CHAPTER 13

CENTRIFUGAL MACHINES: PUMPS, FANS, BLOWERS AND COMPRESSORS

The purpose of the pump and fan is to transport fluids by converting mechanical work
into energy of the fluid in the form of pressure and velocity. The compressor is used to increase
the energy of the compressed fluid in the form of pressure. All units are driven by diesel engines, electric motors or turbines (gas and steam). These units are divided into radial flow (centrifugal) and axial flow types depending on the motion of the flow as it passes through the impeller. In a pump the working fluid is a liquid, whereas in fans and compressors the working fluid is
a gas. Fans are distinguished from compressors by the density change (compression) in the
moving fluid induced by the compressor. The fluid moved by a fan incurs little compression. All
centrifugal and axial machines have performance characteristics that relate the head (pressure)
as well as the efficiency and horsepower to the fluid flow rate. Figure 13.1 shows characteristic
curves for centrifugal machines with different blade curvatures. For stable, efficient operation,
operation must be on the negative slope of the curve or the result will be unstable flow causing
excessive hydraulically and aerodynamically induced vibrations.

Stable
Operation

Figure 13.1. Characteristic Curves for Centrifugal Machines

13.1

PUMPS
PUMP DESIGN AND FUNCTION
The two types of pumps are centrifugal and axial (13.1). The centrifugal pump consists
of a rotating element (shaft and impeller) and a stationary element (casing, bearings and stuffing
boxes). The centrifugal pump (Figure 13.2) uses radial centrifugal action to force the flow from
the inlet (suction) to the outlet (diffuser or volute) at a higher pressure and velocity. The centrifugal pump can be oriented horizontal or vertical. To create a larger discharge head (pressure
and velocity), multistage pumps (Figure 13.2c) are used. The velocity energy is converted to
pressure by either volutes (Figure 13.2a) or stationary diffusers (Figure 13.2b). The volute is an
ever widening spiral from the cutwater to the discharge opening. The axial pump sometimes
called a turbine pump (Figure 13.2d), has the suction and discharge parallel to the impellers
axis of rotation. The flow is both axial and rotational due to propeller like action. Diffuser vanes
(stationary) are used to straighten the flow. However, it is common to prerotate the flow with
diffusers prior to entering the impellers. This type impeller is common in vertical pumps.

Figure 13.2a. Volute Pump

Figure 13.2b. Diffuser Pump

13.2

Figure 13.2c. Multistage Pump (Courtesy Worthington Pumps Inc.)

Figure 13.2d. Turbine Multistage Pump

GENERAL PERFORMANCE CHARACTERISTICS
Important operating characteristics of pumps include:
Capacity, Q gallons per minute (gpm)
Head, H feet (ft.)
Power, horsepower (HP)
Efficiency,
Speed, RPM
Impeller diameter, D (feet)
Specific weight, (lb/ft3)

13.3

Average Fluid velocity, V

Q
ft / sec
A

Elevation, Z (ft above datum) ft

Static pressure, p (lb/ft2)
Acceleration due to gravity, g 32.174 ft/sec2
The total pump head, H represents the net work done on a unit weight of fluid in passing from
the inlet or suction, s flange to the discharge flange d. It is given by Bernoulis equation

p V2

p V2

H
Z
Z
2g
d
2g
s
Unlike fans, the fluids that pumps transport are non-compressible resulting in potentially large interactive forces being transmitted between the rotating and stationary components
(Figure 10.2). In addition, under certain conditions, the liquid can vaporize, then collapse back
into the liquid state causing shock waves that can destroy the impeller of the pump. The presence of abnormal interactive forces and the above described problem called cavitation are a
function of where the pump is operating relative to design conditions (Figure 13.3).

The terms

p
, called the pressure head, represents the work required to move a unit

weight of fluid in the flow direction, V against the pressure p.

