PUMP VIBRATION ANALYSIS FOR THE AMATEUR

by
William E. (Ed) Nelson
Manager, Maintenance Services
Amoco Oil Company
Texas City, Texas

vibration must be kept within acceptable limits. If there is a

high level of vibration, or if it increases markedly during the
course of time, problems of a mechanical or hydraulic nature
are indicated. The decision to remove a piece of pumping
equipment from service for repairs should be based more on
practical economic factors than on any arbitrary standards
for vibration. Production losses, repair costs due to extensive
damage, and personnel safety concerns must be weighed,
and previous experience used to arrive at the final decision
to shut clown or to run the pump.
There are many potential causes of vibration. Sorne of the
causes are shown on the vibration spectrum of Figure 1. The
pump design and construction characteristics must be evalu
ated and the history of the pump's operational reliability
considered, before positive conclusions can be drawn as to
the cause of the vibration and safe operational levels of
vibration. Detailed measurements and analyses of the vibra
tion characteristics, the pump, and the pumping system are
often needed for a reliable diagnosis.

William E. "Ed" Nelson is Manager of

Maintenance Services for Amoco Oil Com
pany in Texas City, Texas. After receiving a
B.S. degree in Mechanical Engineering
from Texas A&M in 1951, he spent four
years as a Naval Engineering Ojficer.
Mr. Nelson has spent thirty-two years in
various engineering, materials management
and management positions at Amoco's Texas
City refinery. His responsibilities include
refinery instrument, electrical, and machin
ery repair. He is also responsible for mobile equipment and mainte
nance training.
He is listed in several editions ofWho's Who in the South and
Southwest and World's Who's Who in Commerce and Indus
try. He has several technical articles and speaks at seminars and
technical meetings.
Mr. Nelson holds memberships in ASME, lMI, the Turbomachin
ery Symposium Advisory Committee, is and a former member of the
International Pump Symposium Advisory Committee. He is a
regisered professional engineer in the State of Texas.

INTRODUCTION
Maintenance is the art of prolonging or restoring the
useful operating condition of a piece of equipment. Many
engineers consider maintenance to be based on industrial
mythology or folklore and as a result engineering principles
are not always applied to maintenance. Vibration monitor
ing coupied with sound engineering judgement could be ope
of the most valuable tools of maintenance but its value is
somewhat restricted at present.. There are frequently two
groups of personnel involved with large machinery; those
who understand the machinery and those who understand
the monitoring equipment. Frequently, there are dif
ferences of opinion. Resolving these differences is hard to
accomplish, because the text books available on the subject of
pump vibration are out of date. More effort must be made to
update the information available and to improve our .abilities
to cope with the problems faced in trying to achieve econom
ical pmping equipment runs. The problem of how to
develop an understanding of what vibration monitoring
data from large single stage or multistage pumps should
mean to the average engineer, operator or m.aintenance
person will be addressed. In short, it is a first aid manual on
what to do for the "patient," the pump, until the doctor or
vibration expert arrives.

_4

0.s

ll\G

1.0

2.0

J.o

ROTOR VlBRATIO!s FREQt:E..<,:CY/!iACHI'. SPUO (CPS/RPSJ

HvOQ"'-'!C

Figure 1. Frequency Spectrum of Pump Vibration Problems.

The first step in any analysis program is to draw a
distinction between vibrations that are system-related and
vibrations from a mechanical or hydraulic problem in the
pump.

SYSTEM RELATED PROBLEMS

First, let's address vibration caused by a pumping system
related problem. A typical pumping system can be divided
into seven elements.
the driver,
mechanical transmission of power through either the
coupling, and in some cases, a gear reducer, or increaser,
the driven pump,
the suction piping section and valves,

CAUSES OF VIBRATION IN CENTRIFUGAL

PUMPS
In every pump, dynamic forces of mechanical and hy
draulic origin are present and a certain amount of vibration
is therefore inevitable. To ensure the safety of the pump, the
109

llO

PROCEEDINGS OF THE FOURTH INTERNATIONAL PUMP SYMPOSIUM

the discharge piping and valves,

the instrumentation for control of the pump flow, and
the foundation.
Sorne system related vibration problems may be obvious,
but many are interrelated with those vibrations emitting
from the pump itself. Sorne typical system-related problems
include the following:
Unfavorable dynamic behavior of the foundation, sup
porting structures or piping (resonance excitation by forces
at rotation speed).
Excitations from the coupling area, especially due to
misalignment of the driver or eccentrically bored coupling
hubs.
Excitations from the vibrations of the driver (motor,
steam turbine, gearing).
Unfavorable incoming flow conditions such as cavita
tion, intake vortex, or suction recirculation.
Flow disturbances in the suction piping due to poor
design and layout, especially in valves.
High pressure pulsations due to hydraulic instability of
the entire pumping system.
Control system/pump interaction during startups or
other periods of low flow.
Methodical investigation will help to identify vibration
sources. The investigation into the causes of the vibration
problem should include a look at ail the possibilities starting
at the most basic parts of the system.

Figure 2, that apply to good foundation design for API or

larger pumps are:
The mass of the concrete foundation should be five
times the mass of the supported equipment.
Imaginary lines extended downward 30 degrees to
either sicle of a vertical line through the machinery shaft
should pass through the BOTTOM of the foundation and
not the sides.
The foundation should be three inches wider than the
baseplate ail the way around the machinery up to 500 hp and
six inches wider for larger machines [I].
These guidelines can be used to quickly evaluate the pump
foundation as a source of vibration potential. Foundation
vibrations are usually at rotational speed. If an inadequate
foundation is identified as the source of a vibration problem,
adding concrete mass may reduce the tendency of the
foundation to respond to vibrations and transfer them. For
example, if a concrete mass equal to the full weight of the
machine is added to the base of the machine, the resulting
velocity energy will be approximately half that of the original
design. If twice the mass is added, the vibration energy will
be eut to about 1/3 of the amount experienced on the original
without the added mass. While often excellent for minimiz
ing vibration, adding mass is a costly remedy after the
foundation installation is completed [2].
On the other end of the spectrum, many ANSI style
chemical process pumps in the range of 5 to 50 hp are
frequently installed with no specially constructed founda
tions or foundations that do not meet the above guidelines.

