impeller rotative speed , ω , and impeller diameter, r 2. An

increase in either variable results in increasing total velocity, V 2.



f=

10

Axial Flow ImpellersAxial flow impellers are usually

included in the general category of centrifugal pumps. However, they do not

use centrifugal force to increase velocity

head. Rather, axial flow impellers operateon the same principles as propellers (Figure

2.4). Axial flow vanes are shaped to push

the liquid in the direction of the pump axis,unlike radial flow pumps that move the

liquid at right angles to the shaft axis.

Mixed Flow Impellers

The third type of centrifugal pump impeller is really a combination of the two just

described. As Figure 2.1b shows, liquid is accelerated radially and axially. Mixed flowimpellers combine goemetric features of both radial and axial flow impellers.

The basic shape of each of the three types of impellers provides a differentcombination of head versus flow characteristics. Radial flow pumps are used in high

(pressure) head, low flow applications while axial flow pumps are used where low

(pressure) head, high flow is needed. Mixed flow pumps occupy a range in between.These head versus flow characteristics are described by a parameter known as a

pump’s specific speed . This meaning of this term is explained in the section below

entitled “Specific Speed”.

Impeller Design Features:

1. Vane Curvature

The vanes of most radial and mixed flow impellers are curved backward (i.e.,opposite the direction of impeller rotation) as shown in Figures 2.2 and 2.3. A few radial

flow impellers are produced with straight vanes.

The backward vane curvature is partly responsible for the decrease in head as flowthrough the pump increases. (The main contributor to this head versus flow characteristicis frictional resistance in the pump flow passages, backward vane curvature serves to

accentuate this effect.) Decreasing head versus increasing flow is desirable from the

standpoint of pump operation as explained in more detail below in the section entitled“Pump Operation: How Pump and System Curves Relate”.

In order to understand how backward vane curvature produces this effect, we need to

refer to the velocity triangles in Figure 2.3. There we see that the total velocity head, V ,at inlet and outlet can each be thought of as the sum of two individual vector

components, U and V T . The value of U depends on impeller radius and on angular

velocity (i.e., shaft rotative speed) and remains constant when these variables are heldconstant. The value of V T rises and falls as a function of the amount of flow through the

impeller (this is no different than, for example, the change in velocity of water through a

garden hose as more or less water flows through it.) In addition, the direction of V T is

tangent to the vane surface. Since V T flows tangentially to the vanes which are directed

Figure 2.4 Axial Flow Impeller.



f=

11

mostly opposite to rotation at the point where they contact the impeller periphery,

increases in V T act to cancel U . If U is held constant, then the net effect of an increase

in V T is a reduction in the total velocity head, V .One can see that the reduction in total velocity head with increasing flow depends

directly on the degree of backward vane curvature. More backward curvature results in

greater decrease in total velocity head with increase in flow.2. Impeller Diameter

The outside diameter of the impeller is another important design parameter that

determines the amount of velocity head that an impeller can develop. Knowing that theone of the two vector components which sum into the total velocity head is U and that it

is directly dependent on outlet radius and angular velocity (or shaft rotative speed), we

realize that increasing or decreasing outlet radius will have a like effect on the total

velocity head, V (Figure 2.3).Increasing the head that a pump can produce by increasing impeller diameter is not

without a cost. Any increase in diameter will also require an increase in the torque

required to turn the impeller. Since power is a direct function of torque and shaft rotative

speed, a larger diameter impeller will require more power for a fixed speed.Minor reductions in impeller diameter (called “trimming ”) are sometimes made in

order to reduce the pump discharge head or to reduce the pump power consumption. For example, the discharge piping could burst if it does not have the strength to withstand the

pressure head created by an oversize pump, or maybe the piping can handle the extra

pressure but the excessive power consumption of an oversize pump is uneconomical.

From the previous discussion of vane curvature and impeller diameter, we know that this problem could be dealt with by modifying any one of several variables that determine the

head a pump will produce. Flow could be increased (which raises V T thereby lowering

V ), shaft rotative speed (or angular velocity, ω ) could be reduced (which lowers U

thereby lowering V ), or vane curvature could be increased (which causes V T to further

cancel U thereby lowering V ), or impeller diameter could be reduced (which lowers U thereby lowering V ). Except for reducing impeller diameter, these parameters are not so

easily changed in most applications. Generally, the process determines the flow and thedriver is often a single speed electric motor. Furthermore, the vanes are an integral part

of the impeller and are not modifiable unless the impeller is changed. This leaves

reducing the impeller diameter as the most cost-effective solution for situations of thiskind.

Pump casings are designed to accommodate a range of impeller diameters which

allows impellers to be removed, trimmed, and reinstalled in the same pump casing. The

pump affinity laws predict the effect that trimming will have on pump performance.

While it is beyond the scope of this paper to explain these laws, they deserve mentioning

because they are so widely used for predicting how impeller trim will affect head versus

flow characteristics.

3. Open versus Closed

This categorization indicates whether the impeller vanes are enclosed by shrouds on

the front, back, or front and back of the vanes. These shrouds are an integral part of theimpeller and thus rotate with it. The shroud of the impeller shown in Figure 2.2 is

represented by the flat disk which supports the vanes.



f=

12

The shrouds serve to keep the liquid flowing through the vane passages in the proper

direction. The pumped liquid is naturely driven back to suction by the high discharge

pressure. Enclosing the vane passages with shrouds helps to keep the liquid flowing inthe intended direction instead of leaking back to suction through open areas between the

vanes and the sidewall of the casing. This leakage represents wasted pumping energy

and a reduction in efficiency.The impeller shown in Figure 2.2 is actually a semi-open impeller because it is

enclosed on one side by a shroud. Strictly speaking, an open impeller is simply a hub

with vanes attached to it [1]. Closed impellers have shrouds on both the front and back sides (Figure 2.14). Closed impellers provide the greatest reduction in leakage and

therefore are more efficient than open impellers. However, they are also more expensive

to manufacture. In addition, a tight clearance running joint must be provided between the

impeller shrouds and the pump casing. This joint is usually lined with replaceable wear rings which also add cost to the pump (see section entitled “Wear Rings” below).

4. Single versus Double Suction

Impellers can be designed with suction inlets, or “eyes”, on one or both sides. Single

and double suction impellers are shown in Figures 2.14 and 2.15 respectively. Doublesuction impellers have several advantages over single suction impellers. First, their

design provides a better balance of the axial forces that occur when pumps are operatedoff of design capacity (see section entitled “Thrust Balancing”). Second, double suction

impellers have a larger suction area than single suction pumps for a given flow which

means that less energy is required to push flow into the suction. In other words, double

suction pumps have a lower Net Positive Suction Head Required (NPSHR) than singlesuction pumps (NPSHR is explained below in “Net Positive Suction Head”).

Single suction impellers are usually preferred because they are less expensive. Single

suction pumps are easier to manufacture and less likely to clog when handling suspendedmatter such as sewage [1].



f=

13

Casing

FunctionIt is apparent that the liquid and impeller must be contained in some kind of vessel

which directs the flow toward the discharge. However, there is more to the case design

than simply catching and containing the high energy liquid as it comes off the impeller.The pump case has another equally important role – it must convert some of the velocity

head into pressure head.

As liquid leaves the impeller periphery its velocity head is very high – in fact it is toohigh and the pressure head is too low for many applications. Some of that velocity head

must be converted into pressure head in order to be useful.

The conversion of velocity head into pressure head occurs in the pump case. Figure

2.5 shows the how velocity headdecreases while pressure head

increases as the flow moves

through the discharge side of the

case. The conversion processfollows the principle of

conservation of energy as stated by Bernoulli’s law. Since total

amount of energy must remain

constant (assuming not losses or gains), pressure head must

increase as velocity head is

reduced.

The way to reduce velocity is by increasing the cross-sectional

area of the flow through the process of diffusion. Simply put,diffusion occurs when flow area

is expanded. The expansion

causes a reduction in velocityand an accompanying increase in

pressure. There are two common

case designs which accomplish

this in an efficient manner (“Efficient” in this instance means no energy losses through turbulence. The energy in

turbulent liquid flow is non-recoverable).

Casing Design:1. Volutes

The most common type of casing design is the single volute casing. The volute

casing is so called because of its spiral shape (Figure 2.6). The volute provides theexpanding flow passage necessary for diffusion to occur.

The smallest point of flow area where the volute begins is called the cutwater . The

cutwater divides the liquid coming off the impeller into two flows with one side flowing

through the volute and the other side flowing toward the discharge.

Pressure

Velocity

Casing

Suction

Impeller Casing

Discharge

Flow Path

Outlet Tips

of Impeller

VanesInlet Tips

of Impeller

Vanes

Figure 2.5 Velocity vs. Pressure Head of Flow Through Pump.

Graph shows the relationship between velocity head and pressure

head of flow through the pump. Casing discharge is designed to

convert velocity head into pressure head while preserving the

total amount of head.



f=

14

Figure 2.7a Balanced radialforces in pump operated at

design capacity. Balanced

radial forces produce a

minimal net force on the

impeller.

F

Figure 2.7b Unbalanced

radial forces in pump

operated below or above

design capacity. Unbalanced

radial forces result in a net

force, F. Direction and

magnitude of F varies with

flow.

The clearance between the cutwater and

the periphery (Figure 2.6) of the impeller is a

critical design dimension because it must besized to strike a compromise between

efficiency and pressure pulsations. Pump

efficiency increases as the clearance betweencutwater and impeller is reduced. However,

if the clearance is too small, then large

pressure pulsations resulting in pump failurecan occur. This is explained in more detail

below in the section entitled “Vane Pass

Frequency”.

Volutes have an additional drawback which has been the cause for many broken

pump shafts and failed seals and bearings.

The pressure of the liquid in the volute acts

on the projected area of the impeller to produce a radial force. Since the pressure actsaround the full circumference of the impeller, it actually produces many radial forces that

act in all directions upon theimpeller (Figure 2.7a). These

radial forces are generally

balanced when the pump is

operated at its Best Efficiency Point (BEP). The sum total of

these nearly balanced forces is a

net radial force that is minimal or,in some cases, practically

nonexistent. However, when

operated above or below BEP, theforces become unbalanced which

can result in a significant net

radial force, F (Figure 2.7b). The

direction and magnitude of thisnet radial force will vary

depending on operating point relative to design capacity. The net radial force can be as

much as fifteen times the force at design capacity (Figure 2.10).(The term Best Efficiency Point (BEP) is described below in the section “Best

Efficiency Point”. However, as the name implies, the pump is producing the maximum

output per input when operated at the BEP.)Despite this drawback, the single volute is still the most commonly used case design.

Part of the reason for this is that it costs less to manufacture than other designs [2].

However, the problems presented by excessive radial forces that result from operating off of design capacity have spawned the following case designs.

Single Volute

Direction of

RotationCutwater

Cutwater-

to-Impeller

Clearance

Figure 2.6 Single Volute Pump Casing. The

spiral shape of the volute can be seen in the figure.

The cutwater divides the flow coming off the

impeller. The clearance between cutwater and

impeller must be optimized for best efficiency and

lowest pressure pulsations.



f=

15

Double Volute

Direction of

Rotation

Figure 2.8 Double Volute Pump Casings contain

an additional volute positioned 180° to volute of

casing sidewall. The double volute balances the

sideloads produced when the pump is operated off

of design capacity.

Radial Load

Double

Volute

Single

Volute

Vaned

Diffuser

Flow

Best Efficiency Point (BEP)

+

+

Figure 2.10 Radial (sideload) force versus flow for

three types of pump casings. Radial force is minimum

when flow is at the Best Efficiency Point (BEP).

Vaned diffuser produces least amount of radial force

when the pump is operated off of BEP.

Vaned Diffuser

Direction of

Rotation

Impeller

Casing

(Concentric

shown)

Figure 2.9 Vaned Diffuser. Note that the number of

impeller and diffuser are not the same. The unequal

number of vanes minimizes pressure pulsations which

would be magnified if the same number of each were

used.

2. Double Volutes

Double volute casings have two opposing

volutes positioned 180° opposite of each other

(Figure 2.8). This feature makes the doublevolute more effective than the single volute at

minimizing the radial loads produced byoperation away from the BEP (Figure 2.10).

3. Vaned Diffusers

Another type of casing design which balances

hydraulically produced sideloads is the vaned diffuser . The vaned diffuser contains several

vanes set around the periphery of the impeller

(Figure 2.9). Essentially, each vane acts as aminiature volute. The distribution of many

evenly spaced “small volutes” makes the vaned

diffuser the most effective of the three casing

designs at minimizing sideloads (Figure 2.10).The vaned diffuser may be contained in either a

concentric or volute shaped casing.

Pump designers pay careful attention to the number of impeller and diffuser vanes.

Equal numbers of vanes and certain other combinations can lead to destructive high

vibration which occurs at the “vane pass frequency” (see section “Vane PassFrequency”). Designers follow guidelines to avoid these combinations of vane numbers.



f=

16

Multiple Stages

Multistage centrifugal pumps are, as the name implies, pumps which contain morethan one impeller (each stage represents an impeller and casing volute). Centrifugal

pumps with as many as fifteen or so stages are not unusual (even twenty stage pumps

have been built). The stages are connected in series so that the discharge from one stageflows into the suction of the stage immediately downstream. Each stage increases the

head by a certain amount with the total head added by the pump being the summation of

all the stages.The reason for designing multistage pumps lies in the fact that efficiency suffers if

too big an increase in head is attempted with a single stage. Thus, in applications which

require high head, increasing head in smaller incremental steps using multiple stages

preserves efficiency.Common applications that require high head multistage pumps are boiler feedwater

pumps, reactor feedpumps, and pipeline booster pumps. Vertical pumps can also be

multistage.

These are a few notable particulars about the design of multistage centrifugal pumps:• cases may be split axially or radially.

• impellers are generally single suction. However, double suction impellers aresometimes used in the first stage since they reduce the required NPSH of a pump.

• the angular orientation of impeller vanes and volutes are offset, or staggered, between stages. The staggered orientation results in a more balanced angular

distribution of the radial sideloads produced at the different stages. In addition,

this helps to reduce pressure pulsations and vibration which occur at vane passfrequency.

Inlet Geometry

The inlet geometry refers to the flow passage in the pump casing from the pointwhere the inlet piping attaches at the suction flange to the point where the liquid contacts

the impeller (Figure 1.2). It is a general term that includes all casing geometry from the

pump’s suction flange right up to the eye of the impeller. (The inlet of a pump is alsoreferred to as the suction.)

The inlet geometry of a centrifugal pump is worth discussing for the reason that is a

major determinant of a pump’s required Net Positive Suction Head (NPSH). NPSH is the positive pressure required at the pump inlet (i.e., suction flange) to prevent the pumped

liquid from vaporizing into bubbles, or cavities, within the low pressure regions of the

impeller vane passages. NPSH is a very important parameter because centrifugal pumpswill cavitate if the available NPSH falls below the required NPSH. NPSH and cavitation

are explained in much greater detail below in their respective sections, however, a brief description is as follows. Cavitation occurs when the pressure (i.e., the Available NPSH)

of the incoming liquid falls below the vapor pressure of the liquid. Cavitation is anundesirable condition because the vapor bubbles can cause serious damage to the

impeller when they collapse. The minimum pressure required to prevent cavitation is

termed the Net Positive Suction Head Required (NPSHR) because it denotes the pressureat the suction flange that is required to prevent the liquid downstream within the flow

passages of the impeller from flashing into vapor bubbles (i.e., cavitating). Since



f=

17

pressure in the impeller must stay above the liquid’s vapor pressure to avoid cavitating, it

follows that the pressure measured upstream at the suction flange must also be kept above

some minimum level. This difference in pressure is caused by the frictional losses thatoccur as the liquid flows through the inlet passage between these two points.

Therefore, a well designed pump inlet will minimize these frictional losses so that

less NPSH is required. A pump that requires less NPSH is advantageous because this pressure must be provided by the inlet piping system and this can result in greater overall

expense of the installation. Good inlet design practices can include large suction side

diameter, smooth flow passages, and other features which reduce frictional losses. The

suction specific speed is a number that provides a way to compare the effectiveness of a

pump at reducing the NPSH required to prevent cavitation (see “Suction Specific

Speed”).

Two commonly used suction configurations are the end-suction (Figure 1.2) and in-line suction designs. It is beyond the scope of this paper to describe these designs,

however, the reader should be aware of their existence and their importance. The

references listed at the end of this paper provide greater detail on these as well as other

types of inlet designs [1,2].

Seals

Seals are a particularly important pump component because they are probably the

most frequent cause of regular pump maintenance and thus the cause of a high percentage

of overall pump maintenance cost [2]. They are also frequently one of the first parts to beaffected by a malfunction. Seals are critical items because the leakage of a hazardous or

toxic liquid can have severe safety and economic consequences.

Other than the sealless designs described in the next section, all centrifugal pumps

require a seal where the shaft penetrates the case. The seal must prevent the high pressure liquid contained in the case from leaking through the joint where the rotating

shaft (or shaft sleeve) and stationary components are in contact. There are two generalapproaches to centrifugal pump shaft sealing: packing and mechanical seals. Eachapproach will be explained separately with a brief comparison to follow.

Packing

Packing is the oldest and one of the most common shaft sealing systems for centrifugal pumps [2]. The main components are shown in Figure 2.11.

The term stuffing box refers to the general part of the pump that houses the packing

assembly. It is the point where the shaft penetrates the casing. The stuffing box can be

either an integral part of the casing or it can be a bolt-on assembly. The stuffing boxcontains the stuffing box throat which is an annular space surrounding the shaft (or shaft

sleeve if there is one). Square cross-section packing rings of a pliable material (usually a

fibrous or metallic substance) are “packed” into the stuffing box throat. This creates theseal between the shaft (or shaft sleeve) and casing. These rings must be held tightly in

place and this is done with the packing gland . The packing gland is a collar that is

compressed against the rings, typically by studs with nuts that can be tightened or loosened to provide compression as needed.

The packing gland is situated on the atmospheric side of the gland assembly so that it

can be accessed for tightening or loosening. On the opposite, or high pressure end, of the

packing rings is the stuffing box bushing . The stuffing box bushing provides a seat



f=

18

against which the rings

can be compressed. The

annular clearance between this bushing and

shaft (sleeve) is tight to

prevent the packing fromextruding into the pump

and loosing compression.

This tight clearance alsolimits the flow of liquid

that can leak in case the

packing fails completely.

An effective stuffing box seal involves a fairly

hard radial “squeeze”, or

compression, of the non-

rotating packing ringsdown onto the rotating

shaft (or sleeve). Thiscreates a hard contact

between the packing

rings and shaft that must

be kept lubricated.Without this lubrication,

the packing ring material

will burn and the shaft sleeve will wear excessively. Thus, a means of lubricating the packing must be provided.