V2
is the velocity head and rep2g

resents the kinetic energy of a unit weight of fluid moving with velocity, V. The term Z, called
the elevation head represents the potential energy of a unit weight of fluid with respect to a reference.

The horsepower required for the pump is

HP
Q =

gpm

H =

ft

QH ( sp. gr. )
3960

The efficiency is the horsepower HP divided by the power input to the pump shaft.

13.4

The specific speed is given by

Ns

N Q
H

where: Q =

gpm

H =

ft.

N =

RPM

The specific speed is used to judge efficiency.

PUMP OPERATION
Figure 13.3 shows a typical pump curve that is a plot of total pump head, efficiency, and
power against flow. The curve shows the best efficiency point (BEP) in terms of flow, head, and
power components (Figure 13.3. These interactive forces can be minimized by proper selection
of the number of vanes and diffusers (Figure 13.3). Under certain operating conditions, when
the local fluid pressure is below the vapor pressure of the fluid, bubbles and/or vapor filled cavities can develop. These bubbles will collapse and cause pulse like random forces on the pump
casing when higher pressure is encountered. In turn, these interactive forces cause damage
and noise (like stones in the fluid) to the impeller and casing. This phenomena is called cavitation and is encountered when the net positive suction head, NPSH, is below the liquid vapor
pressure. Net positive suction head is determined in an installation by considering the atmospheric pressure pa, gage suction pressure, ps, vapor pressure, pop, average fluid velocity, V and
static head Zps.

NPSH

pa ps pp

Zps

V2
2g

Therefore cavitation occurs as a function of where the pump is operating relative to design conditions (Figure 13.4b).

13.5

Figure 13.3. Pump Characteristics for Backward-Curved Vanes

At best efficiency design point, fluid discharge angle matches angle of diffuser and flow is
smooth with minimal disturbances. If flow is decreased (too much back pressure) or is increased (too little back pressure), the fluid angle no longer matches diffuser angle resulting in
recirculation and higher vibration and a loss in efficiency. The dynamic head exceeds the centrifugal head generated by the impeller. As back pressure is decreased, flow through pump increases and fluid discharge angle increases causing cavitation.

Both phenomena give symptoms of noisy and erratic operation as well as damage to impeller
vanes, pump casings and bearings because of eddies and turbulence

13.6

Figure 13.4a. Flow Path Characteristics (after Baxter13.2)

Figure 13.4b. Pump Head versus Flow Curve (After Baxter13.2)

It is therefore apparent that the vibration level that is measured on a pump is highly dependent upon the operating conditions. Important parameters to monitor are:
1.
2.
3.
4.

Back (discharge) pressure

Suction (inlet) pressure
Fluid temperature
Speed

13.7

PUMP FAULT ANALYSIS

A commonly occurring problem with pumps that is often present on vertical pumps is
called rocking mode resonance of the casing. This problem is the result of the first reed (cantilever design) natural frequency of the vertical pump matching the pumps running speed, resulting in high vibration at the running speed frequency of the pump. This problem can easily be
confirmed with a resonance test Chapter 6. Table 13.1 contains some common problems
unique to pumps.
Table 13.1. Pump Fault Analysis*
SPECTRUM/
TIME DOMAIN
Recirculation
1x, vane pass
noisy time
waveform and
spectrum
elevated baseline
Cavitation
1x, vane pass
noisy data and
acoustics
Incorrect Assembly vane pass
elevated vane pass
frequency
1x, 2x or vane pass focused energy at
Structural
Resonance
resonant frequency
1x, 2x or vane pass focused energy at
Shaft Resonance
(Critical Speeds)
shaft critical speed
FAULT

FREQUENCY

Impeller Deflection

1x

focused energy at
1x

TorsionaL
Resonance

vane pass,
resonance

Trapped Foreign
Material
Excessive Wear
Ring Clearance

1x

Impeller-Diffuser
Gaps

1x, vane pass

Improper Inlet
Conditions

1x, random noise

focused energy at
vane pass frequency
increased vibration
at 1x
focused energy at
1x due to critical
speed
vibration at 1x,
vane pass, and
random noise
noisy time and
frequency