PROBLEMS EXTERNAL TO THE PUMP

Piping Design
Fluid flow disturbances in valves and piping create vibra
tional waves which travel freely in the piping, transfer to
pipe hangers, and appear at the pump or at remote loca
tions. Loose fitting valve gates or plugs, excessive throttling
by use of either the discharge valve or a control valve,
improperly sized piping and tortuous flow paths give rise to
noise and vibration. Dynamic hydraulic forces may result
from not understanding the total pumping system. As an
example, in hot reboiler services of a petrochemical plant,
the pump flow is frequently controlled from the tower
bottoms temperature. The control valve, located between
the pump and the reboiler, may cause considerable piping
vibration to occur due to slug-flow pressure pulsations from
the flashing in the reboiler. One possible correction is the
relocation of the control valve between the column or tower
and the reboiler. The same problems may arise in hot water
pumping systems.
Sometimes piping fittings such as block valves, check
valves, and strainers must be stacked vertically to accommo
date pipe rack space limitations rather than to satisfy noise
and vibration elimination concerns. Piping tees, valves,
strainers and elbows may cause flow disturbances. Good
practices that may be used to reduce these problems on the
suction piping include:
Valve stems and tee branches should be installed perpen
dicular to the pump shaft, not parallel to the shaft.
Piping should have about five pipe diameters of straight
run before the suction flange.
To prevent resonant vibrations, long piping runs should
be supported at unequal distances. The pipe wave can travel
through the pipe hangers if they are at equal distances.
Careful and generous use of pipe hangers to prevent the
transmission of vibration through the structure.
Secure anchoring and generous application of expan
sion joints and loops in the pipe [2]

Foundation
Both dynamic and static forces must be considered for soil
support of foundations. Weil designed pump foundations
and base should have the required rigidity to withstand the
axial, transverse, and torsional loadings of rotating
machines. Sorne conservative rules of thumb, as shown in

BASE PLATE
r-'----f'---'-----\---'-__..... EPOXYGROUT
GRADE ELEVATION

FOUNDATION

FOUNDATION'S MASS SHOULD BE APPROXIMATELY

5 TIMES THE MASS OF SUPPORTED MACHINERY.

TYPICAL INSTALLATION
IMAGINARY UNES EXTENDED DOWNWARD 30' TO EITHER SIDE OF VERTICAL lt_SHOULD
PASS THROUGH BOTIOM OF FOUNDATION.

Figure 2. Rules of Thumb for Foundations [1].

PUMP VIBRATION ANALYSIS FOR THE AMATEUR

While these are simple rules, they are frequently over

looked or ignored in piping design.
As it always does when dealing with machinery, the
question arises as to how much vibration can be permitted in
the piping system? One rule of thumb, that works very well,
states that the permissible unfiltered velocity readings taken
on the piping at the midspan of its supports can be three
times the permissible readings taken on the bearing caps.
Experience indicates that bearing cap readings in the range
of 0.5 to 0.6 inches per second are the concern level for a
pump and 1.0 in/sec is the emergency shutdown level. These
pump vibration guidelines would then give 1.5 to 3.0 in/sec
as the limit for piping. Mateo has a very good discussion on
the subject of piping vibration limits [3].
Piping Installation

Dynamic pipe reaction frequently imposes high vibration

forces on pump casings. Heavy forces imposed by pipe will
sufficiently distort casings and bearing housings to create
vibration. Pump casings are designed for specific shaft and
flange loadings. Frequently neither the purnp vendor or any
industry standards define very well the acceptable external
forces and moments about the flanges. In the petrochemical
industry, limits are outlined in API Specifications 61O. These
limits are somewhat vague, although the sixth edition (1981)
and the soon to be released seventh edition (1987) have
addressed this problem in greater detail. Even with the most
accurate values of the forces and moments set by the
standards, piping loads may be harmful to the pump unless
proper engineering practices are followed. Provisions must
be made to independently support piping and provide for
thermal expansion of both the pump and its piping. This is
especially important in hot service pumps. New pipe design
on pumps operating above about 400F should be carefully
reviewed or audited by the owner or other independent
party for correct input and desired results before construc
tion. A good source for guidance in this area is Steiger [4].
Many mistakes have been made when "piping-up" to
pumps, with some resulting in equipment destruction or a
continuing reliability problems that are difficult to diagnose.
The following general practices should be followed when
"piping-up" pumps:
Never connect piping to the pump prior to grouting.
The piping system should be fabricated starting at the
pump flanges, then work toward the pipe rack.
Excessive loads to the pump flanges must be avoided
during piping fabrication by using temporary braces and
supports as required.
Extreme care in the alignment of piping components
must be used during initial fabrication to prevent problems
later. Many times, a completed, tested, and to ail appearances
i
acceptable, piping system has required cutting, ftting,
welding, and retesting, due to Jack of caution during initial
fabrication, with regard to alignment.
Review the piping support design carefully. A good
system design will support the piping loads and forces using
spring hangers and bracing, etc., which do not require
removal during normal maintenance.
The design settings of ail pipe hangers should be clearly
indicated on the hanger nameplate, perferably stamped, so
that time and the weather will not eradicate this important
information.
After ail the fabrication and testing work is complete,
loosen piping flanges. Release ail Iock-pins from spring
hangers and remove any temporary bracing previously

111

used. Adjust the system supports as necessary to free the

pump from an piping strains.
The piping fabrication error that can produce the
largest piping strain is nonparallel flange faces. A feeler gage
will help to detect this problem. Generally speaking, if a
difference can be detected in the two facing planes, piping
strain will exist.
To check for piping strain, place dia! indicators to
monitor both vertical and horizontal movement of the pump
shaft. Suction and discharge flanges should be made up
separately, with indicator readings observed continuously.
Should movement exceed two mils, piping strain is con
sidered excessive and should be corrected. Loosen pipe until
alignment is complete, then tighten and retest for strain.
Misalignment Between Components