This lubrication is typically accomplished by routing a small amount of pumped fluid

into the packing rings via a lantern ring (Figure 2.11). The lantern ring has spaces thatallow the lubricating fluid to flow circumferentially and seep into the joint between the

packing rings and shaft (sleeve) in order to provide the necessary lubrication. Sometimes

it is necessary to use other lubricating fluids if the pumped liquid has poor lubricity or

contains abrasives. The leakage of lubrication can be significant, on the order of severaldrops per minute.

Lubrication of the packing is not the only design feature necessitated by the hard

radial compression. High wear of the rotating shaft surface is inevitable even withlubrication. To avoid periodic replacement of expensive pump shafts, almost all packed

pumps use less expensive, replaceable shaft sleeves made of a hardened or hard-coated

material (see “Shaft Sleeves” below).

Mechanical SealsMechanical seals have been developed to address the shortcomings of stuffing box

and packing gland assemblies. The main components are shown in Figure 2.12.Mechanical seal designs are quite varied but all based on the same general concept.

They do not attempt to seal directly against the rotating shaft as does a packing assembly.

Rather, a mechanical seal moves the joint off the shaft and places it between a pair of

sealing faces, one which rotates with the shaft and the other which is stationary with the

Packing

Gland

Packing Rings (5)

Lantern RingStuffing Box

Shaft Sleeve

Stuffing BoxBushing

Shaft

Lubrication

Figure 2.11 Stuffing Box and Packing Assembly. Packing gland

compresses the packing rings against the stuffing box bushing and shaft

sleeve to seal in the high pressure liquid on left side of assembly. Lantern

ring receives pressurized lubrication via the threaded passage above.

Lubrication flows circumferentially through the lantern ring to lubricate

the packing rings-to-shaft sleeve contact area.



f=

19

case. The mechanical seal faces

are oriented perpendicularly to

the shaft axis and held in contact by one or more springs. Thus,

mechanical seal designs have

been able to eliminate the radialcompression required by stuffing

box and packing assemblies.

While the amount of leakagethrough a mechanical seal is

generally less than through

packing, some is still required

for lubrication. The need for lubrication exists because the

rotating-to-stationary seal faces

would quickly be destroyed if

allowed to run dry.Mechanical seal designs

employ various means of lubrication. The lubricating

fluid can be gas or liquid.

There are other variations in

mechanical seal designs. Theseare described in more depth in

the references listed at the end of this paper [1,2].

Mechanical Seals versus PackingMechanical seals generally leak less than packing, boost mechanical efficiency due to

lower friction losses, require less maintenance than packing (when properly selected,

installed, and operated), and can handle higher pressures. Thus, mechanical seals arewell suited for applications where leakage of pumped liquid through packing would

create safety, environmental, or production problems, such as with toxic or radioactive

liquids.

On the other hand, mechanical seals have a few disadvantages when compared to packing. When they fail, they usually do so much more quickly and catastrophically.

Also, their initial cost is generally higher and they are less tolerant of axial shaft

movement.

Sealless Pump Designs

There are some applications where even the low leakage of mechanical seals isunacceptable. The number of such applications has grown as environmental and safety

regulations have become increasingly stringent. The demand for zero-leakage pumps has

given rise to sealless designs. Two types of sealless pump designs will be mentionedhere [2].

Magnetic drive pumps and canned motor pumps eliminate the shaft-through-case

penetration and its associated seal. This is accomplished by enclosing the rotating parts

in a cylindrical containment shell that tightly wraps the rotor. In magnetic drive pumps,

End

Plate

Spring

Fixed Collar

(rotates with

shaft)

Seal Face

(rotating)Seal Face

(fixed)

Figure 2.12 Simplified illustration which shows the general

elements comprising mechanical seals. Dynamic seal occurs

between the rotating and fixed seal faces. Spring maintains

contact pressure between the faces. Variations in design due to

type of lubricating fluid and contacting/non-contacting are not

shown.



f=

20

an outer magnet external to the shell is rotated by a separate driver motor. The flux from

the outer magnet then passes through the containment shell and turns an inner magnet

attached to the impeller. In canned motor pumps, the motor rotor and pump impeller share a common cavity with the rotor and impeller inside the containment shell and the

stator outside of it. Similar to the magnetic drive pump, there is no need for mechanical

seals because only the flux of the motor penetrates the containment shell.One notable aspect of both types of designs is that they may contain the

configuration by which the malfunction of fluid-induced instability (a.k.a., whirl and

whip) can occur. Briefly, any design configuration that rotates an inner cylinder at adifferent speed inside an outer cylinder with a fluid trapped between them in a close

clearance is susceptible to fluid-induced instability (see “Fluid-Induced Instability”).

Sealless pumps may have this design configuration in possibly two locations. First,

both types of pump have a rotor turning in close proximity to a stationary outer containment shell with the process fluid in between them. Secondly, both designs

typically use radial fluid-film (sleeve) bearings that are lubricated by the process fluid.

As explained below in “Fluid-Induced Instability”, this configuration is one of the

conditions necessary for whirl or whip to occur. Whirl and whip can cause highamplitude vibration that is very destructive.

In general, sealless pumps are not known for vibration stemming from fluid-inducedinstability. However, the potential exists and it has been known to occur.

Wear Rings

If we step back and consider that the basic function of a pump is to raise pressure, we

realize that it is much higher on

the discharge side of the impeller

than it is on the suction side.The pressure difference between

suction and discharge acts todrive the liquid back towardsuction, i.e., in the wrong

direction. The liquid can not

flow backwards through theimpeller vane passages because

centrifugal force drives it in the

proper direction (that is, unless

recirculation is occurring – recirculation is explained below

in “Cavitation”). However, the

suction-to-discharge pressuredifference will cause the liquid

to leak back through any other

available paths. If the impeller is closed, the liquid can leak

back to suction in the space

between the impeller shroud and

the pump casing. If the impeller

Impeller

Case

Wear Ring

(Case Mounted)

Wear Ring

(Impeller Mounted)

Retaining Screws

Leading Edge

of Impeller Vane

Double Flat Rings

Figure 2.13 Wear Rings. Replaceable wear rings are mounted to

case and impeller at the close-clearance gap called the leakage

oint. Clearance should large enough to prevent contact between

rings and small enough to minimize leakage from discharge back

to suction. The “Double Flat Ring” style is shown. Several other

wear ring configurations are also available.



f=

21

is open, liquid will leak back to suction over the vane tops. Such leakage means

decreased pump efficiency because the work previously done to move the leakage from

suction to discharge has to repeated. This work represents wasted energy.In order to prevent this leakage, the clearances between certain points of the impeller

and the case are made as tight as possible. For open impellers, this means that the vane

tops should run close to the case without touching. With closed impellers, clearances between the casing-mounted and impeller-mounted wear rings are kept to a minimum

(Figure 2.13). (Less frequently, the tight impeller-to-case clearance is located at the

periphery of the impeller on the discharge side.) For closed impellers, these clearancesvary from about 0.30 mm (0.012 inches) to about 0.76 mm (0.030 inches) depending on

overall impeller diameter [1].

The tight clearance between these rotating and stationary parts can present a

drawback. Though not intended to rub, these surfaces may contact and wear away whichwill open up the clearance resulting in lowered efficiency. In addition to wear from

surface contact, a corrosive or abrasive liquid can also wear away these surfaces with the

same detrimental effect. To overcome this problem, one or both of the wear surfaces is

usually fitted with a renewable ring – called a wear ring . The name is a bit misleading because under good pump operating practices these rings should never contact.

However, their design allows them to be replaced should wear occur.Shown in Figure 2.13 is a double flat-ring, one of just several wear ring designs.

Other wear ring types are single flat-rings, L-type rings, and labyrinth type rings.

Shaft Sleeves

Centrifugal pump shafts are usually fitted with a sleeve which protects the shaft from

wear at stuffing boxes or where it is in contact with corrosive and abrasive liquids (Figure

1.2). These sleeves are renewable parts meant to be replaced during pump overhaul.They deserve mention because they are a common part of many centrifugal pump

designs.



f=

22

Discharge

PressureDischarge

PressureSuction

Pressure

Single-Suction (Closed) Impeller

Axial

Thrust

Force

Wear

Ring

Figure 2.14 Axial Thrust Force of Single-Suction Closed Impeller. The difference between

suction and discharge pressures produces a net force that acts in the axial direction. The wear ring

is located at the leakage joint between the impeller and case. This narrow joint separates the

suction and discharge pressures.

Thrust Balancing

Centrifugal pumps experience axial thrust because of the difference between thesuction and discharge pressures acting on the cross-sectional area of the impeller. Figure

2.14 shows the pressure distribution surrounding a single-suction, closed impeller. Only

discharge pressure exerts a force on the discharge side of the impeller. However, thesuction side of the impeller has combination of high discharge pressure and low suction

pressure acting on it. The combination of pressures acting on the suction side of the

impeller are lower than the discharge pressure which acts over the entire discharge side of the impeller. The unequal pressure on the two sides results in a net axial thrust force.

The axial thrust force is not a constant. Since it is a by-product of the difference

between suction and discharge pressure, it will vary as this difference varies. Different

operating points (see “Pump Operation” below) will produce changes in this difference.Also, axial thrust will vary with impeller diameter. A single-suction impeller that is

trimmed (trimming is a common practice of machining down the outside diameter of an

impeller so that it will produce less head and consume less power) will have different

suction-to-discharge pressure than an untrimmed impeller. Thus, trimming can alsoaffect axial thrust. Consequently, centrifugal pumps must be designed to handle an axial

thrust force that varies with operating conditions.

Double suction pumps contain opposing impeller vane passages that theoreticallyshould cause axial forces to cancel. Figure 2.15 shows how the impeller is designed to

provide a balanced distribution of suction and discharge pressures. However, other

circumstances often disrupt this balance resulting in axial thrust [1], such as:

• Unequal flow into the two suction eyes. Can be caused by elbows (elbows are bends in the piping) located too close to pump suction.

• Unequal leakage through the two leakage joints. Can be due to uneven wear atthe wear rings.



f=

23

• Unequal discharge pressure between the two sides of the discharge. Can becaused by asymmetrical waterways or an impeller located off-center.

Since axial thrust is a certainty in most centrifugal pump designs and a possibility inothers, all centrifugal pumps incorporate thrust bearings. In addition to thrust bearings,

other balancing devices may also be incorporated. Single-suction pumps may have

balance holes through the impeller that allows suction pressure to leak to the dischargeside. An additional wear ring on the back side of the impeller prevents excessive suction-

to-discharge leakage. The wear ring on the back side of the impeller is situated so that

suction pressure on that side can balance suction pressure on the front side. Multistage pumps may use balancing drums (a.k.a., balancing pistons) and/or balancing disks.These additional balancing devices serve to relieve the thrust bearings of much of the

axial thrust present in the pump. The reader who desires greater detail on the design and

operation of these balancing devices is referred to the references.

Bearings

All rotating machines, including centrifugal pumps, require bearings to support and position the rotor axially and radially. These bearings must maintain relatively constant

rotor position under loads that fluctuate.

The most common types of bearings used in centrifugal pumps are either fluid-film

bearings or rolling element bearings.Fluid-film bearings.

Most fluid-film bearings are oil lubricated. The simplicity and load-carrying capacity

of plain, cylindrical bearings (also referred to as sleeve bearings) make it the mostcommonly used type of fluid-film bearing. However, since sleeve bearings sometimes

experience the malfunction of fluid-induced instability (whirl and whip), other designs

are also used. Figure 3.33 in the section “Fluid-Induced Instability” shows some of thesedesigns and describes how they reduce the potential for fluid-instability.

Discharge

Pressure

Suction

Pressure

Double-Suction Impeller

Suction

Pressure

Discharge

Pressure

Wear

Rings

Figure 2.15 Zero Net Axial Thrust Force of Double-Suction Impeller. Suction and

discharge pressures are balanced. Net axial thrust force is minimized.



f=

24

Some pump designs rely on the pumped liquid for lubrication. The magnetic drive

and canned motor pumps described above in “Sealless Pump Designs” are examples of

pumps that fall into this latter category. Their bearings are completely contained withinthe shell that separates the rotating from stationary parts and so they take advantage of

the pumped liquid in which their rotating parts are immersed. The applications that

require sealless pumps usually do so because of the harmful nature of the liquid to be pumped. The liquid in these applications is often corrosive or abrasive and thus quite

hard on bearings. Filtering screens and hardened bearings are a few of the methods that

manufacturers have used to counteract the detrimental effects of using pumped liquid for lubrication. [2]

Rolling element bearings.

Rolling element bearings are very commonly used in smaller centrifugal pumps.

They include ball, roller, and tapered rolling element bearings. Generally, ball bearingshave the greatest application in centrifugal pumps because they are capable of carrying

both radial and axial loads.

Rolling element bearings are sometimes referred to as “antifriction” bearings. The

term is true in the general sense, however, some friction is still present under normaloperating conditions. For this reason, all rolling element bearings incorporate some type

of lubrication. Lubrication may be grease, oil, and in certain designs, water.

Couplings

Centrifugal pumps require torque from a driver in order to move the pumped liquidagainst system resistance. This torque is transmitted from driver shaft to pump shaft

through a coupling . Since it is impossible to perfectly align driver and pump shafts,

couplings must transmit torque while allowing for variation in alignment within a

specified tolerance.Many pump applications use flexible couplings since they are designed to tolerate

small variations in alignment (excessive misalignment can be a serious pump malfunctionas explained below in “Radial Loads”).

Solid couplings are normally used only where the pump has no bearings and the

motor must support the shaft. Vertical pumps are the primary example of this type of

application. Solid couplings require extremely precise alignment. Not only must theinitial alignment be very precise but it must also remain very stable under operation.

A type of commonly used centrifugal pump that eliminates couplings is the close-

coupled pump. Close-coupled pumps have the pump housing mounted directly onto the

motor housing via close-tolerance fits. This allows the pump and motor to share the sameshaft. Because the shaft is one solid piece, no coupling is required.

An additional advantage of close-coupled pumps is that no bearings are required in

the pump. The pump and motor are positioned close enough so that the motor bearingsalone are sufficient to carry the loads generated by the impeller.

The close-coupled configuration imposes special requirements on the shaft (or shaft

sleeve) material. Because the shafting penetrates into the casing and contacts the pumpedliquid, it must be able to resist any corrosive effects.



f=

25

η

H P

Flow (Q)

Best Efficiency

Point

Head (H)

Power (P )

Efficiency (η )

“Drooping” Head

at Low Flow(Radial Flow Pumps) +

+

Pump Performance Curves

Figure 2.16 Pump Performance Curves. Flow (Q) is the amount of liquid flowing the the

pump. Head ( H) is energy added to the liquid by the pump. Power ( P ) is energy supplied to

the pump by the driver. Efficiency (η ) is a measure of how well the pump converts the energy

supplied to it by the driver into energy added into the liquid. Efficiency peaks at the Best

Efficiency Point (BEP). The “drooping” head at low flow is characteristic of some radial flow

pump curves. The “drooping” chararcteristic is notable because a pump operated in this

region can experience unstable operation (see “How Pump and System Curves Relate”).

PERFORMANCE, OPERATION, AND TERMINOLOGY

Pump Performance Curves: Important Pump Parameters

The sole purpose of a centrifugal pump is to use rotative shaft energy from a driver to

raise the head, or energy level, of the liquid flowing through it. The relationship betweenthe rotative shaft energy input by the driver, the head output by the pump, and the

efficiency of this energy conversion process is expressed in the pump performance

curves.The three parameters in Figure 2.16 are plotted against flow, Q, for a constant speed.

Most pumps in use at this time are not variable speed, thus the curves provided by

manufacturers’ will show pump performance at a fixed speed only.

The SI units of flow are meter 3

per hour (m3/hr) and the U.S. customary units of flow

are gallons per minute (gpm). The SI and U.S. customary units of the plotted parameters

are:

•Rotative shaft energy from driver, P – kilowatts (kw) or brake horsepower (bhp)

•Head added to output by pump, H – meters (m) or feet (ft)

•Efficiency, η - % of energy output (pump head) to energy input (power from driver)



f=

26

Figure 2.16 shows that head decreases as flow increases. This increasing

head/decreasing flow curve shape is often referred to as “rising to shutoff”. The “rising

to shutoff” shape is true for all centrifugal pumps with the exception of some radial flow pumps which “droop” at low flow – that is, have a decrease in head as flow decreases.

Stable pump operation requires that pumps rise to shutoff. Manufacturers generally

recommend that radial flow pumps with drooping curves not be operated in the droopingregion.

The steepness of the curve varies depending on the type of impeller. Radial pumps

tend to have the flattest curves. Mixed flow pumps have steeper curves while axial flowcurves are the steepest of the three types.

Efficiency (η ) is a measure of how well the pump converts the energy supplied to it

by the driver into energy added to the liquid. The pump efficiency curve rises, peaks, and

then falls off. A pump operating at the peak (Best Efficiency Point) is producing themaximum head for the least amount of power input. Pump efficiency is affected by the

following losses:

• hydraulic losses – frictional resistance to liquid flow through the impeller and

case passages• volumetric losses – leakage from discharge back to suction past wear rings

(closed impellers), or the front of vanes (open impellers)

• mechanical losses – friction between mechanical parts such as seals, packingrings, shaft, and bearings

• disk friction losses – frictional resistance of the liquid trapped between rotatingimpeller (which can be thought of as a disk) and the stationary case

These losses increase the amount of power required by the pump to output a desired flow.

The combination of these losses make up the total pump efficiency.The practice of trimming, described above in the section entitled “Impeller”, involves

machining down the outside diameter of an impeller in order to reduce the head output

and the power required. Trimming effectively shifts the entire head curve downwardwithout changing its shape. Manufacturers show how trimming affects performance by plotting several curves for a single pump where each curve corresponds to different

impeller diameter.

System Curves

A pump is not an isolated piece of machinery, it operates within a larger system

(Figure 1.1). When using the term “system” in the context of pumps, we are referring the pipes, fittings, and valves that deliver liquid to the pump suction and carry it away from

the pump discharge.

A pump application engineer tasked with specifying a pump needs to know how

much head is required to overcome the resistance of a given flow through the system. Asystem curve displays this information by plotting head on the vertical axis versus flow

on the horizontal axis.