Piping Structural
Resonance
Acoustical
Resonance
(Piping)
Foundations

1x, 2x, vane pass

focused energy

1x, or vane pass

focused energy at
natural frequency

1x, 2x, vane pass

1x and orders

1x, critical speed

CORRECTION

EXAMPLE

increase flow
through pump

13.6

increase head on
the pump
reassemble casing

13.7
13.10

alter natural
frequency
alter natural frequency or fine tune
balance
increase size of
shaft or bearing
stiffness
change natural or
forcing frequency

13.14

unclog pump

13.23

return internal
components to
specification
alter gaps to optimize efficiency and
vibration
change inlet to increase straight
section length
10 pipe diameters
change natural or
forcing frequencies
Change piping path
or vane pass frequency
improve foundation
design or construction

-13.19
13.21

----

----

*Faults other than common problems such as mass unbalance, misalignment, looseness or rolling element bearing faults.

13.8

Recirculation
Recirculation (Figure 13.4) is an excellent example of what can happen to a pump that
operates against too much back pressure. For a fixed speed pump there is only one back pressure for which the flow angle of the fluid coming off the impeller matches the diffuser angle.
Operation at any other point can result in inefficient operation and excessive vibration. A pump
can experience inlet or discharge recirculation resulting in a loud crackling noise around the
pump suction.

Case history on recirculation. In a pump undergoing recirculation, antifriction bearings

were failing at 6 week intervals on a horizontal split case pump, 2250 gpm capacity at 300 ft. of
total developed head. Visual observation of the pump showed that the rotor was moving in the
axial direction at a low frequency. A random crackling noise was present.
A head curve (Figure 13.5) for the pump was requested to determine where the pump
was operating relative to its best efficiency point. Comparison of the discharge pressure from a
gauge showed that the pump was being operated at 150 psi back pressure versus 125 psi required causing a low flow far to the left hand side of the curve.
The bypass orifice was stamped indicating that it has a 2 inch opening. The pump called
for a 3 inch orifice to ensure the correct minimum flow. It was therefore recommended that the
orifice be replaced. When the orifice was removed, it was found that the hole was actually only
1 inch in diameter. Since the pump was forced to operate as low as 800 gpm, the recirculation
line needed to carry 1450 gpm even though the orifice restricted it to 150 gpm.
Pumps that are forced to operate at drastically reduced flow have pressure that builds up
on one side of the rotor, then the other, due to recirculation. This results in slowly oscillating
axial forces and vibration (Figure 13.6). This can cause rapid failure of anti-friction bearings that
were not designed to take the extra axial loading. It is recommended that all pumps which have
axial shuttling of the rotor be checked to determine if they are operating against excessive back
pressure. The pump in the above case was filling a tank several floors above. When the tank
reached a predetermined level, a control valve closed. This meant that the only outlet for the
pump was the recirculation line. Since the orifice in the recirculation line was too small, the
pump operated against too much head. The piping for the recirculation line had a flow orifice to
restrict flow to 150 gpm. This was removed, and the isolation valve in the line was used to set
the recirculation line flow at a better point until the system could be redesigned. Review of the
design calculations indicated that a flow control valve was needed in the recirculation line to
regulate the flow allowing the pump to run closer to its design point.

13.9

2400

2800

Figure 13.5. Pump Curve (13.2)

Figure 13.6. Vibration on 3BCCW Inboard Pump, Axial

13.10

Cavitation
Cavitation occurs when a pump is operated with insufficient back pressure causing large
flow and a steady low amplitude crackling noise. During baseline vibration monitoring, high vibration levels were discovered on the circulating water pumps at a utility. The vibration was detected in both the horizontal direction on the inboard motor bearing and in the axial direction of
the outboard motor bearing. The spectrum was broad band in nature with no mechanically related identifiable frequencies being observed. Figure 13.7 shows the spectrum of the vibration
on the inboard motor bearing. Mechanically related frequencies can be separated from flow
noise by synchronous time averaging.