A major cause of vibration is misalignment between the

pump and its driver. Machines that normally operate at
elevated temperatures must tolerate vibration during a
temporary cold misalignment, until normal operating tem
perature is reached. Even though extreme caution is
exercised in initially aligning shafts of coupled components,
misalignment may still result from:
loose foundation bolts,
flexible baseplate designs,
piping loads from dynamic, static or thermal causes,
bearing deterioration from internai misalignment or
lubrication problems,
misconceptions about relative thermal growth between
the driver and the pump,
incorrect positioning of driver and pump such that
distance between shaft ends (DBSE) exceeds the flexing
limits of the coupling, and
failure to use dowels to maintain the alignment once, it is
established.
While flexible couplings will accommodate sizable mis
alignment without impairing the life of the coupling itself,
that same amount of misalignment may cause damage to
pump or driver bearings and mechanical seals.
Misalignment of the driver and pump usually results in a
high axial vibration reading. The axial vibration may be as
much as 1.5 times the vertical or horizontal readings. Vibra
tion caused by misalignment generally occurs at the running
speed of the pump, although it may also occur at multiples of
the running speed. Vibration caused by misalignment can be
distinguished from a resonance disturbance by monitoring
the pump during a "coast clown" period. Misalignment
vibration will shift in frequency directly with the rotational
speed. Shaft critical speeds or resonances will not shift.
Unbalanced or eccentrically bored couplings frequently
cause vibration. Each coupling boit assembly (boit, nut, and
washer) must have the same weight; and each coupling boit
hole must have a complete boit assembly for balanced
operation. Even missing set screws and improperly sized
shaft keys can cause mechanical unbalance and consequent
vibration. The coupling should not create axial thrust dur
ing operation even under conditions of shaft misalignment.
A gear type coupling will transmit axial thrust imposed upon
it while operating under a significant driving load due to
frictional conditions between the two gear sets. The coeffi
cient of friction increases as both torque and speed increase
and the axial force actually transmitted through the cou
pling can be up to 1/3 lb/hp transmitted. The thrust load
developed by this "slip-stick" situation can exceed the capa
bilities of the pump or its driver's thrust bearing. This high

112

PROCEEDINGS OF THE FOURTH INTERNATIONAL PUMP SYMPOSIUM

friction condition, called coupling "lockup," can cause vibra

tion at either one or more likely two times running frequency
depending on the extent of damage to the coupling. The
greatest vibration is likely to transfer to the outboard end of
the pump as the rotor is whipped about by the coupling
using the inboard bearing as a fulcrum.
With a membrane or dise coupling, there is little likelihood
of the vibration, caused by coupling misalignment or dam
. age, being two times running frequency due to the axial
"softness" of the coupling.
Gear Boxes
Gear problems can generate lots of vibration data but the
number of sources of the problem make interpreting the
data difficult. There are several frequencies that show up.
One is gear mesh which is always present and is equal to the
number of teeth multiplied by the rotational speed. Since the
amplitude of the gear mesh changes sharply with load,
amplitude alone is not a good indicator of the source of any
problems. Manufacturing errors like excessive backlash on
the gear teeth, eccentricities called "apex wobble," internai
misalignment of the gears or a damage defect will create
sidebands of vibration. These sidebands can be used to
identify which gear is the damaged one. The shock impulse
from a major gear defect can excite the natural frequencies
of the gears in a set to further confuse the interpretation of
the data. In general, the only way to analyze a gear box
problem is to take abundant data on it and to trend any
problems. Snap analyzes are very difficult to make.
Hydrodynamic problems can sometimes afflict high speed
gears. The problem seems to be a combination of circum
stances which makes the lube oil drain stop functioning with
the result that foam or even solid oil touches the spinning
gear wheels. If there is not ample room in the top part of the
box, so that the oil has a chance to flow to the side walls and
down relatively unrestricted, the top of the box floods. This
causes excessive power loss, high temperatures, and vibra
tion. Drain temperature increases as high as I00 F in less
than a minute have been observed. This condition is called
"choking." The vibration frequency of "choking" is wildly
random and of little value in identification.
What triggers "choking" is not really known. There are
seven gear design conditions that seem to be conducive to
"choking."

Low ratios
Downward mesh
Coarse pitch-large teeth profiles
High pitch line velocities
Gears fill a large portion of the box volume
Small sized and/or poorly located drain lines
Relatively high floor in the box

Since gear casing configuration and basic arrangement

are not readily changed, remedies for "choking" usually take
the form of modifying and augmenting the drains and vents.
The bottom half of the box should be equipped with a baffle
or pan so that the oil level and the drain is shielded from the
windage directly hitting it.

VIBRATIONS RELATED TO THE PUMP

The preceding discussion has discussed how measured
vibrations at the pump can actually be caused by system
related vibration problems. Vibrations can also be produced
by mechanical and fluid related problems within the pump
itself. Typical vibration problems of the pump are:

U nfavorable dynamic behavior of the rotor due to

excessive wear ring, bushing, or seal clearances.
Poor support of the rotor because of loose fits on the
shaft or housing in the case of ball bearings. Excessive bore
clearance or lack of "clamp" on the shell OD of sleeve
bearings causes the sarne effects.
Mechanical irnbalance of the rotating parts due to poor
balancing or careless assernbly. Operational influences in
cluding cavitation erosion, deposits, corrosion, damaged
impellers, galled parts, and abrasion can also cause irnbal
ance.
Increased radial hydraulic forces when the purnp is
operated outside of the design flow range. Sorne increase in
vibration is normal when departing frorn the best efficiency
flowrate.
Mechanically defective bearings.
Purnp manufacturer casting and/or rnachining defects.
In a malfunctioning purnp, there are lots of opportunities
to collect massive arnounts of data that can point to alrnost
any defect that can be irnagined. Monitoring systems have
not received the level of acceptance and recognition they
deserve because of a lack of operator support and en
thusiasm, problems of detection-false alarrns or no alarrns,
and the arnount of data generated. The data compiled
exceeds the operator's (and-rnost everyone else's) ability to
cope with it. The end result is a general reluctance to accept
monitoring as the total factor in maintenance of purnps. In
spite of this negative atrnosphere, a change in ideas is
overdue, because of rapidly increasing maintenance costs.
Logic and monitoring data must be tied together to answer
questions if these costs are to be reduced.
Converting Vibration Data into Logical Thoughts
Associating the collected vibration data with the cause of a
system and/or pump related problem is the point where most
of us become confused and ineffective. Where do we start
the analysis process? An analysis method used by one of the
better known machinery troubleshooting experts, Sohre [5],
is to relate the possible vibration causes and problems
according to the ratio of the occuring frequency to the
rotational speed, as shown in Figure 1.