Note that the head plotted in a system curve is the energy lost in the liquid due to thefrictional resistance and elevation change. This is not to be confused with the head

plotted on a pump curve, that is the energy added to the liquid by the pump. Thus,

increasing head on a system curve means more head is being consumed whereasincreasing head on a pump curve means that more head is being produced.



f=

27

System curves generally have a parabolic shape (Figure 2.17). Frictional losses are

responsible for this shape because fluid friction increases with the square of the flow

velocity. Changes in elevation of the flow (i.e., flow uphill or downhill) shift the curveup or down without changing its shape.

Pump Operation: How Pump and System Curves RelateThe pump H-Q curve defines the head a pump will produce at various flows and the

system H-Q curve defines the head that a system will consume, also at various flows.

When the two curves are put together in a single graph (Figure 2.17), the pump operating point is defined. This operating point occurs at the intersection of the two curves. The

pump will supply exactly the amount of head needed to overcome system resistance at

the given flow.

Pump curves that “rise to shutoff” allow the pump to function in a stable operating

mode. The “drooping at low flow” regions of some radial pump curves are unstable because there are two flows for a given head. Pumps operating in the unstable region

tend to “hunt” or fluctuate between the two points as they search for their operating point.

These pressure and flow fluctuations result in surging.

The Best Efficiency Point (BEP)

The efficiency curve in Figure 2.16 shows that there is one particular flow whereevery pump will operate with maximum efficiency - this operating point is known as the

Best Efficiency Point (BEP). A pump operating at its BEP is producing head with the

minimum amount of losses (pump losses are described in “Pump Performance Curves”above).

The BEP is important for two reasons:

Flow (Q)

System

CurveHead (H)

+

+

Pump versus System Curve

Pump

Curve

Pump Operating

Point

Figure 2.17 Pump operating point occurs where the pump and system curves intersect.



f=

28

1. Economics: Operating costs are minimum when a pump operates with maximum

efficiency.

2. Radial Loads: Radial (side) loads are minimum when a pump operates at its BEP(the sections entitled “Casing” and “Radial Loads” explain the source and effects

of operation off of design capacity in more detail).

Specific Speed

Specific speed is a very important parameter because it provides a way to characterize

different pump impeller designs with respect to their head ( H ) versus flow (Q)characteristics.

The differences between radial flow, mixed flow, and axial flow impeller geometries

exist because each design is best suited for providing the different H-Q combinations

required by different applications. Radial flow impellers can deliver high head/low flow performance but not low head/high flow performance. Conversely, axial flow impellers

can only deliver low head/high flow performance. Mixed flow pumps fall in a large

general class somewhere in between axial and radial flow. One impeller type does not

satisfy all applications. Even if one impeller type could function at all the differentcombinations of head and flow, its operating efficiency would be poor. Good economic

practice demands that pumps be optimized for the intended application.Consequently, there needs to be a way to compare pump impellers with respect to

their H-Q optimization. The specific speed, N S , is an index that makes this comparison

possible. The specific speed is a number that can be calculated for every pump using thefollowing equation:

4/3S H

Q N N = (1)

where N = pump rotative speed (rpm),Q = flow at BEP and full impeller diameter (gpm)

H = pump head at BEP and full impeller diameter (ft).The SI unit version of specific speed is N SM where flow, Q, is given in meters

3/hour

and head, H, is given in meters. The conversion factor between the two is: N S =

51.65N SM [2].

As noted, specific speed is calculated at the BEP and full impeller diameter.However, once calculated, the specific speed for a particular pump is constant for

different values of rotative speed, N , and impeller diameter. The pump affinity laws

provide the basis for this fact. The pump affinity laws allow manufacturers and users to predict the effects of impeller trim or speed changes. Further explanation of pump

affinity laws is beyond the scope of this paper. However, they are worth mentioning because they are such important tools to pump manufacturers and users. The referenceslisted provide more detail on this subject [1,2].

In short, specific speed is a function of impeller geometry. If you know a pump’s

specific speed, you can guess its impeller type. Figure 2.18 shows this relationship. The

relationship between impeller geometry and specific speed holds true regardless of impeller size. Radial flow pumps provide high head/low flow and thus have a low

specific speed. In comparison, axial flow pumps provide low head/high flow and have a



f=

29

high specific speed. Figure 2.18 shows that between the two extremes is a continuum of

impeller profiles that will deliver different combinations of head versus flow.

Specific speed is useful when comparing the H-Q performance of different pumps.This is helpful when selecting the best pump design for a particular application. In most

pump applications, flow (Q), head ( H ), and speed ( N ) are predetermined. For example,

flow will be determined by the process requirements, head by the system friction andgravity losses, and speed by the driver to be used. (In practice, driver speed is somewhat

of a rough variable. Electric motors of different speed settings are available.) Combining

these terms into specific speed, N S , lets the pump application engineer select the rightimpeller geometry for the application.

Net Positive Suction Head (NPSH) and Suction Specific Speed (S)

Net Positive Suction Head (NPSH) and Suction Specific Speed (S) are two parametersthat describe how much suction head a pump requires in order to prevent cavitation

(cavitation is explained in detail below in the section entitled “Cavitation”).

Net Positive Suction Head (NPSH)Centrifugal pumps cannot “pull” or suck liquid into themselves. (This is true whether

a pump is operating at full capacity or just starting up.) Instead, liquid must be pushed

into them by a driving or “positive” pressure. If the driving pressure is insufficient, theliquid will turn to vapor (or cavitate) at the point in the pump where pressure drops to its

lowest level. Friction losses cause the driving pressure to decrease as the liquid flows

from the suction flange to the point where the impeller vanes begin to raise the pressure.Since the point of lowest pressure happens to be in the impeller vane passages, this where

cavitation will occur. In order to avoid cavitation, the pressure measured upstream at the

suction flange must be high enough so that the pressure downstream in the impeller

remains above the liquid’s vapor pressure at all times.The driving pressure is properly referred to as the Net Positive Suction Head. As the

term indicates, NPSH is measured at the pump suction flange simply because it cannot be

5 0 0

3 0 0 0

1 0 0 0

Axis of

Rotation

Impeller Profile versus Specific Speed (N S )

Values of Specific Speed (N S )

1 0 0 0 0

1 5 0 0

4 0 0 0

6 0 0 0

8 0 0 0

1 5 0 0 0

2 0 0 0 0

Radial Vane Francis Vane Mixed-flow Vane Axial Vane

(Propeller)

2 0 0 0

7 5 0

Figure 2.18 Relationship between Impeller Profile and Specific Speed. The specific speed number

indicates the H-Q performance of a pump impeller. Radial flow pumps deliver high head/low flow

peformance and thus have a low specific speed. Axial flow pumps have a high specific speed because they

deliver low head/high flow performance. Other impeller types fill the gap in between the two extremes.

(This figure taken after Figure 2.15 in [2]).



f=

30

measured within the rotating impeller vanes where cavitation can occur. The least

amount of NPSH that will prevent cavitation is termed the NPSH Required (NPSHR).

The NPSHR differs from pump to pump because it is dependent on how well the pumpinlet design prevents friction losses. Not only does the NPSHR differ between pumps,

but for any given pump it changes with flow. This is explained by the fact that friction is

velocity dependent and velocity is a function of flow.In addition to pressure, the temperature of the pumped liquid also determines whether

or not cavitation will occur (see “Cavitation”). Consequently, the NPSH Required to

avoid cavitation is based on a certain temperature. The temperature of the pumped liquid must remain below the temperature upon which the NPSH Required values are based in

order to avoid pump cavitation.

Theoretically, cavitation will not occur as long as the NPSHR is less than NPSH

available in the pumped liquid (the emphasis on “theoretically” will be explainedshortly). The Available Net Positive Suction Head (NPSHA) is really a sum (or net) of

several positive and negative pressures acting of the liquid. They are:

• Atmospheric head: the static pressure acting on the liquid, usually atmospheric

pressure measured at a known reference point.• Suction head or suction lift: the elevation head from the reference point to thesuction flange. Positive if reference point is higher than suction flange, negative if

reference point is lower than suction flange.

• Friction head: the friction losses in the piping between suction and reference point, therefore a negative pressure.

• Vapor pressure head: the vapor pressure of the pumped liquid at the operatingtemperature, a negative pressure.

Most pump and fluid dynamics reference books show how to calculate NPSHA. Thereader who desires greater detail on this subject may wish to refer to those sources [2].

In actual practice, cavitation can still occur even if the NPSHA exceeds the NPSHR.

The reason for this lies with the test method that manufacturers use to determine the NPSHR figures. Manufacturers test a pump by operating it at a steady flow with

excessive NPSHA. The NPSHA is then gradually reduced until the onset of cavitation is

detected. The NPSHA at which cavitation begins is figured to be the NPSHR for the

given flow.The problem with this procedure is with the method that manufacturers use to detect

the onset of cavitation. Cavitation is determined to be present when the head produced

by the pump has dropped 3% in response to the reduction in NPSHA. This method ismisleading because cavitation is actually present before the 3% breakaway (so termed

because cavitation causes the curve to “breakaway” from the normal H-Q curve). Thus,

merely maintaining NPSHA in excess of the NPSHR may not be enough to prevent

cavitation damage.Consequently, most pump operators use a margin or ratio between Available and

Required NPSH to avoid cavitation. Some operators use a fixed margin, for instance 5

feet of head minimum difference. Other guidelines on NPSH calculate ratios that varywith certain pump parameters and, sometimes, also the liquid being pumped.



f=

31

Suction Specific SpeedAnother important pump parameter is the Suction Specific Speed (S ). The suction

specific speed is calculated in a manner similar to the pump specific speed ( N S ):

4/3 PSHR

Q N S = (2)

where N = pump rotative speed (rpm),

Q = flow at BEP and full impeller diameter (gpm)

NPSHR = the Required NPSH (ft).

The suction specific speed is similar to the pump specific speed in that it also is an

index. However, while specific speed compares impeller geometry, the suction specific

speed compares the pump inlet geometry. Within certain limits, a high suction specificspeed is desirable because it indicates that the pump produces fewer friction losses

through the inlet (i.e., has a low Required NPSH) making cavitation less likely.



f=

32

3. MALFUNCTIONS OF CENTRIFUGAL PUMPS

GENERAL CONCEPTS

Centrifugal pumps, like other rotating machines, experience the malfunctionscommon to rotating equipment. In fact, those readers familiar with compressors willrecognize their similarity with centrifugal pumps. Both add kinetic energy using the

same principles described earlier and both share some general similarities in construction

and design. However, the fact that pumps handle liquids while compressors handle gasesmust not be overlooked. Liquids are much more dense and viscous than gases and

essentially non-compressible. This causes the symptoms of pump malfunctions to

manifest themselves somewhat differently than compressor malfunctions. The heavy

damping of liquids acts to suppress shaft relative vibration amplitudes and in particular,subsynchronous vibrations. In addition, the high density of liquids creates fluid forces

that are not found in compressors and turbines. These fluid forces have a couple of

effects. First, they are responsible for a few malfunctions that exist only in centrifugal pumps (e.g., Hydraulic Unbalance). Secondly, they often modify the balance

resonances. A pump operated “dry” (no process liquid surrounding the impeller) may

experience a different (often lower) balance resonance than when run “wet” (impeller fully immersed) [4]. (Caution: some pumps may suffer damage from dry running. Dry

runs should only be performed under certain conditions by those fully knowledgeable

with the pump and its requirements for safe operation.)Pump malfunctions are often accompanied by some general signs. These are “high”

vibration, excessive noise, reduced bearing and/or seal life, high bearing temperatures,

and poor performance (higher than normal power consumption or lower than normal

output). The manner in which these signs relate to the various malfunctions will be

outlined in more detail in each of the following sections.Some of the following malfunctions are discussed in depth in the Machine Library

Malfunction Diagnosis articles. Those papers that have been written will be noted andreferenced in each section. The reader who desires more detail can refer to them.

Otherwise, each section will provide a brief explanation of the general mechanism

underlying the malfunction. Then, each malfunction will be related to those things incentrifugal pumps that may act as root causes.



f=

33

PUMP MALFUNCTIONS

High 1X Vibration due to Unbalance

Centrifugal pumps are susceptible to two types of unbalance: mass unbalance andhydraulic unbalance. Mass unbalance in centrifugal pumps is the same malfunction that

we experience with other types of rotating machines. However, hydraulic unbalance isunique to centrifugal pumps. Whereas both types of unbalance produce high 1X

vibration, they are caused by different phenomena and are corrected by different means.

Therefore, the causes and corrective actions of each type of unbalance will be discussed

separately. However, since they produce similar behavior, their symptoms will bediscussed together.

High 1X due to Mass Unbalance

Definition of Mass UnbalanceRotors inherently contain some mass unbalance that causes a 1X vibration. (The rotor

includes shaft, shaft sleeves, impeller, and couplings – anything that is locked to and

moving in unison with the shaft.) This type of unbalance is known as mass unbalance because it originates from the mass of the rotor. The resulting vibration varies directly

with the amount of unbalance. If the vibration exceeds a prescribed level, then damage to

the pump, its driver, or attached structures can occur.Mass unbalance is caused by the fact that the mass center and the geometric center of

the rotor do not lie at the same point. If we consider a single cross section anywhere

along the rotor axis, the mass center is the point about which all the mass is equallydistributed. The mass center can be thought of as the balance point. If you balance a flat,

circular object (like a dinner plate) on the tip of your finger, you are supporting it directlyunder the mass center. The geometric center is different, it is the point within the rotor

about which the geometry (or shape) isequally distributed or symmetric. We are

generally accustomed to thinking of the

rotor mass and geometric centers as one inthe same, however, this is never true in the

real world. Typically, a rotor is slightly

heavier on one side because of manufacturing tolerances, or deposits,

pitting, etc. The heavier side causes the

mass center to be offset in the samedirection. The offset, Rr , between the two

centers is directly responsible for the

condition in a rotor system referred to as

mass unbalance (Figure 3.1).The offset is a problem because, at

rotative speeds below balance resonance

(or critical speed), rotors turn about their

++

Mass

Center

GeometricCenter

Unbalance

Force

R r (Offset between Mass and

Geometric Centers of Rotor )

Figure 3.1 Unbalance Force. A rotor is supported by

bearings and, to a much lesser extent, by seals.

These supports cause a rotor to turn about its

geometric center. However, the offset between the

mass and geometric centers creates an outward

pointing centrifugal force, or unbalance force.



f=

34

geometric center and not their mass center. The offset between the two centers creates an

outward pointing centrifugal force, or unbalance force. This force is identical to that felt

when twirling a string with a rock tied to the end of it. The unbalance force is defined bythe equation for a centrifugal force:

2

Ω= r r R M ForceUnbalance (Eq. 1)

From this equation we see that the unbalance force depends upon M r (the mass of rotor),

Rr (the rotor Smass to geometric center offset), and Ω 2 (the square of the rotative speed).

A few items about the unbalance force are worth noting. First, the outward direction

of the unbalance force explains why it causes vibration to increase. Though this is fairly

obvious, it helps us to visualize the effect of the unbalance force upon vibration when weconsider the fact that it is literally pushing the rotor outward. Secondly, the unbalance

force points outward from the geometric center in the same angular direction as the mass

center. Thus, the angular direction of the unbalance force will shift only if the masscenter shifts. Since the location of the mass center is generally stable, so is the direction

of the unbalance force. The mass center can change only if there are changes in the rotor assembly such as changing shaft bow, deposits, pitting, or a loose rotating part. Sincethese changes are the exception rather than the norm, any apparent shift in the direction

of the mass center should prompt us to look for the reason why the shift occurred.

Thirdly, the unbalance force rotates at the same speed (or synchronously) as the rotor.This last point is notable because it explains the 1X nature of the vibration response to

mass unbalance.

Mass unbalance is discussed in greater depth in the corresponding Machine Library

Malfunction Diagnosis article [11]. The reader who desires more information on thistopic may wish to refer to that source.

Causes of Mass UnbalancePumps usually come from the OEM in a well balanced state. However, the balance

state can worsen if changes occur in the pump over time or due to maintenance. Whilesome examples of such changes are given below, anything that adds or removes mass or

shifts the center of mass may have an adverse affect on a pump’s balance state. The first

three examples illustrate increase in unbalance without the addition or removal of mass;

the mass center shifts simply because parts on the rotor shift relative to each other. Thesecond three examples illustrate an increase in unbalance due to the addition or removal

of mass.

1. Replacing OEM parts with more loosely toleranced non-OEM parts. Non-OEM parts may be manufactured to looser tolerances. Parts with looser

tolerances may shift the mass center because they do not sit on the shaft as concentrically

as an OEM part. For example, large, custom pumps such as boiler feedpumps and theattached OEM couplings are typically well balanced by the factory. However,

replacement of the coupling with a non-OEM part during maintenance may increase the

unbalance. [5]

2. Loose fit impellers.

Loose fit impellers are those which mount to the pump shaft with a clearance fit.

They are typically locked against rotation to the shaft with a key. Unbalance can occur



f=

35

when a loose fit impeller is removed from the pump shaft and balanced on an expanding

mandrel (which is tight fitted), then removed and remounted back on the pump shaft.

The effect on unbalance due to the slight shift in mass center between the expandingmandrel and the looser pump shaft becomes increasingly pronounced with larger

impellers. [1]

3. Shrink fit impellers.Shrink fit impellers are locked to the shaft by an interference fit. The impeller and

shaft experience residual stress in the region of the fit. The residual stress can relax over

time due to temperature cycling, shaft vibration and flexing. As residual stress relaxes,impellers can cock or bow the shaft causing the mass center to shift. [1]

4. Clogs.

Many pumps handle liquids that contain solid objects in the process stream (e.g.,

sewage pumps). While such pumps are usually designed to freely pass objects of acertain size, clogs can still occur. This is especially true for closed impellers. [5]

5. Deposits.

Many pumps process liquids containing substances that can deposit on the impeller.

Deposits will add mass as they collect, however, mass will also be removed if they break off. In either case, deposits tend to collect or break off in irregular patterns that upset the

balance state of the rotor.

6. Pitting (impeller erosion).

Impellers can lose material through pitting. Pitting may occur for a couple of reasons.

First, the pumped liquid may be so corrosive or abrasive that it erodes the impeller and,

especially, the highly exposed vanes. Secondly, pitting can occur even in harmlessliquids for another reason: cavitation. Cavitation is discussed in greater detail below in

“Cavitation”. However, suffice it to say here that cavitation can cause very severe

damage, even eating holes clear through impeller vanes.

Corrective Actions for Mass Unbalance

It should be evident from these examples that mass unbalance is often a symptom of another underlying root cause. While the corrective action will probably include

rebalancing the pump, it should also include diagnosis and correction of any underlying

root cause.