Vane Pass

Figure 13.7. Pump Vibration Spectrum

(Courtesy of Nelson Baxter13.2)
Investigation showed that some of the pumps were only operating against 10 ft. of back
pressure. This seemed quite low so a copy of the pump head capacity curve was obtained
(Figure 13.8). The design point of the circulating water pumps was 156,000 gpm at 38 ft. of
head. The head flow curve ended at 15 ft. of back pressure indicating that operating with only
10 ft. of back pressure was not even considered by the manufacturer. An estimated flow of
200,000 gallons/minute was obtained by projecting the head capacity curve to the 10 ft. discharge pressure level.

13.11

Figure 13.8. Pump Curves (Baxter13.2)

To verify this theory, the discharge valves on the condenser were partially closed to increase the back pressure to a level nearer to the design point. When the valves were partially
closed, the vibration dropped off to an acceptable level.
When one of the circulating water pumps was removed for repair, it was found that there
was serious damage to the suction bell. Based upon the above, it was theorized that this damage was due to cavitation.
A review of the above problem found that the low discharge condition was the result of
operating with only one pump in the header rather than two. However, two pumps produced too
much flow. This mode of operation occurred when the cooling water temperature was low
enough to allow one pump to supply enough water to the condenser to satisfy the back pressure
requirements of the turbine. The unfortunate result was the one pump in operation would cavitate because the discharge head was too low.

13.12

Improper Pump Assembly

Operations complained that vibration was high on a 3600 rpm split case ash sluice pump
motor. Analysis on the pump and motor confirmed this observation. The motor and pump vibration signatures (Figures 13.9 and 13.10) clearly showed that the predominant frequency was
at five times running speed. This frequency was found to be the pump blade pass frequency.
Therefore, the problem was not in the motor but was coming from the pump. The motor 1x vibration was only .1 inches per second.

Figure 13.9. Sluice Pump Motor Vibrations (Courtesy of Nelson Baxter13.2)

Figure 13.10. Sluice Pump Vibrations (Courtesy of Nelson Baxter13.2)

13.13

Since the pump blade pass frequency was at a level of .661 inches per second, it was
recommended that the pump be disassembled. It was found that the casing had been offset
during reassembly after an overhaul, causing the high blade pass frequency.
Structural Resonance
Vertical pumps are known for their casing structural resonance problems the natural
frequency of the unit, foundations, and piping is equal to pump speed or vane pass frequency.
Shafting critical speeds can also be a problem the natural frequency of the shaft on its bearings is equal to the operating speed or vane pass frequency.
This case history on a vertical firewater pump concerns the excessive vibrations the
pump casing, Figure 13.11. From steady state and transient test data, it was determined that a
severe resonance problem existed (natural frequencies on both flanks of the operating speed
(Figure 13.12). A computer model was used to determine means of altering natural frequencies
through addition of internal and external stiffening.
H

Figure 13.11. Schematic Design Fire Water Pump

13.14

Figure 13.12. Impact Test Data

Using all available test data, Figure 13.13 shows the operating deflection shape of the
pump housing for once-per-rev pump vibration (1750 cpm) plotted in velocity (in./sec.) versus
pump elevation. The data show an operating deflection shape with high vibration at the suction
and drive ends of the pump with restraint where it is attached to the floor. Data taken at the
gearbox input shaft housing in the north/south direction, Figure 13.14, show predominately
pump generated vibration. However, the 6x engine vibration level is above good practice 6x
is a major engine order. Figure 13.15 taken in the north/south direction at the top of the gearbox, shows the severity of the problem 1.64 in./sec. rms (17.9 mils) at pump rotational speed.
A computer model of the gearbox and pump was constructed to determine the effect of
various alterations on the vibration characteristics of the pump. The effect of changing mass
and/or stiffness of the various components or adding stiffeners at various locations was studied.