SUBSYNCHRONOUS FREQUENCIES
Five Percent to Thirty-Five Percent Frequency Spectrum
Flow Disturbances
There is a low frequency axial vibration that occurs due to
eddy current flows around the impeller from excessive
impeller shroud to casing clearances, called Gap ".N', and
suction recirculation, both depicted in Figure 3. Flow distur
bances related to suction recirculation and cavitation are
always present in both diffuser and volute type purnps.
These eddy currents are shown in Figure 3. As best efficien
cy point (BEP) operating conditions are approached, the
frequency band width of the recirculation rnay narrow some
what and sometimes the frequency will increase. Assuming
that there is a sufficient margin between NPSH Required
and NPSH Available, the noise or vibration associated with
recirculation can be identified by stepping the operating
flow toward the BEP flow. If recirculation is a cause, the
vibration levels will decrease.
Remedies For Recirculation
Suction recirculation flows are made considerably worse
by operation away from the best efficiency point or BEP. The

PUMP VIBRATION ANALYSIS FOR THE AMATEUR

-_-_- -_--:_A? '{

_

'

113

----_,
""

<)

a) Reduc1ion of impeller diam1:1C-r

b) Reduction of the b-lade

c) Obhq1Je reduction of the bladcs

d) Oblique redu.:tion of the blade>

e) Redul1on of 1mpcllr diamete:r

Figure 4. lmpeller Diameter Reduction

Figure 3. Eddy Flow Pattern at Suction Eye and lmpeller Vane Tzp
at Off Design Flow Operations.
only real remedy is to operate the pump as far out on the
curve near BEP as possible. There are four Rules of Thurnb
about the recirculation phenornenon that should be rernernbered.
As the energy level (measured by the head developed amd the
horsej,ower) increases, the recirculation driving force inrreases for
operations away Jrorn the BEP.
If the npeller diameter has been reduced by trirmning, the flow
dtribution across the exit width of the irnpeller becornes rnore
unstable. The tendency for the high pressure liqul to return to the
low pressure side of the vane and create tip recirculation is greatly
increased.
The onset of recirculation can occur at alrnost anyflow ratefrorn
near 100 percent down to about 40 percent ofBEP deprmding on the
pump design and other factors. Flows below about 50 percent of
BEP should be avoided entirely.
Suction recirwlation, when it occurs, will cause strong AXIAL
vibrations, especially with single stage and double suction horizontal
pumps.
Correction Of Gap "A"
The head developed by a centrifugai pump is frequently
adjusted by trimming the impeller diameter. How the impel
ler is trimmed can influence the vibration levels experienced
on the pump on account of problems associated with exces
sive Gap "A." No hard and fast guidelines for the mechanical
aspects of impeller trimming exist, but there several pump
construction and hydraulic design factors to consider while
making the decision.
For most pumps, eutting the entire impeller, vanes and
shrouds, as shown in Figure 4(a) will often increase the axial
vibration and other problems associated with Gap "A"
(shroud to case clearance), due to the uneven flow distribu
tion at the impeller exit area. Trimming the vanes only tends
to even out the exit flow pattern and reduce the recirculation
tendencies at the exit area. Gap "A" should be about 50 mils
(radial) for minimum axial vibration because of tip recircula
tion [10).

In most cases, it is best to trim only the vanes as shown in

Figure 4(b). The oblique eut in Figure 4(c) yields a more
stable head curve because the tenclency for tip recirculation
and the possibility of suction recirculation being established
is greatly reduced due to a more uniform flow distribution at
the exit area. Structural strength of the remaining shroucls is
a factor in the decision as to how to trim the impeller. There
may be too much unsupported shroud left after a major
reduction in diameter. It is often best to eut the impeller
vanes obliquely, as shown in Figure 4(d), leaving the shrouds
unchanged and solving the structural strength problem
along with improving the exit flow pattern.
The double suction type pump is especially sensitive to
problems caused by increased Gap "A," so trimming the
entire impeller is not a good choice for this style pump.
Forty Percent to Fijiy-five Percent SjJectrum Frequency
Oil Whirl
A condition known as "oil whirl" may occur in lightly
loaded sleeve bearings (Lmder 90 to 100 psi). It is a self
sustaining type of rotary motion and can be explainecl by
considering a wedge of oil traveling around the bearing at its
average velocity or half the surface speed of the shaft. The
flowing wedge of oil lifts and drives the shaft ahead of it in a
forward circular motion within the bearing clearance. This
motion is oil whirl. This oil film flow is due to fluid friction
and has an average speed of Iess than one half of journal
surface speed (42 percent to 4 7 percent), which is the
characteristic frequency of this instability. This vibration
may be caused by excessive bearing clearances or a low lube
oil temperature. Severa! bearing conditions can produce oil
whirl:
light unit loading,
excessive clearances,
high surface speeds,
oil properties (primarily high viscosity due to cold tem
peratures), and
initial deflection of the rotor which generates a nonsym
metrical oil wedge along the axial length of the bearing.
Resonant or Critical Speed
Ail rotors have certain rational speeds at which the shaft
vibrates at its natural frequencies. The natural frequencies
are those frequencies at which the rotor would vibrate if
struck by a heavy blow while stationary.
Many large multistage pumps operate above the first
resonant "dry critical" speed and below the second. Low

114

PROCEEDINGS OF THE FOURTH INTERNATIONAL PUMP SYMPOSIUM

speed pumps typically operate below the first "dry" critical

speed. Just barely enough is known about criticals to enable
the building of machinery which will run. Design numbers
out of a computer and what happens in the real life machine
in the field are often something quite different. Sorne
selected quotations from one of the best brief discussions on
this subject from Gopalakrishnan [6], may be helpful in
understanding this subject.
In a rotating machine handling gas, the critical speed is
basically determined by the mass and stiffness of the rotor
and its bearings. In a centrifugai pump, the presence of
liquid brings in several additional effects. For any centrif
ugai pump to fonction, it must have some close clearance
areas. The number, location, and dimensions of these are
dictated by the design philosophy. These areas include
wear rings (sometimes called seal rings), throttle bushings,
etc., each of which is, in effect, a running fit between a
rotating sleeve within a stationary bushing. The effect of
such fits is evidenced in several ways. First, the liquid in
the bushing resists the radial displacement of the sleeve.
The ratio of this resisting force to the displacement is
called its stiffness. Secondly, the liquid also produces a
restoring force proportional to the transverse velocity of
the shaft movement. The ratio between this force and
velocity is the damping coefficient of the bushings. Finally,
the liquid also produces a restoring force proportional to
the transverse acceleration of the sleeve. The ratio of this
force to the acceleration is frequently termed the virtual
or hydrodynamic mass of the bushing.