High 1X due to Hydraulic Unbalance

Definition and Cause of Hydraulic UnbalanceCentrifugal pumps are subject to another source of high 1X vibration known as

hydraulic unbalance. While hydraulic unbalance is similar to mass unbalance in itsvibration signature, it has a different underlying cause and, hence, a different corrective

procedure.Hydraulic unbalance originates in the fluid forces acting on the impeller. Just as the

impeller acts on the liquid to increase its angular momentum (see “Impellers” above), sotoo the liquid acts against the impeller in an equal and opposite reaction (recall Newton’s

third law). If the liquid does not flow evenly through all of the impeller vane passages,

then these reaction forces will be unbalanced.The references sited in this paper do not describe the exact unbalance mechanism.

However, there is one possible explanation consistent with pump theory that does stand



f=

36

out. Recall from the earlier discussion of velocity triangles (“Impellers”) that the total

velocity head added to the liquid depends on two parameters: U and V T where U

represents the flow velocity due to angular speed at a given radius and V T represents thetangential flow velocity due to flow along the vane surface. Both parameters depend

upon the impeller geometry and its tolerances. If the impeller radius varies excessively

around the periphery, then U will vary excessively. Likewise, if vane angles are notsymmetrical within a specified tolerance, then V T will also vary excessively. Either one

of these or the combination of the two may cause the fluid forces surrounding the

impeller to be asymmetrical. This mechanism points to unacceptably high geometrictolerances as being the underlying root cause of hydraulic unbalance.

Corrective Action for Hydraulic UnbalanceThe corrective actions stated in the references are also in agreement with the

mechanism just described. The only way hydraulic unbalance can be remedied is by

switching to a more precisely manufactured impeller. In fact, it seems to be wellestablished that for a particular manufacturing method, there is fixed amount of

unbalance that will have to be tolerated. The amount of hydraulic unbalance associated

with a manufacturing method is described by one reference as K H , where K H is thenormalized hydraulic unbalance force (lb) [4]. Values of K H for a sand-cast impeller are

as high as .10 while precision-cast impellers (i.e., investment cast) may have values of

.005 to .025 and machined impellers can values as low as .0025.

A recommendation to switch to a more precisely manufactured impeller was thecorrective action suggested in one Bently Nevada Machinery Diagnostic Services case

history involving a vertical slurry pump. The MDS engineer diagnosed a hydraulic

unbalance due to errors in the impeller geometry as the root cause of the high 1Xamplitudes observed. The hydraulic unbalance was confirmed when the high 1X

amplitudes disappeared during a dry run of the impeller.

Vibration Characteristics of Unbalance (Mass and Hydraulic)As noted above, high 1X (shaft relative or casing vibration) is the predominant

vibration component that accompanies mass unbalance or hydraulic unbalance. The high

1X will be especially noticeable when its frequency is close to a balance or structuralresonance.

Effects of High 1X VibrationThere are several tight clearances in pumps that are vulnerable to high vibration.

Seals, packing, bearings, wear rings, couplings all contain these clearances and can be

damaged when forced to operate beyond design limits. Also, high 1X can exciteresonances in adjacent structures and cause their stress levels to exceed design limits (see

“Structural Resonances” below).

Other Sources of High 1X VibrationMany other malfunctions also exhibit increasing 1X vibration. From our

understanding of rotor dynamics (see Machine Library Dynamic Stiffness and Rotor

Response) we know that vibration (or rotor response) is the ratio of unbalance force todynamic stiffness. Therefore, vibration may increase due to reduced system stiffness

while the unbalance force remains constant. Some examples of this are: “softfoot” (i.e.,



f=

37

loose bolts, degraded foundation), bearing or seal wear, shaft crack, alignment changes,

quadrature stiffness changes due to different pumped fluid or lube oil characteristics.

These examples help to illustrate why it is so important to verify the source of the 1X behavior before balancing. Balancing a machine will not address a problem whose root

cause is due to some other malfunction, such as a shaft crack.



f=

38

Figure 3.2 Perfect internal alignment exists in a

machine when the centers of all of the internal

parts are collinear.

Radial Loads (Misalignment and Sideload)

The term radial load refers to any load that acts on the rotor in a unidirectional radial

direction. When discussing radial loads, we only include those loads whose direction and

magnitude are constant or vary gradually over time with process changes. We are not

including radial loads, such as unbalance, that rotate with the rotor. Rotating radial loadshave different symptoms and produce different effects from (relatively) constant radial

loads.

Radial loads act to push the rotor to one side of the bearing. This effect is not entirelyunwanted. For instance, sleeve bearings are less prone to fluid-induced instability when

the rotor does not ride in the center of the bearing. Therefore, sleeve bearings are usually

designed to take advantage of normally existing radial loads, like gravity, to keep the

rotor from operating in the bearing center. In contrast, when discussing radial load as amalfunction, we are referring to loads that exceed the design of the machine. Radial

loads in excess of design limits can lead to serious pump damage if not detected and

rectified in their early stages.

Two common sources of excessive radial load in pumps will be discussed here: 1)misalignment and 2) the radial load on an impeller (or sideload ) that occurs when

operating a pump too far away from its Best Efficiency Point (the Best Efficiency Point,or BEP, is explained above in “The Best Efficiency Point”). Since these two types of

radial loads have different root causes, their underlying mechanisms and their corrective

actions will be explained separately. However, since misalignment and sideload manifestsimilar behavior, their symptoms and effects will be discussed together.

The reader who desires more discussion on the topic of misalignment can refer to the

corresponding Machine Library Malfunction Diagnosis article [12].

Radial Load due to Misalignment

Definition of Misalignment

Misalignment is a very important source of radial load because it is responsible for so

many pump failures. One reference [1] even stated that “Outside of serious unbalance of pump components, there is no single

contributor of poor mechanical

performance more significant than poor alignment.”

In order to understand how

misalignment creates radial loads, wemust first understand the broader concept

of alignment. Perfect internal alignment

exists when the centers of all of amachine’s bearings, interstage

diaphragms, and seals are located on the

same line and that line is the centerline of

the machine (Figure 3.2). Two machineswould be in perfect external alignment if

the centerlines of their shafts were on the



f=

39

Parallel Misalignment

Angular Misalignment

Aligned

Figure 3.3 Two machines are in perfectexternal alignment (top) when their shaft

centerlines are collinear within an allowable

tolerance zone (red). With parallel misalignment

(middle) the shaft centerlines are offset, but

parallel. With angular misalignment (bottom), the

shafts are oriented at different angular

orientations in space. Misalignment usually

involves a combination of parallel and angular

misalignment. In the figure, the shafts are shown

centered in the bearings. In reality, gravity loaded

shafts would rest in the bottom of the bearings

Top View

Side View

Figure 3.4The 3-dimensional misalignmentproblem is usually broken down into two 2-

dimensional problems.

same line (Figure 3.3, top). In practice,

some degree of internal and external

misalignment always exists. Flexiblecouplings are designed to accommodate a

certain amount of misalignment, and that

amount will depend on the type of coupling being used. When the misalignment

exceeds the allowable tolerances for the

coupling in use, the machines are said to bemisaligned.

There are two basic types of external

misalignment. Parallel misalignment occurs

when the centerlines of two machines havethe same angular orientation, but are

separated from each other (Figure 3.3,

middle). Angular misalignment occurs

when the centerlines of two machines havedifferent angular orientations (Figure 3.3,

bottom).An additional type of “misalignment”

involves the axial position of two machines

coupled together. Coupled machines can

have correct parallel and angular alignment but still suffer incorrect axial alignment.

The tolerance for axial position for two

machines will depend on the type of coupling that is used. Rigid couplings have

a very low tolerance for axial positionerrors, while disk pack and diaphragm

couplings have somewhat more, but stillsmall tolerance for error. Gear couplings

have a higher tolerance for axial position

errors.Misalignment is a three-dimensional

problem. Each machine has a centerline

that exists at some orientation and position

in space, and the centerline of an adjacentmachine will have a different orientation

and position. To make external alignment problems easier to solve, the three-dimensional centerlines of the machines are

projected on two perpendicular planes

(Figure 3.4). Then, the alignment problem can be treated as two, separate, two-dimensional problems.



f=

40

Cold Misaligned

Hot Aligned

Figure 3.5 Machines are deliberately

misaligned cold so that, when they reach hot

operating temperature, thermal growth will align

the machines.

Causes of MisalignmentThe condition of misalignment can result from any one or a combination of several

causes. When discussing these causes, it is important to distinguish them from the

condition of misalignment. While the condition of misalignment may cause undesirableeffects in a pump, we must not forget that the misalignment itself usually results from an

even lower level of root causes. This is important to keep in mind because, if we re-align

the pump without diagnosing and correcting

the underlying cause, then misalignment mayreappear as the situation worsens.

1. Thermal growth.

As the temperature changes during astartup, the linear dimensions of a machine

can change in complicated ways, with the hot

parts growing more than the cooler parts.Dimensional changes in the machine supports

and casing can cause the machine to rise or

fall and/or change angular orientation as itheats up. Any adjacent machine will also

change, and that change will most likely

follow a different pattern. Thus, if the two

machines were aligned in a cold condition,they would become misaligned in a hot condition. For this reason, machines are

deliberately misaligned in the cold condition, and the cold misalignment is carefully

calculated to produce correct alignment in the hot condition (Figure 3.5).Because the temperature of a machine can vary with load, alignment can also change

with load, and it may be difficult to set a cold alignment that produces acceptable hot

alignment for all anticipated operating load conditions.

2. Foundation problems.

Foundation problems can cause a shift in machine position over time. Foundation

problems can include cracked grouting, a loose soleplate, and loose anchor bolts. Oil

soaked concrete can lead to deterioration of the foundation and a loss of support strength.

3. Soft Foot.

Soft foot is a condition where one or more machine feet are not coplanar after

tightening hold down bolts. When one foot is not properly supported (the soft foot),tightening down the soft foot will warp the machine casing. Softfoot can be caused by

inadequate shimming or by an excessive number of shims, which can produce a springy

support. There should be no more than 3 to 4 shims under a foot. Soft foot can also becaused by a warped or bowed soleplate, an improperly installed soleplate, improper

machining of feet or the soleplate, or a foot not parallel to the soleplate. A warped or bowed machine casing can also cause soft foot.

4. Piping strain.

Piping strain can warp a machine casing and cause misalignment by moving the

bearing supports. Pipe strain can result from loose piping hangers or bent, broken, or

missing piping supports. Poor piping fit can put tremendous loads on the machine casing.Piping should never be forced to mate with the machine through the use of force.



f=

41

5. Improper alignment.

While not often the case, it is always possible that the current alignment is not

sufficient and therefore needs to be redone. If other root causes have been carefullyconsidered and ruled out, then it is possible that the pump and its driver simply need to be

re-aligned.

Radial Load due to Pump Sideload

Definition of SideloadingThe hydraulic forces that act radially on the impeller cause the second common

source of radial load in centrifugal pumps. These forces combine to create a resultantforce known as a sideload. Sideloads are explained in greater detail above in “Casing”.

However, in short, high sideloads occur when centrifugal pumps are operated off of their

Best Efficiency Point (BEP). Single volute pumps are especially vulnerable to sideload.While the magnitude and direction of sideloads vary with flow, they meet the basic

criteria for radial load in that they are relatively constant and only change gradually when

compared to rotative speed.

Sideloads can reach very high magnitudes - high enough to break pump shafts, causerubs or do other serious damage. The magnitude and direction of sideload can undergo

extreme variations. In fact, the sideload at shutoff (no flow) can be as high as 10 to 15

times greater than at BEP and that direction can change by almost as much as 180degrees [1].

Vibration and Temperature Characteristics of Radial LoadExcessive radial load can be suggested by one or more of the following

measurements:

1. High Bearing Temperature.High fluid-film bearing temperature is often the first warning of a possible high radial

load condition. The high radial load can cause high shearing stresses in the lubricatingfluid of an overloaded bearing. The extra work done in overcoming these higher thannormal fluid stresses produces extra heating of the fluid. The fluid, usually oil, transfers

this excess heat to the bearing babbitt.

Oil drain temperature is not a very useful indicator of the temperature in the bearing.It is limited because, at that point, the oil exiting the bearing is a mixture of oil that has

passed through the load zone of the bearing and oil that has bypassed the load zone. It is

best used for plant heat load calculations or oil temperature regulation, but it only provides a vague picture of the machine condition.

Resistive Temperature Devices (RTDs) or thermocouples that are imbedded in the

bearing babbitt can provide better warning. Ideally, the RTDs should be installed at

several different circumferential positions in the bearing. The are two reasons for this.First, in some machines, the direction of the radial load on a bearing can vary with

operating conditions, and it can be difficult to predict where the maximum load occurs in

a bearing. Second, if a machine becomes misaligned, load shifting can produce radialloads that act in unpredictable and unexpected directions.

While high bearing temperatures indicate overloading, an abnormally low bearing

temperature indicates that the load in that bearing may be below normal. Given the loadshifting that takes place with radial loading, one bearing may show an unusually high



f=

42

Slow

Roll

Normal

Running

Position

Assumed

Radial Load

Abnormal

Running

Position

Misalignment Load

Figure 3.6 Comparison of normal and abnormal

shaft centerline behavior during a startup of a

typical, horizontal, gravity loaded machine with plain, cylindrical, fluid-film bearings. Here, the

radial load is assumed to be vertically downward.

The machine is rotating X to Y (CCW). The dashed

circle defines the bearing clearance. As speed

increases, the hydrodynamic oil wedge becomes

stronger, and the normal rotor moves up and

slightly away from the bearing wall (green). When

misalignment forces are present, the behavior can

be quite different (red), and the rotor can end up

operating in an unusual quadrant in the bearing.

(Note that rotors operating in tilt-pad bearings

normally tend to move straight up toward the radial

load with increasing speed.)

1 23 4

Figure 3.7 The misaligned machine train is

shown with the operating average shaft centerline

positions for each bearing. Note that, for this

example, the rotor positions in bearings 1 and 4 are

approximately normal, while the rotor positions in

bearings 2 and 3 are in opposite quadrants,

indicating a possible misalignment condition.

temperature, while another, adjacent

bearing may show an unusually low

temperature. For this reason, bearingtemperatures should be monitored and

compared along the machine train.

2. Average Shaft Centerline Position.For a horizontal, fluid-film bearing

machine train which is perfectly aligned,

and in which gravity is the primary radialload, the average shaft centerline position

will change during a startup or shutdown in

a typical way (Figure 3.6, green).

Normally, the shaft position angle will besomewhere between 0° and 45° from the

direction of the applied load. If the

machine train experiences abnormal radial

load, then load shifting will cause changesin the behavior of shaft centerline plots

(Figure 3.6, red). For example, the radialload due to misalignment can be in a

different direction, and the direction and

amount of the misalignment load can

change as the machines heat up. Heavilyloaded bearings will have operating

eccentricity ratios that are higher than

normal, while lightly loaded bearings willhave operating eccentricity ratios that are

lower than normal. If the misalignment

becomes severe enough, shaft operating positions may move to unusual locations,

such as near the top of a bearing (Figure

3.6).

Differences in operating position can be most apparent across a coupling

between two machines, where the rotor

may operate in opposite quadrants of the bearings (Figure 3.7).

Average shaft centerline plots should

be examined at every axial position andcompared for signs of abnormality.

Average shaft centerline plots are most

useful when clearance circles are knownand included on the plot. That way,

operation in an abnormal quadrant can be

more easily detected. Shaft centerline plots

should be compared to previously archived



f=

43

data and examined for changes.

3. Orbits.

Orbits can be very helpful for diagnosis of radial load. Normally-loaded, healthyrotors that operating in plain, cylindrical fluid-film bearings tend to produce direct, or

unfiltered, orbits that are elliptical in shape and where the frequency is predominately 1X

(Figure 3.8). The ellipticity of such orbits can fall into a wide range and still beconsidered normal.

Elliptical and lemon-bore bearings tend to normally produce orbits that are more

elliptical than those produced by plain cylindrical or tilt-pad bearings. Also, the major axis of the ellipse in elliptical and lemon-bore bearings tends to be aligned with the

bearing geometry.

Because radial loads can change magnitude and direction with load, orbits can vary in

size and shape with load. Also, any resonances will affect the size and appearance of theorbit.

Because of the many possibilities, a database of normal operating orbits for a

particular machine should be archived for later reference.

The shape of a direct orbit is sensitive to the amount of the radial load that acts on therotor. As the radial load is increased, the orbit will become more flattened, and part of the

orbit path may partially follow the curvature of the bearing. (Note that elliptical andlemon bore bearings tend to normally produce more elliptical orbits than would occur

with plain cylindrical bearings. For these bearings, the orientation of these elliptical orbits

tends to be more aligned with the bearing geometry.) The orbit may also become banana

shaped, containing a 2X vibration component that is visible on spectrum plots (Figure3.8C). 2X components can be amplified if the rotor operates at half of a resonance speed.

In extreme cases of radial load, the rotor may become so constrained that the orbit

follows a line that matches the curvature of the bearing (Figure 3.8D) or, if unbalance issmall, may shrink to nearly a point. Assuming that unbalance is the primary source of

rotor vibration, the details of the orbit behavior will depend on the degree of radial load,

the amount of unbalance forcing in the rotor, and the attitude angle and eccentricity ratio

A B C D

Figure 3.8 Unfiltered orbits. Each orbit shows eight shaft revolutions. Orbit A is a normal orbit

from a generator bearing on a small steam turbine generator set. The orbit is mildly elliptical and

predominately 1X. Orbit B is from a Frame 6 gas turbine bearing. The orbit shows evidence of

constraint along the lower edge, suggesting a misalignment problem. Orbit C is from the exciter

bearing on a 500 MW steam turbine generator set. Note the highly elliptical, banana shape. The

banana shape will produce a 2X vibration component which would be strongest in the horizontal

direction. Note the curvature of the right side of the orbit, which suggests that the shaft may be

following the geometry of the bearing boundary. Orbit D is from a HP steam turbine bearing. The

orbit is highly flattened, suggesting a high, misalignment-induced radial load. (Note that line orbits

can occur for other reasons.)



f=

44

1 23 4

Figure 3.9 The misaligned rotor of Figure 5 is

shown with possible orbits. All orbits

correspond to the same operating speed. The orbit

size (vibration amplitude) is partially controlled

by the bearing stiffness, which is a function of

eccentricity ratio. Thus, the heavily loaded

bearing 2 orbit is small, while the lightly loaded bearing 3 orbit is relatively large. The bearing 2

orbit partially follows the contour of the bearing.

Bearings 1 and 4 are approximately normally

loaded.

in the bearing.

A rotor that is unloaded in a bearing

because of misalignment or sideload mayoperate at a low eccentricity ratio and have an

orbit that is nearly circular. Because elliptical

orbits are the norm, a circular orbit suggestsan unusually low radial load that could be due

to misalignment or sideload.