13.15

Figure 13.13. Pump Operating Deflection Shapes

13.16

Figure 13.14. Horizontal at Gearbox Split Case North/South

Figure 13.15. Horizontal at Top of Gearbox North/South

13.17

The model shown in Figure 13.16 was formed from the best available information. The
pump model was verified with test data. Impact tests were used to simulate the natural frequencies, and operating data were used to simulate the forces on the pump and to establish an
operating mode shape. This model was then modified in various ways to obtain the modifications to the pump that would have the best chance of solving the problem.

Figure 13.16. Pump Model

13.18

The cause of the excessive vibration on the fire water pump as shown in Figure 13.12 is
resonance. Natural frequencies exist at 1545 cpm and 1830 cpm with the pump operating at
1750 rpm. The test data and model show that the principal vibration at operating speed is amplified by the 1830 cpm resonance. Thus, lowering the operating speed of the pump would improve the vibration levels.
Means of altering these natural frequencies to eliminate the excessive vibrations were
studied using the computer simulation model. Also, the use of a dynamic vibration absorber
was evaluated. It would appear that the addition of mass to lower the 1830 cpm natural frequency would be the easiest solution. The following actions were considered.
1. Lower the pump speed 60 rpm move forcing frequency.
2. Insert additional bolts or springs between the spacer and/or the pump and between the
spacer and the foundation.
3. Add stiffeners at the gearbox and at the water line in the sump.
4. Add mass to the gearbox.
5. Add a dynamic vibration absorber at the top of the gearbox.
The valley between the two natural frequencies shown in Figure 13.12 indicates that
lowering the pump speed by 60 rpm would eliminate some of the amplification of the resonance
by the 1830 cpm natural frequency. However, the forcing frequency would be 9% away from
the natural frequencies on each side. This would require good speed control.
The test data and the model suggest flexibility being introduced by the bolted joints at
the spacer would lower the higher natural frequency. This modification by the pump manufacturer created an unstable mounting. Additional rigidity would increase the 1830 cpm natural frequency; however, it may also increase the 1545 cpm natural frequency. Probably impossible
because stiffening could occur in the north/south direction.
It appears that stiffeners added between the gearbox base and the concrete foundation
would raise the upper natural frequency but not adversely affect the lower natural frequency in
the east-west direction. The addition of stiffeners at the water line would raise both natural frequencies above the pump operating speed.
The addition of weight to the gearbox will lower both natural frequencies; however, the
added weight must not overstress the pump casing. This was the final fix 785 lb. Added to the
gearbox split line.

13.19

A dynamic absorber, Figure 13.17, was designed for the pump. The absorber is designed to vibrate while the pump stands still. This fix is dependent on the availability of a suitable attachment point on the top of the gearbox. The dynamic vibration absorber is a massspring system with its natural frequency equal to the pump operating speed. A series of plates
(mass) are adjusted to a position on the threaded pipe (spring) where the natural frequency of
the mass-spring system is equal to the pump speed.

Figure 13.17. Dynamic Vibration Absorber

Shaft Resonance (Critical Speeds)
The motor-pump shaft will have natural frequencies that depend on the shaft sizes and
lengths, couplings, bearings, and casings. These natural frequencies can be determined by an
impact test if the shaft is supported on rolling element bearings. Otherwise natural frequencies
are determined by transient testing (Chapter 6) or computation. Critical speeds are excited by
either pump speed or van pass frequency.
Excessive Pump Shaft Deflection
A lowest bidder supplied centrifugal pump (Figure 13.18) was placed into service to
pump water against 300 ft. of head. It was determined that the head on the pump caused up to
20 mils deflection of the pump shaft. It was anticipated that this excessive deflection would
cause premature bearing failures. Figure 13.19 shows vibration data from the pump that had
retrofitted proximity probes. The data show 4 mils pk to pk vibration; however, a portion of that
response is due to a rough shaft at the probe locations. For a three inch diameter shaft
roughly 5 to 6 mils clearance, this vibration level is excessive.