A pump rotor's "dry" critical speed will not remain con

stant for life as it may be influenced by such operating factors
as the mass, stiffness, and damping characteristics of the
rotor and its supporting structure. A small amount of
damping will cause the critical to exhibit a sharp vibration
peak over a narrow range of spee. A large amoun! of
damping such as the leakage at balancmg drum or wear n.r:igs
can cause the critical to exhibit a broad peak over a w1de
range of speed.
The "wet" critical can be at almost any frequency above or
below the running speed since it also changes with wear at
the leakage points. Normallf the critical is not seen at ail_ due
to suppression by the dampmg effect of the leakage pomts.
Sorne quotes from the literature may further clarify the
problems [7].
The forces occurring in the wear rings and any balanc
ing drum have a considerable influence on the vibrational
behavior of the rotor. The balancing drum has a great
influence because of its large area and high pressure
differential. The restoring force acting in opposition to
rotor deflection has the effect of increasing the natural
frequency, while the forces acting in the tangential di ec
'.
tion can lead to instabilities in the form of self-excited
vibrations.
At rest, the rotor has a slightly lower natural frequency
in liquid because the liquid has the effect of an additioal
mass. With increasing rotational speed, the pressure d1f
ferential across the labyrinths and balancing piston in
creases, as a result of which the above mentioned restor
ing forces are built up and the natural frequency, which
then depends on the speed, is raised.
Vibrations excited by unbalance forces occur at the
rotation frequency (synchronous vibrations) or at a higher
harmonie frequencies, without the occurrence of subsyn
chronous vibrations. Subsynchronous vibrations only be
corne noticeable in the case of self-excited vibrations.
These can occur, for example, in boiler feed pumps with
heavily worn wear rings, since with increasing wear ring
clearance, not only the natural frequencies but also the
rotor damping rates are reduced.

Oil Whip

If the cpm of an oil whirl condition is near a rotor

resonance or critical that is below rotational speed, oil whirl
excites and locks on to this resonance, producing oil whip.
Oil whip is a forward circular orbit of the shaft at the
resonance speed. Since it is a shaft vibration occurring at the
natural frequency of the shaft when the speed of the shaft is
two or more times the natural frequency, the speed can be
less than or greater than 50 percent of running speed. This
might be considered the stabilized version of the half
frequency whirl, the constant frequency being due to the
persistence of the first natural critical of the rotor.
Severity of Oil Whirl or Whip

The shaft vibration resulting from oil whirl or whip can

vary from completely acceptable levels to highly destructive
levels. Since the whirl is in the oil film itself, the basic
tendency is very destructive, because the capacity of the
bearing is reduced to near zero. As a rule of thumb, oil whirl
movement should not exceed 50 percent of the bearing
clerance. At higher vibration levels, the machine must be
shut clown immediately to avoid serious damage.
Correction of Oil Whirl or Whip

There are several methods that can be used either indi

vidually or in combination to eliminate oil whirl or whip:
Install new bearings with reduced clearances.
Preload the bearing by a pressure dam or reducing the
net area of the bearing, raising the unit loading above 150
psi.
Raising the oil temperature. ln rare cases, lowering the
temperature can be effective in breaking up the oil whirl.
lnstall tilting pad journal bearings which cannot whirl.
Three of these suggestions involve bearing redesigns.
Numerous bearing designs have been developed to reduce
the tendency for a shaft to whirl or whip in a bearing. These
designs can be listed in the following order, from the worst to
the best, of ability to prevent whirl or whip:
Design
Type

Anti-whip
Characteristic

Cylindrical
Elliptical
Pressure Pad
Multiple Groove
Tilting Pad

None
Slight
Fair
Fair
Excellent

The tilting pad bearing "fix" is the best one because it is not
wear sensitive and it is more reliable. The bearing surface of
this design is divided into three or more segments, each of
which is then pivoted at the center. Each pad tilts in order to
form a wedge shaped oil film, which tends to force the shaft
toward the center of the bearing. This is the most effective
anti-whip design.
Rubbing

Sorne partial internai rubs will cause "half frequency"

vibrations because they excite a resonance frequency. Other
rubs may produce synchronous or random multiples above
running speed. Viewing the orbit of the vibrations on a scope
may aid in determining if a rub is taking place.
Sixty Percent to Eighty Percent Frequency Spectrum
Hydraulic In Nature

This random frequency vibration is particularly found in

multistage pumps. Its causes are hydraulic unbalance, the

PUMP VIBRATION ANALYSIS FOR THE AMATEUR

effects of leakage points such as wear rings, balancing drums

or center stage bushings and impeller-volute interaction,
when operating away from the design point. At flows below
about 50 percent of BEP, these dynamic forces may be very
strong with both diffuser and volute types of pumps. Reme
dies such as anti-swirl devices can reduce the effects of
leakage points such as at the balancing drum bushing of
multistage straight through flow pumps, but these are trial
and error efforts. Total correction is difficult, since very little
is known about proper volute design.
Mechanical In Nature

Subsynchronous vibration components between 0.5 and

0.75 times rotational frequency can occur in high speed
multistage pumps because of excitation of a natural frequen
cy by loose bearing housing fits and/or excessive bearing
clearance. When these frequencies appear, the situation has
to be taken very seriously; the rotor can be destroyed in a
matter of minutes or sometimes in seconds. Both the bearing
housing and bearing clearances should always be checked
during the assembly of any pump.
One Hundred Percent Frequency Spectrum
Synchronous Speed
Apparent Unbalance

A once per revolution (synchronous) vibration is frequent

ly cited as a sure sign of simple mechanical unbalance. The
analysis is given that a short session on a balancing machine
will correct ail problems associated with the pump rotor. This
premise is more often incorrect than correct. The once per
revolution can also be a symptom of several other rotordy
namics problems other than mechanical unbalance. These
other problems resemble mechanical unbalance in nature
but have entirely different origins. The resulting vibrational
forces cause stresses in the rotor and its support structure.
The vibrations from these forces can result in:
excessive wear of bearings, bushings, seals, couplings,
and gears.
fatigue failure of rotating components such as shafts,
impellers, wear rings coming loose and spinning.
fatigue failure of stationary structures such as bearing
supports, foundation bolts, and piping.
transmission of vibrations to adjacent machinery and
causing of additional problems.
generation of noise.
The origins of the problems that produce synchronous
vibrations are many and varied. Sorne are built into the
pump while others are generated by poor operational or
maintenance practices. The following discussion covers
some of the more commonly encountered causes.

115

pattern wear, core shift, poor metal flow, and shrinkage.