Multiple orbits should be displayed for every axial position in the machine train and

compared with each other (Figure 3.9). If

startup or shutdown data is available, these

multiple orbits should be examined over theentire speed range of the machine for evidence

of high radial loads. The orbits (which contain

dynamic position information) should be

correlated with average shaft centerline plots(which contain average position information)

over the length of the machine train.

4. Vibration.

Assuming that the source of vibration

originates in the rotor (for example, due to unbalance), the amount of casing vibration

will depend on the transmissibility of rotor vibration through the bearings and into thecasing. (Casing vibration will also depend upon how well the machine is mounted to the

foundation.) The very high fluid-film bearing stiffness that exists at high eccentricity

ratios acts to more effectively couple the rotor to the casing. Thus, in a radial loadedmachine, the rotor may transmit more vibration to the casing, and the machine may

experience higher than normal casing vibration. Rotor shaft relative vibration, because of

the increased constraint on the rotor (increased Dynamic Stiffness), may decrease asmore of the vibration energy is transmitted to the casing.

If, because of radial load, a particular bearing is unloaded, the rotor may become

more decoupled from the casing (transmissibility will decrease) at that location, and the

casing vibration there may decrease. Under this circumstance, shaft relative rotor vibration may increase as the rotor support Dynamic Stiffness decreases.

Thus, either increases or decreases in casing vibration could be an indication of a

misalignment or sideload condition. An increase in casing vibration coupled with adecrease in rotor shaft relative vibration (and vice versa) suggests either condition.

Remember that casing vibration can increase if the machine support structure

weakens or loosens, or if the machine develops a soft foot. The reduced stiffness of themachine support allows vibration to increase. Sometimes, tightening loose foundation

bolts will reduce casing vibration back to normal levels.

Parallel misalignment at the coupling can produce “cranking” of the rotor shafts. Thiswill usually produce a 1X and 2X shaft relative vibration component that exists over the

entire speed range of the machine. The vibration may transmit to the casing, but only

shaft relative measurements will reveal the cranking action at slow roll speeds. The 2X



f=

45

component occurs because of opposed high spots reacting in different bearings. The

effect is similar to the 2X generated in a bicycle crank.

Misaligned disk and diaphragm couplings can produce an axial “pumping” action thatresults in axial vibration. This axial forcing is available to excite any rotor system axial

resonant frequencies. It is also possible for the axial vibration to couple into lateral

vibration, showing up in radial vibration measurements. Properly functioning gear couplings are much more axially compliant and less likely produce axial vibration.

Effects of Radial LoadAs mentioned above, radial loads for which the machine is designed (preloads) can be

beneficial because they suppress fluid-induced instability in sleeve bearings by

positioning the rotor at higher eccentricity. (This effect is discussed in greater detail in

the section entitled “Fluid-Induced Instability” and in the corresponding Machine LibraryMalfunction Diagnosis article [15].)

Aside from the positive effect of suppressing fluid-induced instability, radial load can

wreak havoc with many critical parts of a centrifugal pump and be the primary cause of

pump failure. The adverse effects of radial load most commonly include:1. Rub.

Extreme radial load can cause the rotor to wipe bearings and seals or to rub at wear rings (see “Rub” below). A rub at fluid-film bearing can result in metal to metal contact

and wiping of the bearing babbit. A rub at seals or wear rings can open up clearances,

resulting in higher leakage flows and a loss of efficiency.

2. Shortened Bearing Life.Bearings can be damaged by high radial load even in the absence of a rub. Normally

loaded fluid-film bearings have a babbitt temperature of 160°F to 180°F (70°C to 80°C).

Overloading of a fluid-film bearing will produce higher shear forces in the oil, resultingin higher oil and babbitt temperatures. Bearing babbitt will start to creep at 240°F

(115°C) and melt at 260°F (125°C), leading to bearing failure.

Rolling element bearings are also highly sensitive to radial load. Rolling element bearings have finite lifetimes that are a strong function of radial load. The L10 life (the

time that 90% of similar bearings will survive) for a point contact ball bearing goes down

as the third power of the applied load. Thus, load shifting due to misalignment or

sideloading can, by increasing the load, drastically reduce the useful life of a rollingelement bearing.

3. Damaged Seals or Packing.

Mechanical seals and packing are designed to operate within certain limits of shaftdeflection and position. Radial load can deflect shafts and push seals and packing outside

of their design limits. Re-occurring failure of mechanical seals or packing may indicate

excessive shaft deflection due to radial load.

4. Cracked Shafts.

Radial load is cited by more than one reference as a common reason for broken pump

shafts [1, 2]. Radial loads that deflect shafts beyond their design limits can create highreversal stresses. These stresses can fatigue the shaft and cause it to break.



f=

46

5. Shortened Coupling Life.

Incorrect alignment can shorten coupling life by producing extreme heat in elastomer couplings. Also, gear couplings can experience extreme wear and dry element couplings

can experience high fatigue.



f=

47

Axial Rub

Radial Rub

Combination Rub

Figure 3.10 Rub can occur in the radial direction,

the axial direction, or a combination of both.

Clearance

Center

Average

Position

Clearance

Boundary

Dynamic

Position

(Orbit)

Rub

Figure 3.11 Partial Rub. The dynamic motion of

the shaft centerline (orbit) is added to the average

shaft centerline position. When the total

displacement exceeds the allowable clearance,

rub occurs.

Circular

Orbit

Figure 3.12 Full Annular Rub (1X, forward).

The rotor maintains contact with the clearance

surface of the stator throughout the entire

vibration cycle.

Rub

Centrifugal pumps are susceptible to rubs in much the same way as other types of rotating machines. Beyond the similarities, though, there is a key difference. The

subsynchronous vibration that can accompany rub in other types of rotating machines is

seldom, if ever, seen in pumps.Aside from this key difference the discussion of pump rubs basically mirrors the

general topic. The reader who desires more discussion on the topic of rub can refer to the

corresponding Machine Library Malfunction Diagnosis article [13]. The main pointsconcerning rub will be summarized below and particulars about pumps will be noted.

Also noted will be the pump components that are especially vulnerable to damage by rub.

Definition of Rub

Rub is an undesired contact between a rotating and stationary part. Normally, bearings serve the purpose of separating

the rotating part of a machine from thestationary part. When machine parts moveto a position where contact can occur at

places other than the bearings, the parts

“rub” on each other, hence its name.

Machines with rub can suffer seriousdamage because the rubs create stresses for

which the machine was not designed.

Rub contact can occur in the radialdirection or in the axial direction, or in a

combination of both (Figure 3.10). This

section will be concerned primarily withradial rub.

Rubs can force contact between the

rotor and stator that lasts for just a fractionof the total time required to complete a full vibration cycle or it can force contact that



f=

48

lasts for the entire duration of the vibration cycle. Thus the duration, or dwell time, of the

rub contact can vary significantly. Rubs that force contact between rotor and stator for

only part of the vibration cycle are referred to as partial rubs (Figure 3.11). Rubs thatforce contact between rotor and stator throughout the entire vibration cycle are referred to

as full annular rubs (Figure 3.12). Partial rubs can be further separated into Normal-

Tight and Normal-Loose while full annular rubs can be separated into forward andreverse. These subgroups are explained in reference [13].

Rubs produce significant forces that act on the rotor. The combination of these forces

with the different type of rubs tend to produce unique patterns of vibration characteristics.By observing these patterns, we can identify rub as the malfunction at hand and hopefully

identify its source as well.

These effects and their symptoms will be described below. However, we must first

lay the groundwork by discussing the causes of rub.

Causes of RubMachines are designed from the outset to prevent unwanted contact from taking

place. Thus, for a rub to occur, something else in the machine must have moved out of

the design position (or allowable position range) to some position that results in contact.For this reason, rub is almost always a secondary malfunction. There is usually another

malfunction that is the root cause of the rub.

1. Radial Load

Rub can be caused by radial loads such misalignment or sideload (see “Radial Loads”above). Misalignment can be either internal or external. Internal misalignment affects the

position of the internal parts relative to the rotating shaft. For example, an out of position

seal or diaphragm could cause this. External misalignment affects the position of machines relative to each other. External misalignment can produce unwanted loads on

the rotor system and cause the rotor to move out of normal operating positions inside

either or both machines. Sideloads can have effects similar to misalignment. A heavysideload can deflect a shaft sufficiently to force a rub where clearances are tight (seals,

packing, wear rings) or to create high fatigue stresses which lead to a shaft crack.

2. High Vibration

High vibration produces a large amount of dynamic motion of the rotor inside themachine. This dynamic motion, which is described by an orbit, is in addition to the

operating average shaft centerline position (Figure 3.11). Rub will occur when the

instantaneous position of the rotor exceeds the allowable clearance.Recall from the preceding “Unbalance” section that, among other sources, high

vibration can be caused by hydraulic or mass unbalance. In one case history, hydraulic

unbalance was the root cause of a rub between the impeller vane tips and casing of avertical slurry pump [6].

3. Axial Thrust

The normal flow of liquid through a centrifugal pump creates high axial thrust forces.

Centrifugal pump designs include a wide variety of features and devices to accommodatethese forces [1]. However, the axial thrust balancing within a pump can fail for any

number of reasons or external factors may create axial thrust for which the machine was

not designed. For example, a 90° piping elbow situated too close to the inlet of a double

suction pump can upset flow to the suction and cause one suction eye to receive more



f=

49

flow than the other. The uneven flow between the two suction eyes will create uneven

pressure on the two sides of the impeller with a resulting axial thrust force.

4. Locked SealsSometimes, a floating seal may lock up. If the seal locks up, the rotor may rub on the

seal.

Effects of Rub on the Rotor System

The general concept that rub is an unintended contact between rotating and stationary

parts is quite straightforward. However, the effects of rub upon the rotor system and themechanisms by which these effects produce the vibrations characteristic of rub are fairly

complicated. These mechanisms and their effect on vibration are discussed at length in

reference [13]. The reader who wishes greater depth of discussion can refer to thatarticle.

Vibration Characteristics of Rub

1. Changes in 1X Vibration

Steady State: At steady state, rub will

produce changes in 1X vibration becauseof rotational energy transfer to lateral

vibration energy and because of changes

in stiffness. A light rub is more likely toincrease 1X vibration amplitude, while

heavy rub can severely constrain the rotor

and reduce 1X vibration. Heavy rub canalso result in more energy transfer to the

machine casing, causing an increase of

1X vibration on the casing.If the rotor system is operating near a

balance resonance, the 1X vibrationamplitude can increase or decrease

depending on which side of the resonancethe machine is operating at. The

resonance is moved to a different speed

because of the rub-induced stiffnesschange.

Thermal bow effects due to rub can

produce changes in the amplitude or phase of the 1X response vector.

Occasionally, these changes can be

continuous over time (Figure 3.13). Startup and Shutdown: During

startup and shutdown, rub-induced

changes of the balance resonance speed

can produce changes in observed behavior through the resonance (Figure

3.14). For this reason, it is always good to

have reference startup and shutdown

0° 270°

180°90°

*

**

*

*

**

**

**

** *

**

*

*

***

*

*

*

**

*

**

******

*** * *

*****

***

* **

* **

****

***

*****

*

*

*

* *

**

** *

**

*

** *

*

*

****

*

* * * ** * **

***

**

*

******

03:30:08

03:34:02

3.0 mil pp Full Scale

* **

Figure 3.13 Steady synchronous rub at an oil seal in

a thrust bearing box. The 1X response changescontinuously, completing one revolution of the polar

plot in about four minutes.

Speed

P h a s e

A

m p

Rub

Startup

NoRub

Rub

Shutdown

Figure 3.14 Startup and shutdown partial

(Normal-Tight) rub behavior. 1X Bode plot

comparisons of experimental startup and

shutdown data for no rub, light radial rub during

startup, and light radial rub during shutdown.



f=

50

Rub

Figure 3.15 Full spectrum cascade plot showing a radial rub during startup. The first balance

resonance occurs at about 1800 rpm (horizontal red line). The rub initiates at about 4400 rpm during

the higher vibration associated with the second balance resonance (blue line). Note that the rub

vibration tracks the 1/2X order line and has significant reverse components. Also, at the onset of

1/2X, the rotor speed is twice the rub-modified first balance resonance frequency at about 2200 cpm.

The inset displays eight shaft revolutions of the direct orbit at 4400 rpm. Note the locked Keyphasor

dots that indicate that the vibration frequency is a pure integer ratio (1/2X in this case).

Bode and polar plots available for reference.

2. Subsynchronous Vibration

Subsynchronous vibration, if present at all, will usually be small due to the heavydamping of the pumped liquid. Subsynchronous vibration may not be present because

the requirements for ½X are that running speed be more than 2 times the rub-modified

natural frequency, more than 3 times the same rub-modified natural frequency for 1/3Xand so forth for each subharmonic. Thus the rotative speed must be high relative to the

critical speed (balance resonance) and this is not often the case. If this requirement is met

and subsynchronous vibration is present, then amplitudes are usually low because of theheavy damping provided by the pumped liquid.

If a rub does produce subsynchronous vibration, it will follow the rules laid out in

reference [13]. These subsynchronous frequency components will be pure integral

fractions of running speed and they will follow the rules for Normal-Tight and Normal-

Loose partial rubs. The full spectrum in Figure 3.15 shows subsynchronous behavior



f=

51

1/3X

2/3X

Figure 3.16 Full spectrum of rub showing

harmonics. Rub is producing 1/3X, with a

harmonic at 2/3X. The 1X line probably contains

both 1X rotor and a harmonic of the 1/3X. 4/3X,

5/3X, etc., supersynchronous harmonics are also

visible. Note that significant reverse components

exist for many of the harmonics.

typical of a partial rub. The reverse frequency components are strong indicators of a

partial rub.

Note that it is not possible to absolutely verify that a vibration frequency is a pureinteger ratio by using spectrum. There is always some uncertainty in the displayed

frequency on a spectrum plot because of the limited resolution of the spectrum. A direct

(unfiltered) orbit with Keyphasor dot display should be used to verify the integer relationship.

3. Supersynchronous and Reverse Precession Vibration

Rubs, particularly partial rubs, can produce sharp changes in the rotor

trajectory as the rotor rebounds from the

contact surface. Sharp changes in direction

will produce harmonic frequencies onspectrum plots. For example, if the

vibration frequency is predominately 1X,

then it is possible to see 2X, 3X, etc.

harmonics in the spectrum (Figure 3.16).In the unlikely event that the predominant

rub-induced vibration frequency is 1/2X, itis possible to see 1X (as a mixture of

normal 1X rotor response and the harmonic

of 1/2X), 3/2X, 2X, 5/2X, etc.

Reverse components are also often present in full spectrum plots. Because the

rub usually involves tangential friction

forces in the reverse direction, fullspectrum plots with supersynchronous

vibration will show significant reverse precession components at the supersynchronous

frequency.

4. Average Shaft Centerline Position Changes

Rub can produce a dramatic change in the trajectory, or orbit of the rotor. Because of

the change, the average shaft centerline position can change. This can be a very

pronounced effect for light rub, less so for heavy rub. Sudden changes in average shaftcenterline position during startup, shutdown, or steady state operation can be

symptomatic of a rub.

5. OrbitsDirect (unfiltered) orbits should be examined and correlated with any other unusual

activity that may be taking place in the machine. Changes in direct orbit shape should be

noted. For example, the sharp changes in trajectory produced by rubs will be apparent inthe orbit shape.

Very importantly, only direct orbits with Keyphasor dots can be used to verify that

subsynchronous vibration is a pure integer ratio. On an orbit, the number of displayedKeyphasor dots yields the denominator of a frequency ratio. For example, two Keyphasor

dots could indicate 1/2X, 3/2X, 5/2X and so on. If the frequency really is locked to an

integer ratio, then the Keyphasor dots will remain locked in place through subsequent



f=

52

vibration cycles. If they move in position steadily, then the frequency is not a pure integer

ratio.

Vibration that consists of a mixture of 1X and rub-induced 1/2X can produce orbitswith complicated shapes (Figure 3.15). This orbit shows the path of the shaft centerline

for eight shaft revolutions. The two stationary sets of Keyphasor dots show the vibration

to be pure 1/2X.6. Loss of Efficiency

Rub can cause extreme wear of contacting parts. Seals and wear rings can be

especially vulnerable, and, because machine efficiency often depends on tight clearances,wear at these interfaces will usually result in degraded operating efficiency.

Machines with a significant loss of efficiency should be carefully inspected for

evidence of wipes at seals, bearings and wear rings. Look for discoloration of parts due to

high temperature, scratched or smeared bearing babbitt, and damaged turbine blades,compressor blades, or pump impellers.



f=

53

Shaft Crack

Centrifugal pumps are vulnerable to cracked shafts for a variety of reasons. Thereferences cited below contained several examples of pumps that have failed due to

cracked shafts [3, 5].

The problems created by cracked or totally broken pump shafts are not difficult toimagine. The consequences of a broken shaft can range from inconvenient and costly to

catastrophic and dangerous depending on the circumstances under which the failure

occurs.

Definition of Shaft Crack

A shaft crack can be thought of as a slowly growing fracture of the rotor. If undetected in an operating machine, a crack (also called a fatigue crack) will grow over

time until the remaining, reduced cross section of the rotor is unable to withstand the

static or dynamic loads that are applied to it. When this happens, the remaining rotor section will fail in a fast brittle fracture mode. The sudden failure will release the large

amount of energy that is stored in the rotating system, and the rotor will fly apart. Shaftfractures have caused machine parts to penetrate the machine casing and even penetrate building walls. Damage due to this kind of failure is catastrophic and can cause serious

injury or death to anyone unfortunate enough to be standing near the machine at the

moment of failure. Obviously, shaft crack detection is a very serious matter, and

machines that are suspected of having a crack must be treated with the utmost respect.Shaft cracks begin in regions of high local stress. Shafts are subjected to large-scale

stresses due to static or dynamic bending and torsional twisting, static radial loads, or

residual stresses from heat treatment, welding, or machining operations. These larger-scale stresses can be concentrated by geometric factors such as step changes in shaft

diameter, shrink fits, keyways, drilled holes, or other discontinuities.

Further stress concentration can occur at the microstructure level where surfacemachining imperfections, chemical surface damage, or material discontinuities (such as

produced by slag inclusions or chemical impurities) can produce high, local stress

concentrations. All of these stresses combine to produce a local stress field that changeswith time (i.e., with shaft rotation). The end result can produce a small local region where

stresses exceed the maximum that the material can withstand, and a microcrack will form

in the material. Because shaft bending tends to produce the highest stresses at the outer

surface, shaft cracks usually, but not always, start at or near the outer surface. Sometimes, because of chemical or other processing problems in the rotor billet, a microcrack may

exist inside the shaft before it is put into service.