13.20

Figure 13.18. Double Suction Pump

Figure 13.19. Proximity Probe Data

13.21

Torsional Vibrations
A torsional vibration analysis was conducted on a vertically mounted horizontal sewage
pump (Figure 13.20) because of a series of motor shaft failures. The 300 hp induction motors
drive 24 x 18 type SSEV pumps through a flywheel. The motor speed (400 rpm to 705 rpm)
was controlled by an adjustable speed pulse width modulated type inverter.
Torsional

vibration

tests

(Figure

13.21) were conducted on a motor over its

operating speed range. These data showed
the presence of excessive torsional vibrations
at speeds between 400 rpm and 525 rpm. It
was found from a computer model developed
for the unit that a condition of torsional resonance exists.

It is excited in a range of

speeds by the interference of the pulse width

modulation (PWM) motor excitation at a frequency of 5 times motor speed and the first
mode torsional natural frequency at 38.8 Hz
(Figure 13.21).

The computer model, vali-

dated by the test data, showed excessive

stresses sufficient to cause fatigue failure of
the motor shaft in its keyway.
Subsequent design studies were conducted using the computer model to determine how to lower the system torsional vibraFigure 13.20. Pump Configuration

tions. It was determined that a soft coupling

of the Holset type would lower the natural

frequency of the system enough to remove the resonant conditions from the pump operating
speed range. The addition of high damping in the coupling will remove the concern of passing
through the system first mode natural frequency during startup and coast down. Further, the
high damping will provide protection from torsional shocks and/or low frequency pulse excitation.

13.22

38.8

328
11.6
33
.042
Q

Figure 13.21. Torsional Peak Hold Test

A Koppers Max-C flex-rigid size 4.5 coupling, type WB with SBR high damping blocks
was evaluated within a modified motor (4 diameter shaft)-flywheel-pump configuration. It was
found to have acceptable torsional vibration characteristics.
Foreign Objects
A predictive maintenance engineering assistant noticed abnormal noise coming from an overhung high-pressure service water pump (Figure
13.22). The pump delivers untreated lake water at
2,500 gpm (1,780 rpm) (Motor 350 hp).
Aside from the abnormal noise and above
average horizontal pump vibration, all looked good.
Figure 13.22. High-Pressure
Service Water Pump

Horizontal pump vibration was .40 in./sec. on the

inboard pump bearing and .60 in./sec. on the outboard pump bearing. This is about twice what the
normal vibration.

13.23

Vibration signatures on the motor showed little vibration. Pump vibration signatures
showed all vibration was the 1x running speed frequency (Figure 13.23). Since there was no
axial or vertical vibration, this indicated some type of balance problem.

Figure 13.23. Pump Vibration

In a similar case history with identical vibration signatures, a piece of metal was lodged
in the impeller. Therefore, it was recommended that the impeller be inspected for foreign objects. The impeller was removed for inspection and a four inch (4) piece of round stock was
found in the bottom of the pump casing. This piece of metal fit snugly into a hole in the bottom
of the impeller. The pump was put back together and test run. Vibration levels dropped to normal limits.

Excessive Wear Ring Clearance

The wear rings in a multistage pump, Figure 13.2c, that prohibit flow from a high pressure stage to a low pressure stage are a seal similar to the plain journal bearing. Even though
the wear ring has greater clearance than the normal bearing they act like bearings. These
bearings help to support the multistage pump shaft and raise the natural frequencies. For this
reason, excessive wear ring clearance can cause critical speed problems by lowering the system stiffness.