Good quality control at the foundry or the use of investment
castings is the only correction for this condition. It is impossi
ble to eliminate or even to reduce hydraulic unbalance by
mechanical balancing of the impeller, since the cause is
hydrodynamic and its action is unpredictable. Sorne pumps
can exhibit synchronous response and some subsynchro
nous response because of hydraulic unbalance. The only
remedy for the problem is to change the impeller to one of
better quality with equal and uniform liquid channels.
Bowed Shafts

Bowed shafts are a common cause of unbalance. Every

effort should be made to avoid conditions that help create
shaft bows because they are difficult, sometimes impossible,
to straighten and/or balance out. Here are a few causes of
shaft bow:
Gravitational sag. Large pump shafts with enlarged clear
ances which are allowed to sit in one position for a long
period may develop bows. Upon startup, the shaft may
vibrate violently.
Thermal distortion. During the startup of hot oil pumps, a
bow may develop in the rotor shaft because of heat distribu
tion. The rotor must be slowly brought up to speed while
uniformly heating it to avoid creating a large thermal bow.
The casing may also be bowed due to heat distribution.
Thermal Expansion. Thermal growth can cause vertical
and horizontal misalignment that must be accommodated by
the coupling. Internai misalignment can also result. The
relative growth of a typical multistage hot oil pump of the
barrel or double casing design is shown in Figure 5. The
casing grows away from the support point while the rotor
grows toward the support. Since the thrust bearing is at
tached to the casing, part of the rotor growth is canceled out.
SUCTION

DISCHARGE

DATUM PLANE

KEY-____...

ENLARGED VIE'J
SECTION A-A

AMOCO OIL
Rtf.J-HOT DIL PUHPS CXIM-

Figure 5. Thermal Expansion of Hot Pump.

Mechanical Unbalances

The impellers of many centrifugai pumps are sand cast

ings. Investment cast impellers are greatly preferred to sand
castings. A sand cast impeller will have some degree of
internai physical variations, of dimensions or of mass, which
will appear as either ( 1) mechanical unbalance or (2) hydrau
lic unbalance. A good dynamic balancing can correct some of
the problem but not ail of it.
Hydraulic Unbalance

Hydraulic unbalance is the result of uneven flow patterns

and volumes between vanes or liquid passages due to their
uneven spacing, caused during the sand casting process by

Contrai of this growth is very important to both internai

and external alignment. The heat expansion of pump parts
can affect the mechanical performance of the pump in
several ways.
Usually the parts of the rotating element are of different
materials than the pump case, and the running clearances
may be reduced below a safe minimum due to their different
rates of thermal expansion.
Shaft sleeves, impeller rings, and case wearing rings,
which are tightly fitted, may become loose when hot, because
of a difference in expansion, especially during quick start
ups.

116

PROCEEDINGS OF THE FOURTH INTERNATIONAL PUMP SYMPOSIUM

The rotating element may get out of alignment when the

case is hot due to heat distortion of the case. Internai leakage
further aggravates the thermal distortion, since uneven
heating occurs along the leakage path. If the pump is put on
stream cold and is brought up to temperature gradually over
several hours, then there is sufficient time for all parts to
reach the same temperature with little case distortion. How
ever, if hot oil is charged to the pump suddenly, the various
pump parts do not reach the same temperature at the same
time. Usually, the upper part of the outer case reaches its
highest temperature an hour or more before the bottom
because the flow tends to be in the top of the pump near the
inlet and discharge nozzles [8].
In barrel or double wall type pumps, shut clown leaves
liquid trapped in two areas of the casing. One area is the
inner liquid channels of the bundle and another area is the
annulus between the bundle and the casing (barrel) that is
sealed at the discharge end. Thermosyphon action occurs in
each area. Field experience indicates a temperature of up to
l 75 F can develop. This temperature differential causes
greater expansion of the casing top half, which pushes the
bearings downward and the pump center upward, in rela
tion to the center line, so that a half turn of the rotor can
cause binding at leakage points such as wear rings and
balancing drums. The amount of deformation depends of
the location of piping nozzles including flush, cooling and
balance line connections. Severa! hours must elapse before
the pump is uniformly cool [9].
Putting the rotor on turning gear is a common practice to
solve the problem of thermal sag of the rotor. Recommencled
turning gear speed is usually 35-60 cpm, which is enough to
establish an oil film wedge in the bearings. Such low speeds
are not enough to create circulation in the liquid channels of
the pump and thereby aggravating the thermosyphon ef
fects and causing more distortion [9].
In dealing with pumps that will operate at temperatures
greater than 500 F, extreme care must be taken to ensure
that ail auxiliary piping be welded to the case and the case
rough machined prior to final heat treatment. This will
prevent distortion due to stress relieving by temperature
cycles in service.
Shaft Rubbing
Rubs or contacts of stationary parts like oil baffles, glands,
seals, or bushings on the rotor will cause hot spots which can
bow the shaft. Thermal bows in shafts can cause vibration at
running speed or sometimes twice running speed. The bow
tends to act like an unbalance condition giving the one times
running speed characteristic. The tendency is toward two
times running speed in very severe rubs. For every rotation
of the shaft, a vertical impulse on the outboard and inboard
end of each bearing will occur, thus there are two impulses.
lmproperly Fitting Parts
Loose components may cause vibrations that shorten
equipment life. One example is the unbalance imposed on a
rotating assembly by the shift in mass center of loose rotor
components. The design fit for a pump impeller cannot be a
heavy fit as used in turbines or compressors because of
assembly problems. Loosely fitting nonrotating and rotating
parts generate vibrational waves that travel freely through a
machine and its structure. To avoid these problems the
following guidelines for rotor assembly are important:
The impeller to shaft fit must be accurately controlled
and be a slight interference fit. Many impeller castings are

not properly stressed relieved, so bore dimensions may

change while in hot service, or when they are heated for
removal. Careless or excessive application of heat during
installation or removal can cause distortion and Joss of fit.
The impellers must be precision balanced on a solid
mandrel.
The shaft, thrust collar, balancing drum and ail other
rotating parts must be precision balanced.
Ali fits and runouts; impeller to shafts, keyways, etc.,
must be carefully checked.
Stack the rotor (and diffusers if final assembly) 111 a
vertical position.
Heat the impellers as necessary.
Quick cool the toe or eye area of the impeller with the
flow of shop air from a ring about two inches larger than the
shaft.
Dia! indicate the rotor on the shaft, wear ring and
bushing areas for runout due to stacking deviations. Runout
should be below two mils. If runout exceeds this level, look
for skewed impellers or a heat bowed shaft.
Check balance the entire rotor.
Lass of Balance
A progressively worsening mechanical unbalance may be
caused by uneven wear or corrosion. Pump rotors are
particularly susceptible to corrosion and erosion, major
causes of mechanical unbalance. Water may corrode boiler
feed pump impellers. Catalyst or ash fines may erode slurry
pump impellers. Cavitation or suction recirculation flows
may pit and erode any centrifugai pump impeller where
marginal suction conditions exist.
Sleeve Bearings-Excessive Clearance
Excessive clearance or damage in sleeve bearings is nor
mally one times running frequency. The traditional cylindr
ical bore bearing has some basic problems for use in the
larger and higher speed pumps. To keep bearing tempera
tures at desired levels, larger clearances must be used. This
runs counter to the need for keeping rotor vibration to
acceptable values. Excessive clearance or damage can nor
mally be cleterminecl by comparing the vibration level read
ing on the bearing housing, as close to the shaft as possible,
with the vibration reading taken on the shaft itself. The
pickup must be in the same plane and direction when taking
both readings. Before bearings are replaced, a careful
dimensional check should be made of the bearing bore and
the shaftjournal surface to spot any bore size change, out of
roundness, or shaft wear.
Sleeve Bearings-Mechanical Looseness
Looseness of bearing caps or brackets usually results in
one times running speed vibration, a large number of
harmonies and sometimes partial frequencies in the vibra
tion spectrum. Vibration levels are generally higher in the
vertical plane than in the hmvnntal plane. Bearing housing
bores and a bearing shell Of liner diameters are sized to
impose an exact "crush" or holding force. Check the bearing
shell outside diameter to the housing bore to be sure there is
sufficient "crush" to hold the bearing. This clearance should
be checked using "plastigage" strips between the top of the
shell and the bearing housing while five mil shims are placed
on both sicles of the split line of the housing. The squeeze of
the "plastigage" should indicate the clearance. The "crush"
should be at least metal to metal, with one or two mils of
interference preferred.