Shafts, because of their rotation, are subject to periodically changing, or cyclical,

stresses and can fail even though the actual maximum stresses remain well below theyield strength of the shaft material. Failures that occur via cyclical, or reversing, stresses

are referred to as fatigue failures. Shafts can encounter reversing stresses for a number of reasons, Figure 3.17 illustrates just one of these. If a rotor orbits about the center of the

rotor system in pure 1X precession, the stress at any particular outer fiber will see no

change in stress. However, if the rotor is offset from the axis of the rotor system

(typically the case because of a radial process load or gravity), then rotor outer fibers willsee a 1X variation in stress. In addition, a 1X elliptical orbit (which is also typical)



f=

54

Rotor Centered

Rotor Offset With

Elliptical Orbit

1 20

0

T e n s i o n

0

Shaft Revolutions

T e n s i o n

1 20

C o m p

Figure 3.17 An example of outer fiber stress

variation for a rotor in simple bending. When the

rotor moves about the center of the system (top)

in a 1X circular orbit, the stress is constant. When

the rotor is displaced from the system center in a

1X elliptical orbit (bottom), the rotor seesvariable stress with a mixture of 1X (from the

displacement) and 2X (from the ellipticity).

Transverse

Crack

Torsion

Crack

Figure 3.18 Transverse and Torsional Cracks.

A transverse crack results from pure bending

stress in the shaft and propagates directly into the

shaft. A crack resulting from pure torsional stress

forms a spiral at 45° to the long axis of the shaft.Most shafts contain a mixture of bending and

torsion stress. The local stress field at the crack

tip, which can be influenced by local geometry,

determines crack propagation direction. The

crack tip propagates perpendicular to the

direction of the maximum local tensile stress.

produces 2X stress cycling. Thus, even

under normal, 1X operation, real rotors live

in a complicated stress environment thatcontains a mixture of 1X and 2X stress

cycling. Any sub- or supersynchronous

vibration that may be present will producean additional complicated pattern of cyclic

stresses in the shaft.

Once initiated, and if sufficiently highcyclic, or alternating, stresses are present,

the crack tip will slowly propagate in a

direction perpendicular to the orientation of

the local maximum tensile stress at thecrack tip. The orientation of this stress field

is affected by the type of stress (bending or

torsional) and by any geometric factors. If

a rotor is subjected only to simple bendingstresses, then the stress field will be

oriented along the long axis of the rotor,and the crack will propagate directly into

and circumferentially across the rotor section (Figure 2). Pure torsional stress will

produce a crack that is oriented at 45° relative to the long axis. The crack will propagate

into the rotor, but the crack will tend to form a spiral on the shaft surface. In rotor systems, the stress field usually contains a mixture of bending and torsional stress.

Bending stress is usually the dominant

component; thus, the crack will usually propagate into the rotor more or less as a

transverse crack. However, other crack

geometries are possible.As the crack propagates, less and less

material is available to transmit loads in the

rotor shaft, and the local stress across the

remaining shaft material becomes higher and higher. At some point, the section will

become so small that, during the next load

application, the local stress intensity willexceed the fracture toughness of the

material. The fracture toughness is a

measure of the material’s resistance to fastfracture and is a function of the alloy, heat

treatment, the material temperature, and the

rate of loading of the shaft. When thefracture toughness is exceeded, the

remaining section will undergo a fast

brittle fracture, and the rotor will break in

half.



f=

55

a b c

Figure 3.20 2x snapping action of cracked

shaft. The effect of asymmetric rotor stiffness

can be demonstrated with a simply supported

ruler. In the figure, identical downward loads (red

arrows) are applied to a ruler in three positions.The responses are shown as blue arrows. The

ruler has minimum stiffness and largest

deflection at position (a). It has maximum

stiffness and smallest deflection at position (b).

At intermediate position (c), the ruler has a

perpendicular (quadrature) component of

deflection. If the ruler is rotating with a similar

unidirectional load, a snapping motion will be

seen two times per revolution (2X).

Rotor shafts are usually manufactured out of materials with high fracture toughness.

Rotor shaft cracks have exceeded 90% of the shaft cross sectional area before final

fracture, although certainly one should not depend on this. It is not an easy matter todetermine crack size in a rotating machine, and any machine suspected of having a shaft

crack should be shut down as soon as possible.

Effects of Shaft Crack upon Rotor System

1. Reduction of Shaft Stiffness

Shaft cracks, like other malfunctions, have their own unique effects upon a rotor system. The first of these is the overall reduction of shaft stiffness. This occurs in a

relatively straightforward manner and can be visualized by comparing the stiffness of a

thin shaft with that of a thicker shaft. Given that all other things are equal, the thinner shaft will bend more easily than the thicker one because it has less stiffness, i.e., less

cross sectional area moment of inertia. A crack has the same effect because it reduces the

cross sectional area moment of inertia of the shaft. Less cross sectional area moment of inertia, or stiffness, means that the shaft will now show greater deflection in response to

the forces which act on the rotor. This leads to a change in rotor bow and 1X behavior.

2. Asymmetric Shaft StiffnessAnother effect of cracks upon pump shafts is asymmetric stiffness. Asymmetric

stiffness means that the shaft is stiffer in one direction than in another. Considering the

fact that cracks generally start at the surface and work inward, stiffness will be reduced

more in the direction of crack growth than in the direction perpendicular to crack growth(Figure 3.19).

The significance of crack-induced asymmetric stiffness is that it can produce a

Lower

Stiffness

Higher

Stiffness

CrackRemaining

Shaft Section

Figure 3.19 Asymmetric stiffness of a cracked

shaft. As a crack grows, the remaining shaft

section shape becomes asymmetric. This

produces an asymmetric rotor shaft stiffness thatrotates with the rotor.



f=

56

1X Uncomp

1 Mar

1800 RPMGenerator Outboard X

Time

A m p l i t u d e

P h a s e L a g

1 Apr 1 May 1 Jun 1 Jul 1 Aug

Figure 3.21 Example 1X APHT plot showing

possible 1X vibration changes due to a crack. The

machine runs at a constant speed, and amplitude

and phase change slowly over time. Near the end,

shaft stiffness will drop rapidly as the crack

weakens the section (red arrow), and vibrationamplitude will increase rapidly. The details of

each machine’s behavior are different.

Speed

P h a s e

A m

p

1 2 3 4

HighStiffnessLow

Stiffness

Figure 3.22 As a crack propagates, rotor shaft

stiffness will decrease, and resonant speeds may

move downward. Here, the changes in 1X

amplitude and phase that could occur are shown

for four different operating speeds.

snapping action of the rotor whenever a strong sideload is present. Since sideloads are

usually present in centrifugal pumps to one degree on another (refer to “Casing”), so is

the snapping effect associated with shaft crack. The snapping action produced bysideload acting on a cracked shaft occurs twice per shaft revolution and, hence, shows up

as a 2X frequency component (Figure 3.20).

Both of these effects are explained in more detail in the corresponding MachineLibrary Release 2 Malfunction Diagnosis article and the reader who desires fuller

explanation of these effects can refer there. These effects help explain the vibration

characteristics that follow in the next section.

Vibration Characteristics of Shaft Crack

1. The First Rule of Crack Detection (1X)The first rule has to do with 1X filtered

vibration: If a rotor is cracked, it is very

likely to be bowed. And that bow is likely tochange over time. A change in rotor bow

will change the effective location and

magnitude of the heavy spot, which will produce a change in 1X rotor response.

Thus, continuous changes in 1X amplitude

and/or phase are the best primary indicator

of a shaft crack. As the crack grows and theassociated bow develops, 1X amplitude and

phase will change in such a way as to

produce a non-repeating pattern on a Bodeor polar plot over time (Figure 3.21). The

time scale of this change can range from

months to weeks in the early stages of crack growth, to weeks to days as the rotor begins

to seriously weaken, and to hours as the

rotor nears catastrophic failure.

As failure nears, 1X vibration amplitudewill usually increase rapidly. At this point,

1X vibration is likely to be the dominant

source of vibration in the system, so overalldirect vibration will also increase rapidly.

Thus, steady increases in unfiltered peak-to-

peak vibration over time should be takenvery seriously and investigated.

This increase in 1X vibration is not anabsolute - it can also decrease (Figure 3.22).

As a crack propagates across the shaft,reduced stiffness may shift balance

resonances downward to a lower speed.

Depending on the mode shape and thelocation of the crack, some resonances (or

modes) of a rotor could be affected more



f=

57

Vertical

Frequency

Relationship

R o t o r S p e e d ( r p m )

Precession Frequency (kcpm)

Figure 3.23 Full spectrum cascade startup data from a rotor with a

shaft cross-section asymmetry and a unidirectional radial load. A

significant 2X component appears when the running speed (left axis)

reaches ½ of the balance resonance speed near 3600 cpm. Note that a

reverse 2X component is also visible but is smaller than the forward

component. Thus, the 2X component is forward and elliptical.

than others. Also, rotor mode shapes may change, depending on the location of the crack

and how it effects the stiffness distribution of the shaft.

Because of the changing bow of the rotor, the amplitude and/or phase of the 1Xfiltered slow roll vector is also likely to change as the crack propagates. Slow roll vectors

should be compared to historical slow roll data.

Occasionally, a diagnostician may encounter a machine with a “balance problem.”Perhaps the machine had no history of such a problem before. Changing rotor bow due to

growth of a shaft crack will change the location and magnitude of the effective heavy

spot of the rotor. If this happens, a previous balance correction may soon be renderedineffective, and the 1X vibration will increase again. If the root problem is a shaft crack,

repeated attempts to balance the machine will not solve the problem.

Changes in 1X rotor behavior in resonances are an indication that something has

changed in the rotor system. A significant downward shift in a balance resonance speed isa clear indication that the stiffness of the rotor system has decreased. One then has to ask

why this has happened. A weakening shaft due to a crack is a possibility.

1X vibration is usually very sensitive to the presence of a shaft crack because of the

relationship between crack growth and rotor bow, and it can provide significant earlywarning of a crack.

2. The Second Rule of Crack Detection (2X)The second rule has to do with 2X filtered vibration: If a rotor with a crack has a

steady, unidirectional radial load, then a strong 2X response may appear when the rotor

is turning at half of a

balance resonance speed .In addition, this 2X

component is likely to be

predominately forward(although it may be

elliptical). The 2X

snapping action of therotor produces lateral and

torsional impulses in the

system. Because these

impulses occur twice per revolution, the rotor will

respond at the 2X

frequency. If a resonanceexists at twice running

speed, then the 2X

vibration will beamplified. This forms a

vertical relationship on a

half or full spectrumcascade plot (Figure 3.23).

Note that a rotor could

pass through such a 2X

speed relationship during



f=

58

0°

**

*

**

**

**

*

**

**

*

270°

180°

90°

501119084840

38353231

3031

2934

26674359

5541

6998

7101

7146

7193

6034

*

*

*

*

**

*

*

*

* **

* *

**

**

* *

*

*

**

*

*

***

*

*

180°

90°

270°

0°5961

69476987

7029

7068

7107

7127

71947171

7149 1440

12001257

1059

1039

1560

1599

1638

1657

1676

1715

17561853

8752170

48685008

425

525

3236776

R O T N 19 June 28 June

4 mils pp Full Scale

2X Filtered Data Note Phase Change in Leading Direction

Figure 3.24 2X filtered polar plots showing data from two startups of a machine with a rotor crack.

The startup on 28 June shows drastically different behavior than the earlier startup on 19 June. On the

later startup 2X amplitude is larger, and some unusual, leading phase behavior is visible.

0

90

180

270

360

10.0

8.0

6.0

4.0

2.0

0.01SEP 25SEP 19OCT 12NOV 6DEC

SPEED: 1187 rpm2X Filtered

P h a s e L a g

A m p l i t u d e

Figure 3.25 2X filtered APHT plot of a ReactorCoolant Pump with a crack. The pump operated

at a constant speed of 1187 rpm. As the shaft

weakened, the reduced shaft spring stiffness

caused a resonance that was originally above

twice running speed to move down and pass

through twice running speed. Note that, when the

machine was shut down, the 2X amplitude was

decreasing.

startup or shutdown, or a rotor could normally operate at half of a resonance speed.

Obviously, it is less likely that such a relationship would exist at normal operating speed.

For that reason, and because of the additional requirement that a unidirectional radial load be present, a crack may or may not produce significant 2X vibration at running speed.

Experience with other machines types has shown that 2X vibration does not appear

when operating at design speed in about 75% of shaft cracks. However, this is one of theinstances where a common machine malfunction can manifest itself differently in

centrifugal pumps than in other types of machines. While 2X is not always present for

the reasons given in the preceding

paragraph, 2X vibration is about as commonin centrifugal pumps as are changes in 1Xvibration. This is because centrifugal pumps

usually experience sideloading which, along

with asymmetric shaft stiffness, creates thetwice per revolution snapping action that is

responsible for 2X vibration.

Like 1X vibration, 2X vibrationamplitude and/or phase can change as the

crack propagates through the rotor shaft.

Startup and shutdown 2X Bode and polar

plots should be examined for any evidenceof change (Figure 3.24). Also, 2X amplitude

and phase should be trended during steady

state operation. In one case, a reactor coolant pump developed a crack while the pump was

operating at a constant speed. As the crack

propagated, the rotor shaft stiffness droppedso much that a resonance that originally



f=

59

existed above twice the operating speed moved down in frequency and passed completely

through the 2X frequency before the pump was finally shut down (Figure 3.25).

Other Malfunctions With Similar Symptoms

1. 1X Behavior

Many other malfunctions can produce a change in 1X rotor response.A loose bearing support or soft foot can cause a change in 1X vibration. Usually, but

not always, this is manifested as an increase in 1X vibration amplitude. This is the kind of

malfunction that could develop over time with a slow increase in 1X vibration. Because itmimics the behavior of a shaft crack, it can be very difficult to determine the root cause.

If casing measurements are available, an increase in casing vibration with little or no

increase in shaft relative might suggest a soft foot problem, while an increase in shaft

relative with little increase in casing might suggest a crack. But, there are no firm ruleshere.

Thermal growth and subsequent changes in alignment can affect the rotor bearing

stiffness and produce changes in 1X vibration. Thermal bow of a rotor such as a

generator can also produce a similar change, as could an alignment change. Thesechanges in vibration should stabilize once the machine reaches thermal equilibrium at

steady speed and load.Rub can cause changes in both 1X and 2X vibration. These changes can be sudden,

occurring at operating speed, or the changes can show up as changes in transient behavior

during startup or shutdown. Rub can disappear if the part in contact wear away (this can

happen in seals). Or, if the rub is severe, rub contact may be maintained for aconsiderable time. However, rub is not as likely to produce a steadily increasing 1X

vibration level over a long period of time.

A loose rotating part can produce changes in 1X response. If a part moves to adifferent angular or axial position on the rotor, the resulting total unbalance of the rotor is

likely to change, and the 1X amplitude and/or phase will change accordingly. Loose parts

can shift occasionally, producing stepwise changes in 1X response, or they can shiftcontinuously, producing a continuously changing response. Continuously moving parts

will tend to produce a cyclic, repeating behavior on a polar or APHT plot. A loose part is

not likely to produce a steady, long-term increase in 1X vibration amplitude.

Clogged debris in an impeller can produce significant differences in heavy spotlocation in a machine. This will produce corresponding changes in 1X vibration response

and cause a machine to go out of balance.

A locked gear coupling can also produce a sudden step change in 1X vibration.The key to crack identification is to realize that a developing crack is likely to

produce a steady and accelerating increase in 1X vibration amplitude over time as the

shaft stiffness decreases. While some malfunctions will produce periodic changes in 1Xvibration amplitude and/or phase, shaft cracks will tend to produce non-repeating patterns

on polar and APHT trend plots, with the 1X amplitude trending to ever higher levels.

2. 2X Behavior Nonlinearities in rotor system stiffness can cause harmonics (2X, 3X, etc.) of running

speed to appear in spectra. Nonlinear stiffness can be caused by high eccentricity ratios in

fluid-film bearings or by rub impacting. Also, coupling problems can produce 2X

vibration.



f=

60

Pt. B Pt. C Impeller

Bearing Journals

Pump Shaft

Sideload

C

B

Fig. 3.26 Typical Points of Shaft Crack on Overhung Pumps. Shaft deflects under the

heavy hydraulic sideload. High deflection causes the shaft to undergo one cycle of alternating

tensile and compressive stress per each shaft rotation. These fatigue stresses are maximum at

the shaft surface and are further concentrated (or multiplied) at the step changes in diameter at

shoulders located at points B and C. Cracks can initiate in these regions of high fatigue stress.

(After figure from reference [7].)

If any source of 2X vibration exists in a machine, it will be available to excite a

resonance at half of a balance resonance speed. Thus, the presence of 2X at half a

resonance, while suspicious, is not in and of itself confirmation of a crack.

Causes of Shaft Crack in PumpsRecall that cracks occur in regions of high localized stress. Pump shafts are subject to

all the same factors that create and concentrate stress in other machine types. In addition,

pump shafts are equally, if not more, subject to high fatigue stresses that result in fatigue

failure. Some of the common sources of stress are listed below.

1. Stress Concentrators

Step changes in diameter such as shoulders will concentrate stress. Overhung pump

shafts often have shoulders situated in the region where the bending stress fromsideloading is high (Figure 3.26, Pts. B and C). Pump designers will calculate the

maximum stress expected in operation and try to minimize it with fillets and other good

design practices. However, actual operating conditions may present higher stresses thosefor which the pump shaft was designed. For example, one reference [7] saw “…several

pumps break shafts at point C or B because the pump was designed to run with packing

for support on long overhangs (large C & B dimension) and the pump was later changedto run with mechanical seals, i.e., no support. The seal performance was poor also

because of excessive deflection.” (Figure 3.26)

2. Residual Stress

Residual stresses may be leftover from the manufacturing process or may be

unwittingly created by well-intentioned maintenance practices. For example, a high- pressure boiler feedpump shaft that bent (for reasons not stated) was straightened by

peening with a blunt nose chisel in order to return it to service. The shaft developed a

crack over the course of about a year which became evident when the stuffing boxes

began to leak excessively (as rotor bow increased – recall the first rule of shaft crack).While not a foregone conclusion that the peening alone caused the shaft crack, it is quite



f=

61

possible a main contributor since the resulting compressive stresses may have exceeded

the maximum allowable [3].

3. Radial Loads

Sideloads: As noted previously, centrifugal pumps are usually subjected to high

sideloads which increase as the pump is operated further away from its BEP. Pumps are

often operated further off of design capacity than designers originally intended. This cansubject pump shafts to very high levels of fatigue stress.

Misalignment: Misalignment is also a common of source of excessive radial load

resulting in high fatigue stress [1].



f=

62

SpringForce

Radial

Load

Pressure

WedgeTangential

Force

Figure 3.27. The circulating fluid in a bearing

or seal forms a pressure wedge when the rotor

is displaced from the center (left). This reaction

force can be separated into a tangential force (red)

and a spring force (green). The spring force tries

to move the rotor back toward the center of the

bearing, but the tangential force tries to move the

rotor tangentially. The tangential force is

ultimately responsible for destabilizing the rotor.

Fluid-Induced Instability

DefinitionFluid-induced instability is a large amplitude, usually subsynchronous vibration of a

rotor that is caused by rotor interaction with a surrounding fluid. The term “instability” is

somewhat of a misnomer. When a rotor operates in fluid-induced instability, it is actuallyoperating in a stable limit cycle of high vibration. But the rotor is unstable in the sense

that it is operating outside desired operational limits.

The large amplitude, subsynchronous vibration can cause rotor-to-stator rubs on seals, bearings, impellers, or other rotor parts. The vibration can also produce large-amplitude

alternating stresses in the rotor, creating a fatigue environment that could result in a shaft

crack. In addition, the bearing surface is subject to alternating stresses that can lead to

fatigue failure of the babbitt.Fluid-induced instability is a potentially damaging operating condition that should be

avoided.

Cause of Fluid-Induced InstabilityWhen a fluid, either liquid or gas, is trapped in a gap between two, concentric

cylinders, and one is rotating relative to theother, the fluid is set into motion around

the gap (Figure 3.27). This situation exists

in fully lubricated (360° lubricated) fluid-film bearings, in seals, around impellers in

pumps, or when any part of a rotor is

completely surrounded by fluid trapped

between the rotor and the stator. In thissection, we will talk primarily about fluid-

film bearings of basic cylindrical shape.However, it should be understood thateverything written here about bearings also

applies to seals, pump impellers, and any

other region in a machine where a liquid or gas is trapped in a small clearance between

a rotor and a stator.

When a rotor moves away from the

center of a bearing, the converging fluidforms a pressure wedge (Figure 3.27, left).

The pressure profile creates a force that can

be separated into two components. A direct component or spring force, F s, exists that acts like a spring and points back toward the

center of the bearing:

r K F s = (1)

where K is the effective spring constant of the bearing at that eccentricity ratio, and r is

the distance from the center of the bearing.



f=

63

At the same time, a quadrature component acts in a tangential direction in the same

sense as rotor rotation. It turns out that this tangential force, F t , is a function of bearing

damping, fluid circulation, rotor speed, and distance from the equilibrium position:

r jD F t λ = (2)

where D is the bearing (or seal, impeller, etc.) damping, λ (lambda) is the Fluid

Circumferential Average Velocity Ratio, Ω (capital omega) is the angular velocity of the

rotor (the speed of the rotor in radians/sec), and r is the distance from the center of the

bearing.

The j is 1− . Practically, all that means is that the action occurs at 90° relative to thespring force, F s, in the direction of rotor rotation (Figure 3.27, right). (See reference [16]

for much more detail.)

What is λ ? Put most simply, λ is a measure of the amount of fluid circulation in the

bearing. It is defined as the ratio of the average angular velocity of the fluid to the angular

velocity of the rotor. For a plain cylindrical, fully lubricated (360° lubricated) bearing λ is

typically a little under ½, around 0.49 or so. But the value of λ can be influenced by thegeometry of the bearing, the rate of end leakage out of the bearing, the eccentricity ratio

of the rotor in the bearing, and the presence of any pre- or antiswirling that may exist inthe fluid.

Note that the strength of the tangential force depends not only on the rotation speed,

Ω , but also on the strength of fluid circulation around the rotor (λ ). It is much stronger (λ is much higher) when the rotor is surrounded with fluid (the fluid-film bearing is fully, or

360° lubricated). Properly loaded fluid-film bearings are normally only partially

lubricated, and λ is usually small. Thus, properly loaded bearings are unlikely to be a

source of very large tangential forces unless the bearing becomes flooded with an excessof lubricant. Note that fluid-film bearings can become unloaded , for example because of

misalignment, transition to fully lubricated operation, and generate high tangential forces.The spring force, F s, acts to stabilize the rotor because it pushes the rotor back toward

the center of the bearing. However, the tangential force, F t , acts to destabilize the rotor by pushing it at a right angle (i.e., tangentially) to the bearing center. If conditions are

right, the tangential force will drive the rotor in a large amplitude, forward, circular orbit

constrained only by the stiffness of the surrounding stationary cylinder (e.g., bearing,seal, pump casing). When this occurs, the rotor is undergoing a fluid-induced instability

(whirl or whip).

Reference [15] explains these conditions in detail and describes the difference between whirl and whip. Additional detail on this subject is beyond the scope of this

paper. The reader who desires more detail is referred to that article.

At this point, the primary concept to note is that the “ingredients” for fluid-inducedinstability can exist in several regions of a centrifugal pump. This includes impellers, therotors in magnetic drive and canned motor pumps, as well as in bearings and seals.



f=

64

Frequency (kcpm)

R o t o

r S p e e d ( r p m )

W h i r l

Whip

High Eccentricity

Natural Frequency

Low Eccentricity

Natural Frequency

T h r e s h o l d

o f S t a b i l i t y

Figure 3.28 Full spectrum cascade plot of a rotor system startup. The rotor system starts into fluid-

induced instability (in whirl) at about 2400 rpm, the Threshold of Stability. At this time, subsynchronous,

forward precession begins at a frequency near 0.475X. The initial whirl frequency is about 1300 cpm, which

is the low eccentricity natural frequency of the rotor system. As speed increases, the whirl orbit becomes

larger, the bearing becomes stiffer, and the rotor system natural frequency shifts to a higher frequency.

Thus, the whirl tracks at a sub multiple of running speed. At about 2900 rpm, the high 1X rotor vibration

associated with a balance resonance causes the rotor to operate at a high dynamic eccentricity ratio. The

resulting higher bearing stiffness pushes the Threshold of Stability temporarily above running speed, andthe fluid-induced instability disappears. After the resonance, 1X vibration declines, the orbit diameter

decreases, the bearings stiffness decreases, and the Threshold of Stability once again falls below running

speed; thus the fluid-induced instability reappears. When the rotor dynamic motion reaches high

eccentricity, the rotor shaft becomes the weakest spring in the system, and the instability frequency locks in

to the high eccentricity natural frequency in whip. The orbit inset shows the orbit of the rotor inside the

bearing in whirl, and the magenta circle shows the approximate bearing boundary. At this dynamic

eccentricity ratio (about 0.6), the bearing controls the spring stiffness of the rotor system (see Figure 3). The

pair of Keyphasor dots are shifting slowly in a direction opposite to rotation. This indicates that the

fre uenc of vibration is a little less than 1/2X.

Vibration Characteristics of Fluid-Induced Instability

1. Subsynchronous VibrationThe primary symptom of fluid-induced instability is forward, subsynchronous

vibration. The frequency of the subsynchronous vibration due to oil whirl is usually less

than 0.5X (Figures 3.28 and 3.29). For pumping whirl (whirl originating in the pumped

liquid surrounding the impeller), this frequency can occur in the range of 0.7X to 0.9X



f=

65

Frequency (kcpm)

R o t o r S p e e d ( r p m )

Whip

First

Balance

Resonance

Figure 3.29 A rotor system can enter fluid-induced instability whip

directly without encountering whirl first. In this case, the rotor operates

at a high eccentricity ratio within the bearing, and the bearing stiffness is

much higher than the shaft stiffness. The rotor enters whip in a bending

mode that corresponds to the high eccentricity natural frequency. The

first balance resonance for this mode can be seen at approximately 2200

cpm. Harmonics of the whip frequency are also visible. The whip orbit is

also shown inside the magenta bearing clearance. Note the jumble of

Keyphasor dots and the very high dynamic eccentricity ratio of about

0.9. Shaft stiffness is the weak spring (Figure 3); thus, the natural

frequency cannot be modified and the subsynchronous frequency

remains constant.

due to the increased

fluid circumferentialaverage velocity

generated by the

impeller [4], [9]. The

frequency of thesubsynchronous whirl

vibration is related to

the fluid swirling rate,

lambda (λ ), of the fluid

causing the instability.

In whip, however, the

frequency of vibrationwill lock to a rotor

system bending mode

(Figure 3.29). The

subsynchronous whipfrequency can range

from 0.3X to 0.8X or higher depending upon

the fluid circumferential

average velocity ratio

(λ ) of the fluid causing

the problem.

Unlike rub, fluid-

induced instabilityalmost never produces a

pure integer ratiovibration frequency

such as 1/2X, 2/3X, 1/4X, 1/3X, etc. Instead, fluid-induced instability produces irrationalfraction frequencies. However, if the lubricating film between rotor and stator breaks

down or if the large amplitude instability vibration causes a rub elsewhere, then fluid-

induced instability can lock to an integer ratio.The subsynchronous vibration caused by fluid-induced instability is almost purely

forward (Figures 3.28 and 3.29). This is a very useful way to discriminate between rub

and fluid-induced instability as a root cause. Rub tends to produce significant reversecomponents at the subsynchronous frequency.

During a startup or shutdown, whirl due to fluid-induced instability will track running

speed at some sub multiple (Figure 3.28), while whip tends to lock to a constantfrequency (Figure 3.29). As can be seen in the figure, it is possible for whip to suddenly

appear without any whirl.

Fluid-induced instability is always associated with a natural frequency of the rotor

system (usually the lowest mode). Often the balance resonance associated with that modewill appear during startup as 1X vibration (Figure 3.29). However, if the lowest mode of

the rotor is supercritically (over) damped (as can happen with rigid body modes), then the



f=

66

rotor will not have a resonance on that mode, and the 1X vibration associated with the

mode will not be visible during startup. This is the case for the machine in Figure 3.28.

2. OrbitsIf the vibration at the measurement plane is dominated by fluid-induced instability,

then the direct, unfiltered orbit will be predominately forward and circular (Figure 3.28).

Orbits that are filtered to the instability frequency will always be approximately circular and forward.

The behavior of the Keyphasor dots will depend on the relationship of the

subsynchronous frequency to running speed (the Keyphasor trigger frequency). Ingeneral, the number of Keyphasor dots visible is related to the denominator of the nearest

subsynchronous integer ratio. For subsynchronous frequencies near 1/2X, two Keyphasor

dots will be visible. If the subsynchronous frequency is slightly below 1/2X, then theKeyphasor dots will slowly drift in a direction opposite to rotation. If the subsynchronous

frequency is slightly above 1/2X, then the Keyphasor dots will slowly drift in the same

direction as rotation. Vibration near 1/3X will produce a set of three Keyphasor dots in

the orbit that behave in a similar way. Vibration near 2/5X (0.4X) will produce an orbit

with 5 Keyphasor dots.When the subsynchronous vibration is not near an integer ratio, the Keyphasor dots

will tend to form a chaotic pattern consisting of great many dots (Figure 3.29). Note that, under the right circumstances, rub will produce subsynchronous vibration

at a pure integer ratio with locked Keyphasor dots. These dots will not drift around the

orbit with time and will tend to stay in thesame location. Because rub produces

integer ratio subsynchronous vibration

frequencies (such as 1/2X), Keyphasor dots from a subsynchronous rub orbit will

form a locked integer set. This is a very

powerful tool for discriminating between

fluid-induced instability and rub. LockedKeyphasor dots imply rub, while moving

Keyphasor dots imply fluid-induced

instability.Whip orbits, because of the lower

subsynchronous frequencies at which it

usually occurs, are more likely to showchaotic Keyphasor dot behavior than

whirl orbits.

If the vibration at the measurement

plane contains a mixture of 1X andsubsynchronous vibration, then the orbit

will be more complex in shape. The

subsynchronous vibration will cause theorbit to continually change shape, but the

motion of the Keyphasor dots (for frequencies close to an integer multiple) will still tend

to migrate in a small circle (Figure 3.30).

Fluid Instability Rub

Figure 3.30 Direct orbits showing a mixture of 1X

and subsynchronous vibration for eight shaft

revolutions. The fluid-induced instability frequency

is slightly less than 1/2X in whirl (the orbit is from a

location some distance from the source), while the

rub frequency is exactly 1/2X. In the instability orbitthe Keyphasor dots slowly migrate against rotation

(black arrows) in a circular path (red), while the rub

orbit dots are locked in place



f=

67

Whip

Low

Speed

High

Speed

Radial

Load

Direction

Figure 3.31 Average shaft centerline plot showing the

transition from stable behavior (black) to fluid-induced

instability whip (blue) inside the bearing for the data

shown in Figure 6. The dashed circle shows the bearing boundary. As the instability develops, the average

eccentricity ratio in the bearing approaches zero.

500

3600

Radial

Load

Direction

Figure 3.32 Normal shaft centerline

plot from a shutdown of a steam

turbine generator. During startup, the

properly loaded shaft centerline

would start at the 500 rpm point and

move up to the right for X to Y

(CCW) rotation. Compare to Fig. 7.

3. Average Shaft Centerline Position

In classic fluid-induced instability, the journal will move about the center of the bearing at a subsynchronous frequency in a forward, circular orbit. As the rotor orbit

grows larger in whirl or whip and begins to move around the bearing clearance, the

average eccentricity ratio will begin to approach zero. That is, the average shaft

centerline position will approach the bearing center (Figure 3.31). Thus, it can be veryuseful to correlate the onset of subsynchronous vibration with movement of the shaft

centerline toward the center of the bearing.

A related issue concerns a potential cause of fluid-induced instability. Machines with

fluid-film bearings are usually designed to operate in a partially lubricated condition at ahigh eccentricity ratio position. The shaft centerline plot of a normal machine has a

typical behavior (Figure 3.32). If a machine becomes misaligned, then one or more

bearings in the machine may become partially unloaded. When this happens, the shaftcenterline operating position will move to an abnormal position closer to the center of the

bearing. Operation near the bearing center is more likely to result in full lubrication of the

rotor journal, causing fluid-induced instability. Thus, the shaft centerline plot can providea clue as to the root cause of the fluid-induced instability that is taking place in themachine.

Corrective Actions for Fluid-Induced Instability

1. Reduction of Fluid Circulation

The fluid circulation is what creates the destabilizing tangential force. λ , the Fluid

Circumferential Average Velocity Ratio, is a measure of the strength of the fluid



f=

68

Figure 3.33 Bearing geometries that break up

fluid circulation in the bearing reduce the value

of λ and promote stability.

Pressurized

Fluid

Pressurized

Fluid

Figure 3.34 Antiswirl injection involves

injection of pressurized fluid tangentially into a

bearing or seal in a direction opposite rotation.

The injected fluid disrupts circulation and greatly

circulation. Anything that acts to disrupt fluid flow around the clearance will help rotor stability.

Control of λ can be difficult for an end user and can be most easily accomplished at

the OEM level. This is commonly done by utilizing bearing geometries that depart fromsimple cylindrical shapes (Figure 3.33). Tilt pad bearings are an example of this. Because

the pads are not continuous, fluid flow is disrupted in the bearing and stability is

enhanced.Antiswirl injection involves injecting working fluid tangentially into the bearing or

seal in a direction opposite to rotation (Figure 3.34). The injected fluid acts to slow down

the overall average fluid angular velocity and reduce λ . This technique has been

successfully applied in both bearings and seals and has proven to be very effective.

2. Proper Loading of Hydrodynamic BearingsFluid-induced instability often originates in hydrodynamic bearings that are

insufficiently loaded. Misalignment can shift the load from one bearing to one or moreother bearings in the machine. The lightly loaded bearing will tend to position the rotor

closer to the center of the bearing.

If a machine that once ran acceptably now exhibits fluid-induced instability, check the shaft centerline plot to see where the rotor is operating in the bearing clearance. If the

rotor is found to be operating in a particular bearing at a low eccentricity ratio while

adjacent bearings are highly loaded, then the external and internal alignment of themachine should be checked. Correct alignment should result in properly loaded bearings.

At the design level, fluid-film bearings in a machine should be designed with an

adequate load. Over designed bearings could result in a fluid-induced instability problem.3. Adjustment of Supply PressureHydrostatic bearings normally operate in a fully lubricated (360° lubricated)

condition. In these types of bearings, the spring stiffness of the bearing, K B, is strongly

influenced by the lubricant delivery pressure in the bearing. Thus, increasing the bearingsupply pressure will increase the rotor system stiffness, K , and may eliminate the

instability.



f=

69

Seals act like hydrostatic bearings. The rotor in the seal area is normally completely

surrounded by the working fluid of the seal. Thus, increasing the seal fluid supply pressure may increase the stiffness of the seal and, if the seal is the source of the

instability, eliminate the instability.

Hydrodynamic bearings, on the other hand, normally operate in a partially lubricated

condition at a relatively high eccentricity ratio. Increasing the lubricant supply pressuremay actually flood the bearing, causing it to operate in a fully lubricated condition. This

is likely to destabilize the rotor system. If a hydrodynamic bearing is suspected of being

the source of the fluid-induced instability, then reducing lubricant supply pressure mayeliminate the flooded condition and stop the instability. Obviously, care must be taken to

avoid reducing the supply pressure to such a low level that causes damage to the bearing.

4. Adjustment of Lube Oil TemperatureFluid viscosity affects both the bearing stiffness, K B, and the bearing damping, D.

Thus, changing the fluid viscosity may have a significant effect on the fluid-induced

instability.

It is difficult to predict ahead of time how changes in oil temperature will affect the

speed at which fluid-induced instability occurs (called the Threshold of Stability). Insome cases, a change in oil supply temperature of only a few degrees has produced

dramatic changes in the fluid-induced instability behavior of the machine. Further explanation of the effect of lube oil temperature on fluid-induced instability can be found

in Reference [15].



f=

70

Structural Resonances

Rotors are not the only parts of a pump and its driver that can resonate, other machine

parts such as casings and brackets can resonate as well. In fact, it is not just machine

parts that can resonate, any object with properties of elasticity (i.e., spring stiffness) and

mass will resonate when excited at its natural frequency. This includes virtually everyobject found in the industrial environments that normally surround centrifugal pumps.

The piping, piping support brackets, the pump pedestal and other support structures,

roofs, walls, floors, etc. – in other words, any surrounding structure will resonate whenexcited at its natural frequency, hence the term structural resonance.