117

PUMP VIBRATION ANALYSIS FOR THE AMATEUR

Ball Bearings-Looseness
Unfortunately, many pump builders do not indicate the
proper bearing fits for shaft and housings to guide shop
repairs. The original dimensions of both the housing and the
shaft will change from time to time from oxidation, fretting,
damage from locked bearings and other causes. Every
bearing handbook has tables to aid in selecting fits. The
vibrational effect of looseness on the bearing fits is different
for the housing and the shaft.

Table 1. Vane Passing Frequencies

Housing Fits
Ball bearing fits in the bearing housing are of a necessity
slightly loose for assembly. If this looseness becomes exces
sive, vibration at rotational speed and multiple frequencies
will result. Obtaining a loose fit of the antifriction bearing
outer race when mounted in the bearing housing is done by
measuring the bearing OD to verify compliance with re
quired dimensions. Do not install bearings with ODs outside
of the given tolerance band since this might result in either
excessive or inadequate outer race looseness.

confusing. Vane passing frequency is the number of impeller

vanes times cpm for a double volute pump, IF the number of
impeller vanes is an even number. IF the number of impeller
vanes is an odd number, it is the number of impeller vanes
times cpm times two for the same pump. IF it is a diffuser
pump, the vane passing frequency is the number of impeller
vanes times the number of diffuser vanes times cpm. A quick
reference for this information may be viewed in Table 1.

Rules of T humb:
Bearing OD to housing clearance-about 3/4 mil loose
with 1-1/2 mils max.
Bearing housing out of round tolerance is one mil max.
Bearing housing shoulder tolerance for a thrust bearing
is O to 1/2 mils per in of diameter off square up to a maximum
of two mils.

Correction of Gap "B"

Gap "B," as shown in Figure 2, is a major factor in the
magnitude of the vibration level from vane passing frequen
cy. Careful machining of the volute or diffuser tips to
increase gap "B" while maintaining gap "Pl' has been used
for over fifty years to reduce the vibration levels of vane
passing frequency. The pulsating hydraulic forces acting on
the impeller can be reduced 80 percent to 85 percent, by
increasing the radial gap from about one percent to six
percent. Definitions and recommended dimensions for the
radial gaps are presented in Table 2. Locations of Gap "X'
and Gap "B" are shown in Figure 3. Many pump handbooks
state that shortening the diffuser or volute channels will
reduce efficiency. Experience has shown that there is little or
no loss of overall pump efficiency when the diffuser or
volute inlet tips are recessed. The maintenance of efficiency
in actual pumps probably results from the reduction of
various energy-consuming phenomena: the high noise level,
shock, and vibration caused by vane passing frequency, and
the stall generated at the diffuser or volute inlet [ 1 O].

Shaft Fit
A loose fit of the shaft to the bearing bore will give the
effect of an eccentric shaft, at a one times running frequency
vibration pattern. The objective of the shaft fit is to obtain a
tight fit of the antifriction bearing inner race, when mounted
on the shaft. The bearing bore should be measured to verify
inner race bore dimensions. Do not install bearings with an
ID outside of the given tolerance band since this might result
in either excessive or inadequate shaft tightness.
Rules of Thumb:
Fit of bearing inner race bore to shaft is 1/2 mil tight for
small sizes-3/4 mil tight for large sizes.
Shaft shoulder tolerance for a thrust bearing is O to 1/2
mil per inch of diameter off square up to a maximum of 1
mil.
Soft Foot
An improperly shimmed foot on a pump or its driver can
cause vibration. As any residual unbalance of the rotor
passes over the "soft foot," the support structure deflects,
giving a characteristic of one-time running frequency.

No. lmpellr
Vanes

Volute Style
(Double)

Diffuser
Style

Even

No. vanes xCPM

No. impeller vanes

X diffuser vanes
xCPM

Odd

No. vanes X CPM

x2

No. impeller vanes

x diffuser vanes
xCPM

Remedies for Vane Passing Vibrations

Random Positioning of Impellers

In the diffuser style pump, a complete set of hydraulics
can be created specifically for each pump application be
cause the diffusers are cast separately from the case and can
be readily changed. In the volute design, the volutes them
selves can be relocated only by a very expensive pattern
change. In multistage volute pumps the degree of posi
tioning of the volutes is severely limited by case design, so it is
necessary to randomly eut the keyways in the impellers to
Table 2. Recommended Radial Gaps for Pumps.

Pipe Strain
Piping strain has already been discussed as part of the
pumping system. It generally has the effect of creating
misalignment.

Type

SUPER-SYNCHRONOUS

Diffuser
Volute

Vane Passing Frequency

Vane passing frequency vibration is a hydraulically in
duced vibration at a frequency determined by the number of
impeller vanes, the number of stationary vanes and the
pump cpm. Detection of this problem sounds easy but can be

*B = 100 (R3-R2)
R3 = Radius of diffuser or volute inlet
R2
R2 = Radius of impeller
NOTE: If the number of impeller vanes and the number of
diffuser/volute vanes are both even, the radial gap must be
larger by about 4%.

Gap "A"

50 mils
50 mils

Gap "B"*
Percentage of Impeller Radius
Minimum

Preferred

Maximum

4%
6%

6%
10%

12%
12%

ll8

PROCEEDINGS OF THE FOURTH INTERNATIONAL PUMP SYMPOSIUM

assure that vanes on adjacent impellers are not aligned.