The vibration of a structural resonance can transmit into a pump and damage

bearings, seals, couplings, and other vulnerable pump components. Conversely, vibrationcan originate from within the pump and damage an adjoining structure, although this

probably less likely since the pump will tend to be more sensitive to vibration than

adjoining structures. In either case, it is important that structural resonances, if present, be identified and corrected if believed to be problematic.

Definition of Structural ResonanceResonance is the peaking of the amplitude of vibration that occurs when a periodic

force excites an object at its natural frequency. Anyone who has heard a rattle in their car

has experienced a structural resonance. The offending rattle occurs because some part isloose enough (that is, its stiffness is lowered) so that it vibrates when the frequency of the

exciting forces (for example: engine, vibration of tires on road surface, etc.) match the

natural frequency of the part. That is why the rattle may appear and then disappear withchanges in speed. The rattle is most pronounced when the frequency of the exciting

forces are closest to the natural frequency of the loose part. Recall that the natural

frequency, denoted by ω n, of an object is defined by the equation:

M

K n =ω (1)

where K and M are object’s spring stiffness and mass respectively. This simple

relationship between stiffness and mass explains why stiffening the part by either tightening or reinforcing it silences the rattle. The stiffened part now has a natural

frequency that is beyond the range of the frequency of the exciting forces and is thus

unable to resonate.Machines “live” in a complex vibration environment. Vibrations of a wide range of

frequencies originate from both inside and outside the pump. The frequencies of

vibrations originating from within the pump typically include 1X, but can also includesubsynchronous and supersynchronous frequencies. Vibration external to the pump can

originate within adjacent machines and also from adjoining processes (process liquid

flowing through a pipe can excite vibrations as well). All of these vibrations of variousfrequency and amplitude combine to make up a complex source of excitation.

Even though common sense tells us that vibrations do not stop at some artificial

boundary surrounding the pump but rather will transmit in and out, the interaction

between a pump and its adjoining structures is sometimes overlooked. The interactivity



f=

71

Structural

Resonance

180

240

300

360

60

120

180

10

8

6

4

2

00 1000 2000 3000 4000 5000 6000 7000

P h a s e L a g

( d e g r e e s )

Bode Plot, 1X Compensated

A m p l i t u d e

( m i l s p p )

Figure 3.35. Structural resonance on Startup and Shutdown

Plots. The structural resonance appears as deviations from the

normally expected ampitude and phase lag curve shapes. The

circled portions of the curve show increasing amplitude and phase

lag. However, the deviations could just as easily have been

decreases.

of a pump with its environment means that there are additional sources of vibration that

need to be considered when attempting to solve problems of high vibration.

Vibration Characteristics of Structural Resonance

Startup and Shutdown data: Structural resonance can cause vibration amplitude and phase lag curves on startup and

shutdown plots to deviate from

the normally expected shape.The deviation usually occurs

over a limited frequency range

relating to the structuralresonance. The deviation can

be either an increase or a

decrease in vibration and phase

lag depending on the phaserelationship between the rotor

and structural vibrations

(Figure 3.35).

Steady State data: As with

startup and shutdown plots,

steady state plots can also showan increase or decrease in

vibration amplitude and phase

lag depending on how the

vibrations combine. A trend plot may show a change in

vibration if something

adjoining the pump undergoesa change. Such changes could

include a broken pipe support bracket, change in an adjoining process, etc.

One must make careful note of the fact that these changes are not unique to structuralresonances. Other malfunctions can also cause change in vibration. For example, shaft

crack can also produce changes in 1X vibration amplitude and phase over time.

Effects of Structural ResonancePump seals, bearings, and couplings are typically the parts most affected by high

vibration, including those produced by structural resonance. In one case history [5], two pumps that had operated for several years with only minor problems and annual

maintenance began to experience multiple seal, bearing, and coupling failures and to

require monthly overhauls. The cause of the failures was traced to the pumps’ discharge piping which were found to resonate at 1X the pump running speed. The discharge

piping had been rerouted about the same time that the pumps began to fail. The new pipe

run had left them insufficiently supported (i.e., lower stiffness) and thus able to resonateat a frequency that coincided with the pump speed. Pump reliability was restored to

original levels once vibration in the piping was reduced.



f=

72

Cavitation

Any discussion of centrifugal pump malfunction must include the important subject

of cavitation. Cavitation is one the most common centrifugal pump malfunctions. It is

capable of causing severe performance loss and pump damage resulting in significant

financial impacts to pump owners. Consequently, cavitation is also the focus of muchinvestigation on the part of researchers.

Definition of CavitationThe term cavitation refers to the formation of tiny vapor bubbles, or “cavities”, within

the pumped liquid that subsequently collapse with tremendous force. There may be gas

bubbles of some other dissolved substance in the pumped liquid, such as air, that areexpanding and collapsing along with the vapor bubbles. However, true cavitation refers

to the vaporization and subsequent collapse of the pumped liquid itself.

The vapor bubbles are capable of causing severe damage when they collapse againstthe metal surfaces inside the pump. One reference contained a photograph showing an

impeller vane that had been eroded completely through by cavitation [2].Only liquid handling machines experience cavitation because liquids by nature will

boil into vapor and then condense back into liquid given the right conditions.Compressors do not experience cavitation because the gas they handle already exists in a

vapor state and remains so throughout the entire compression process.

Mechanism of CavitationThe mechanism of cavitation is actually the process of liquid evaporation and

condensation. Thus, if we understand the conditions for evaporation/condensation, then

we will have defined the general physical conditions that cause cavitation.There are two ways to evaporate, or boil, a liquid: 1) increase the temperature of the

liquid to its “boiling” temperature, or 2) decrease the pressure acting upon the liquid to

less than or equal to its vapor pressure. This natural phenomenon is straight out of standard thermodynamic principles that show that evaporation is dependent on bothtemperature and pressure. Anyone who has cooked boiled foods at high elevation has

discovered that more time is required because boiling temperature lowers as the

atmospheric pressure lowers. For example, water at sea level will boil at 100 °C (212 °F)

versus 94 °C (202 °F) at 1524 meters (5000 feet) above sea level. Since condensation is

merely the opposite of evaporation, vapor will condense when 1) its temperature is

lowered below the boiling point, or 2) its pressure is raised above the vapor pressure.

Applying this to centrifugal pumps, we see that cavitation will occur when either the pressure inside the pump drops below the liquid vapor pressure or the temperature of the

liquid inside is raised above its boiling point. While cavitation most often occurs because

of the former, it is also quite possible for increased temperature to cause cavitation.Because the pressure inside a pump is a function of the Net Positive Suction Head

(NPSH), the lack of available NPSH is the primary cause of pump cavitation. (See

section “Net Positive Suction Head” for greater detail about NPSH.)



f=

73

Figure 3.37 Collapse of Vapor Bubbles. Cavitation produces vapor bubbles which can erode metalsurfaces inside the pump when they collapse against those surfaces.

Rotation

Formation of

Vapor Bubbles

Collapsing Vapor

Bubbles

Figure 3.36 Location of vapor bubbles. Cavitation occurs where pressure is

lowest. This is along the trailing side of the impeller vanes.

The location in

the pump wherecavitation will occur

is the point where

pressure is lowest.

Figure 3.36 showsthat this point is

adjacent the trailing

(low pressure) sideof the impeller

vanes.

All liquids havethe potential to

cavitate since all

liquids follow the

principles of

thermodynamics.However, liquids

differ in the severityof cavitation damage

they may cause.

Denser liquids, likewater, cause more damage when their vapor bubbles implode than do less dense liquids,

such as hydrocarbons. Also, liquids with larger differences between liquid and vapor

specific volumes will create larger implosion forces when the vapor cavities collapse.

Effects of CavitationAs it was alluded to earlier, cavitation vapor bubbles do not form and then gently

“pop”. Rather, they form and collapse in a few thousandths of a second. As the bubbles

collapse, they implode with tremendous pressures estimated on the order of 104

atmospheres (Figure 3.37).

The effects of the local shock wave produced by these collapsing bubbles can rangefrom inconsequential to extremely damaging. Cavitation may be little more than an

annoyance due to the severe noise produced (although cavitation can occur without

noise). The presence of noise due to cavitation does not necessarily mean that it iscausing damage to the pump. Some pumps may noisily operate in cavitation for years



f=

74

without failure. The tendency for cavitation to damage a pump depends on impeller

material and design and operating conditions.Of greater concern than noise is the loss of hydraulic efficiency. A cavitating pump

has less liquid flowing through it because the lower density vapor cavities block flow.

Reduction of hydraulic efficiency always accompanies cavitation, whether or not the

losses are significant depends on the amount of cavitation present.At worst, the local shock wave from the implosion of the vapor bubbles can erode

impeller vanes through the removal of material from metal surfaces. The severity of

erosion can vary from surface pitting to holes clear through the vanes. This can occur ina matter of a few weeks. The loss of material on the impeller upsets mass and hydraulic

balance resulting in high vibration that can damage seals and bearings.

Characteristics of Cavitation

1. Reduction in Pump Head

The loss of efficiency described above will be recognizable as a drop in the head

produced. A three percent drop in head has traditionally been used as an indicator of cavitation. However, the onset of cavitation starts before drop in head reaches three

percent. Thus, a pump might be cavitating even without a significant drop in head.

If the pump head has dropped because of cavitation, that does not necessarily mean it

is damaging the pump. Whether or not cavitation will harm a pump depends on severalfactors including the impeller material and the nature of the pumped liquid.

Drop in head will also occur if pump rotative speed is reduced. Changes in speed

should be verified before assuming cavitation is present.

2. Noise

As described above, cavitation may or may not be accompanied by noise. If it is, the

noise tends to be a steady “crackling” noise [1]. This is in contrast to the noise of recirculation (another type of cavitation that is described below) which has been

described as being a random crackling with high-intensity knocks [1].

3. VibrationCavitation increases vibration amplitude over a broad frequency range. The vibration

increases can be high enough to damage seals and bearings.

4. Visual Indicators

Erosion on the low pressure side of the impeller vanes is a sign that cavitation iscaused by insufficient NPSH. Cavitation caused by recirculation will erode other areas of the vanes (this is described in more detail below in “Cavitation caused by

Recirculation”).

Corrective Actions

1. Increase the Available NPSH

Increasing the NPSH provided by the system will raise the pressure in the pumpabove the liquid vapor pressure. The suction side piping should be evaluated for the

presence of bends, elbows or other fittings that might be reducing the pressure at the

pump suction to unacceptably low levels.

2. Decrease the Required NPSH

Another way to prevent pressure in the pump from falling below the liquid vapor

pressure is to reduce the Required NPSH. Required NPSH is a function of the frictionloss experienced by the liquid as it flows from the suction flange to the point of lowest



f=

75

Discharge RecirculationSuction RecirculationFigure 3.38 Suction and Discharge Recirculation. Suction recirculation occurs in the impeller

eye. Discharge recirculation occurs at the discharge tips of impeller vanes.

pressure in the impeller vane passages. Examples of changes that will reduce inlet

friction losses are:Use a pump design with a lower Required NPSH. For example, double suction

pumps generally require less NPSH than a comparable single suction pump because the

double suction eyes provide a larger inlet passage with lower frictional losses.

Use a pump that runs at lower rotative speed. A lower speed pump will have to belarger in order to deliver the same head versus flow of a smaller pump with comparable

performance.

3. Cooling the suctionCavitation can also be prevented by lowering the liquid temperature on the suction

side of the pump. Temperature at the suction must be reduced to the point where the

liquid stays below its boiling temperature when flowing through the impeller vanes.(“Boiling temperature” is pressure dependent just like vapor pressure is temperature

dependent, the boiling temperature of the liquid inside pump is not the same as the

boiling temperature of the liquid at atmospheric pressure.)

Cavitation caused by RecirculationDefinition

The type of cavitation discussed in the preceding section results from insufficient NPSH. However, a malfunction known as recirculation can also cause cavitation. The

distinction between cavitation caused by insufficient NPSH and that caused by

recirculation is important because they have different corrective actions.The term recirculation refers to a reversal of flow within the pump. The normal

direction of flow through the pump is from suction to discharge. However, under certain

conditions liquid will reverse direction and flow toward the suction instead of continuing

to discharge as intended.

Flow reversals create vortices and cavitation occurs at the center of the vortices. Theflow reversal and its associated cavitation occur in two main areas (Figure 3.38).



f=

76

All impellers will recirculate if flow drops below a specific level. The flow at which

recirculation occurs is impeller dependent and cannot be changed without modifying thedesign [1].

Symptoms of Recirculation

1. NoiseLike the cavitation discussed in the preceding section, recirculation can also produce

noise. However, recirculation noise tends to be of greater intensity than the noise from

low-NPSH. It has been described as a random knocking sound, as if a loose bolt or nutwhere being rattled around inside the rotating impeller.

It may be possible to distinguish whether the recirculation is occurring in the suction

or discharge depending where the noise is of highest intensity. Suction recirculationnoise will be most noticeable near the pump suction while discharge recirculation noise

will be louder at the pump discharge.

2. Vibration

Recirculation cavitation can result in increased radial and axial vibration.

3. Pressure PulsationsRecirculation causes a sudden increase in the magnitude of pressure pulsations.

These pulsations are detectable using pressure transducers [1].

4. Visual Indicators

Recirculation cavitation erodes the pressure side of vanes unlike low-NPSH

cavitation which attacks the low pressure side of vanes.The location of erosion indicates whether the recirculation is suction or discharge.

Pitting near impeller eye indicates suction recirculation while pitting near discharge end

of vanes indicates discharge recirculation.

Corrective ActionsSince insufficient flow through the impeller causes recirculation, this flow must be

increased. Two ways to accomplish this are:

1. Increasing pump output.2. Rerouting some of the pumped liquid from discharge back through the suction.

One caution with rerouting liquid back through the pump is that temperature can rise to

unacceptable levels. The mechanical work of the impeller upon the liquid flowingthrough it results in a slight temperature rise. The rerouted liquid can cause heat to

accumulate in the pump thus raising the temperature unacceptably.

If neither of these is acceptable, an additional possibility exists. The susceptibility of an impeller to cavitation damage depends on several factors, one of which is the type of

material used in its construction. If cavitation cannot be eliminated, then switching to an

impeller of harder material presents an additional option.



f=

77

Vane Pass Frequencies

Definition of Vane Pass FrequencyVibrations which occur at a frequency related to the number of impeller vanes, the

number of casing vanes, and pump rotative speed are known as vane pass frequencies.Because single volute pumps have only one vane (i.e., the cutwater - Figure 2.6), vane

pass frequencies are usually an integer multiple of rotative speed where the integer

multiple is the same as the number of impeller vanes.The source of excitation for vane pass frequencies is the interaction between the

cutwater of the pump casing and the nonuniform velocity and pressure distribution of the

liquid exiting the impeller vane passages. The finite vane thickness and slower movingliquid adjacent to the vane surfaces (called a boundary layer ) create variations in the

velocity and pressure of the flow exiting the impeller periphery [4]. In addition, flow

from each impeller vane is forced to make an abrupt change in direction as it passes thecutwater [1]. These variations in velocity and pressure at each vane exit are repeated

around the circumference of the impeller in a pattern that is evenly spaced with the vanes.As these variations in flow pass the cutwater (or casing vane tips in the case of double

volutes and vaned diffusers), a hydraulic reaction force excites the rotor at the vane passfrequency.

Corrective ActionsVibrations at vane pass frequency represent another source of excitation of structural

resonances and also additional stress to the pump and driver. The most effective method

for minimizing these vibrations is to maintain sufficient radial clearance betweenimpeller and cutwater (or casing vane tips). In truth, this clearance must be correctly

designed into the pump during its initial design. One reference recommended a clearance

of not less than 5% of the impeller diameter [1]. Other than trimming, pumps do notcome equipped with means to move the impeller further away from the cutwater. Anadditional means suggested for reducing the vane passing forces is the sharpening of the

trailing edges of the vane tips [4].



f=

78

4. REFERENCES

[1] Karassik, I.J., Krutzsch, W.C., Fraser, W.H., Messina, J.P. "Pump Handbook,"

Second Edition, McGraw-Hill Book Co., New York, NY.

[2] Volk, M.W., "Pump Characteristics and Applications," Marcel Dekker, Inc., NewYork, NY.

[3] Karassik, I.J., "Centrifugal Pump Clinic," Marcel Dekker, Inc., New York, NY.

[4] Corbo, M.A. and Malanoski, S.B., "Pump Rotordynamics Made Simple,"

Proceedings of the 15th

International Pump Users Symposium, TurbomachineryLaboratory, Texas A&M University, College Station, Texas.

[5] “Boiler Feed Pumps”, “Vertical Pumps”, “Horizontal Pumps”, Applied DiagnosticsWorkshop - Book 1, Bently Nevada Corp., Minden, NV.

[6] “Vertical Slurry Pump”, Machine Diagnostics Case Histories, MachineLibrary,

Bently Nevada Corp., Minden, NV.

[7] Jackson, Charles, "Centrifugal Pumps – Maintenance & Design, Shafts, Bearings, &

Sleeves" Issue Number 2, CJ On Pumping, 7/8/70.

[8] Fox, R.W., MacDonald, A.T., "Introduction to Fluid Mechanics," Fourth Edition,

John Wiley & Sons, Inc., New York, NY.

[9] Ibrahim, A. and Sace, E., "ADRE for Windows – instrumental in solving a complex

vibration problem on a boiler feedwater pump," Orbit, Bently Nevada Corp., v.19,

No. 1, March 1998.

[10] Eisenmann, Robert C., Sr., and Eisenmann, Robert C., Jr. "Machinery Malfunction

Diagnosis and Correction," Hewlett-Packard Professional Books, Prentice-Hall, Inc.,

Upper Saddle River, New Jersey.

[11] Hatch, Charles T., "Malfunction Diagnosis: Unbalance and 1X Vibration,"

MachineLibrary, Bently Nevada Corp., Minden, NV.

[12] Hatch, Charles T. and Fahy, Dave "Malfunction Diagnosis: Misalignment,"


[13] Hatch, Charles T., "Malfunction Diagnosis: Rub," MachineLibrary, Bently

Nevada Corp., Minden, NV.

[14] Hatch, Charles T., "Malfunction Diagnosis: Shaft Crack," MachineLibrary,

Bently Nevada Corp., Minden, NV.



f=

[15] Hatch, Charles T., "Malfunction Diagnosis: Fluid-Induced Instability,"


[16] Muszynska, A., "One Lateral Mode Isotropic Rotor Response to Nonsynchronous

Excitation," BRDRC Report No. 4, 1991, pp. 1-31, MachineLibrary, Bently Nevada Corp., Minden, NV.

137132221-centrifugal-pumps-pdf.pdf

Documents