Frequently, this random selection of keyway positioning is
not clone when manufacturing the impellers and the vanes
on the impellers line up, producing a high vane passing
frequency vibration. The alignment of impellers '.1nd v<:>lutes
in each stage should be carefully observed durmg 1tness
testing or reassembly on a new pump, or when replacmg the
impellers during maintenance.
Correction of Vane Shape

Impellers manufactured with blunt vane tips can also

cause trouble by generating hydraulic disturbances in the
volute and impeller exit area even when the impeller
_ s the
correct distance, "Gap B," from the cutwater. This d1stur
bance may be partly or entirely eliminated by tapering the
vanes by "overfiling," removal of metal on the LEADING
face of the vane as shown in Figure 6 and Table 3. The same
improvement can be achieved by sharpening the impeller
blades on the underside or TRAILING edge of the vane as
shown in Figure 7. This method enlarges the outlet area of
the liquid channel and can result in up to five percent more
head near the best efficiency point, depending on the outlet
vane angle. At least 1/8 in of tip thickness must be left as
shown in Figure 7.

\\

Il

Figure 6. Impeller Vane Overfiling.

Table 3. Length of Blend For Over Filing.

Impeller Diameter

"A" Distance of Blend

10" & Below

10-Vis through 15"
15-Yl6 through 20"
20-V16 through 30"
30" & Larger

l-V211
2-V2"
3-V2''
5"
6"

There are some guidelines that can be established to

evaluate bearing deterioration. For example, a ball passes
over defects on the inner race more often than those on the
outer, because the linear distance around the diameter is
shorter. A very good discussion on early detection of the
bearing distress frequencies is contained in the literature
[12]. One word of caution on the use of the formulas in the
reference is that parameters for the same size bearing will
change with the manufacturer, so conflicting data can occur.
Also, considerable dimensional data on the bearings is
required to utilize the formulas.
Without going into the formulas [12], there are four
dimensions of a ball bearing that can be used to establish
some feel for the condition of that bearing:
A defect on outer race (ball pass frequency outer) occurs
at about 40 percent of the number of balls times running
speed.
A defect on inner race (ball pass frequency inner) causes
a frequency of about 60 percent of the number of balls
multiplied by running speed.
Ball defects (ball spin frequency) are variable with lubri
cation, temperature and other factors.
Fundamental train frequency (cage defect) occurs at
lower than running speed values.
A simple check for verification of poor bearing condition
is made by shutting off the pump and observing that the high
bearing frequency remains as the pump speed reduces. This
high frequency signal will normally remain until the pump
stops. The frequency indication is normally from five to 50
times the running speed of the machine.

CONCLUSIONS

Max. sharpening
Lcavc at
lcast2 mm

Figure 7. Sharpening of Impeller Vanes [7].

Antifriction Bearing Defects

Antifriction bearing defects are difficult o deect i;11 th_e

early stages of a failure, because the resultmg vibration 1s
very low and the frequency is very . high. If monitoring _is
performed with simple instrumentation, these low levels will
not be detected and unexpected failures will occur. The
vibration frequencies transmit well to the bearing housig,
because the bearings are stiff. They are best measured w1th
accelerometers or shock pulse meters. Spencer and Hanson
[11] have a good discussion of the reliability ot: four different
monitoring systems for rolling element bearmgs.

Vibration analysis that limits itself to collecting and "trend

ing" data is not economical. The data must be interpreted as
to a base cause and more importantly, corrections developed.
If the corrections are not developed, a major portion of the
economic incentives of monitoring systems are lost.
Vibration in pumps generally will corne from the sources
discussed briefly herein. Most of these sources are best
corrected before they occur. Good sound eng.ineerin P:ac
tices applied to the sizing and selection o eq1;11pmen 1s vtal.
Proper planning, design and construct10n mstallauon 1s a
major factor in successful operational startup. Proper
maintenance will keep the pump operating successfully.
If corrections are necessary during operation, a systematic
approach with vibration analysis instrumentation and a good
grasp of the construction peculiarities ?f th_e pmp can
identify vibration disturbances. After 1denuficauon, the
causes can be either minimized or corrected so that trouble
free operation can be achieved.

REFERENCES

1. Dodd, V. R, and East, J. R. , ASME Pump Engineering

Seminar, South Texas Section, Houston, Texas (1979).

PUMP VIBRATION ANALYSIS FOR THE AMATEUR

2. Meyerson, N. L., "Vibration: Its Causes, Identification

and Cures," Power and Fluids, Northington Corpora
tion (Date unavailable).
3. Maten, S., "Field Criteria for Pipe Vibration," Hydrocar
bon Processsing Quly 1984).
4. Steiger, J. E., "API 610 Baseplate And Nozzle Loading
Criteria," Proceedings of the T hird International Pump
Symposium, Turbomachinery Laboratories, Department
of Mechanical Engineering, Texas A&M University,
College Station, Texas (1986).
5. Sohre, J. S., "Turbomachinery Problems and T heir
Correction," Sawyer's Turboml!.chinery Handbook, II, Ch. 7
(1980).
6. Gopalakrishnan, S., and Husmann, J., "Sorne Observa
tions on Feed Pump Vibrations," EPRI Conference,
Cherry Hill, New Jersey (1982).
7. Sulzer CentrifugalPump Handbook, Sulzer Brothers Limit
ed, 1986.
8. Karassik, 1. J., Centrifugai Pump Clinic, Marcel Decker,
Inc. (1981).

119

9. O'Keefe, W., "How to Avoid Transient-Caused Problems

ln Boiler Feed Pumps," Power, Qune 1986).
10. Makay, E., and Diaz-Tous, 1. A., "Feed Pump Reliability
and Efficiency Improvements Resulting from Hard
ware Modifications," EPRI Conference, Cherry Hill,
New Jersey (1982).
11. Spencer, D. B., Hansen, J. S., "A Better Way To Monitor
Bearings," Hydrocarbon Processing Qanuary 1985).
12. "Effective Machinery Maintenance Using Vibration
Analysis," Dynamic Signal Analyzer Applications, Aplli
cation Note 243-1, Hewlett Packard (1983).

ACKNOWLEDGEMENT
James H. Ingram of Sterling Chemicals in Texas City,
Texas, spent many hours reading and critiquing this paper.
T he input of his experience and knowledge has helped
greatly in the writing of the paper and in successfully solving
many pumping problems over the years.

120

PROCEEDINGS OF THE FOURTH INTERNATIONAL PUMP SYMPOSIUM