13.24

Impeller-Diffuser (Volute) Gaps/Blade Pass

Figure 13.24 shows the gap of a single stage pump. The size of the gap presents a
tradeoff between high efficiency and excessive blade pass vibration. Also, the number of vanes
on the impeller and diffuser determine what magnitude of blade pass vibration measured on a
pump13.3.

Figure 13.24.

Single Stage
Pump

Acoustic Resonance and Poor Inlet Conditions

A vertical pump similar in design to that shown in Figure 13.24 had excessive vibration
at the drive bearing. Structural resonance was the suspected reason for the elevated vibration
at 3 times operating speed (vane pass frequency). However, after elaborate impact testing no
natural frequency could be identified. Figure 13.25 shows a transient test results from double
integrated accelerometers and a pressure transducer. The figure shows peaks in the structural
vibrations and pressure pulsations indicating acoustic and structural natural frequencies. Af-

13.25

ter a thorough investigation, it was found that the excessive vibration came from a combination
of resonances as well as high excitation due to poor inlet conditions. The peak vibration measured in the pump operating speed range was .37 ips. The acoustic resonance was in the discharge piping which had a length of approximately 130 ft. to the discharge header.
Therefore

fn
VS

4850 ft./sec.

130 ft.

Fn

4850
18.65 Hz
2 x 130

Vs
2L

Water: Acoustic Velocity

Solutions to solve this problem include changing the natural frequencies, tuned absorbers (13.5), and reduction to excitation.
Speed 270-390 RPM

Figure 13.25. Vibration and Pulsation at Vane Pass Frequency

13.26

FANS AND BLOWERS

Many centrifugal fans use a volute or scroll type casing the flow enters axially and
leaves tangentially. Blading may be fixed or adjustable (sometimes during operation). A typical
fan performance characteristics is shown in Figure 13.26. The basic curve is fan pressure versus flow through the system head or pressure varies as the square of the flow. The fan will
operate satisfactorily at the intersection of the system characteristics and the fan pressure characteristic (Figure 13.26). The system characteristic can be changed with an outlet damper control. Variable vane, pitch, and speed controls alter the fan characteristics. Characteristics of
fans mounted in series and parallel must be considered as a system to avoid excessive vibration and premature bearing failure.

Figure 13.26. Fan Characteristic Constant Speed

To ensure stable operation, the slopes of the pressure-flow curves of the fan and system
should be of opposite sign (Figure 13.26). When the slopes of the fan and system characteristics are of opposite sign, any system disturbance tending to produce a temporary decreased in
flow is mollified by the increase in fan pressure. The condition which accompanies unsteady
flow is pulsation that occurs when the operating point of the fan is to the left of the maximum
pressure on the fan curve (the surge point). Inlet dampers normally can be used to position the
fan operation to the right of the surge point. At low capacities, flow reversal or puffing can occur. Air puffs in and out of the let. If the back pressure is greater than the fan discharge pressure flow reversal will occur. Flow separation in the blade passages of the impeller can cause
unsteady flow and vibration.

13.27

Fans can be plagued by impeller eccentricity, asymmetric supports, loose fan wheels
and structural flaws. Improper isolation mounts often present a problem.
In addition to flow noise generation by size and directional changes in inlet ducting,
acoustic resonance can be set up when a fan vane pass frequency matches the acoustic natural frequency of the air in the duct work. Fans are subject to critical speed and structural resonance problems because of the nature of their mounting on skids, isolators, and flexible frames.
Ducting and casing resonances can be excited by high frequency variable frequency drive induced excitations.
Table 13.2 shows a number of common fan faults some of which will be illustrated
with brief case histories.

COMMON MECHANICAL PROBLEMS

Mass unbalance (Figure 13.27) and blade pass (Figure 13.28) are two common fan
problems. Balancing is frequently done with the isolators blocked. However, it is better to not
block the isolators because the goal is to minimize the vibration symptom. It should be noted
that balancing of a resonant fan is difficult and should be avoided when possible. It is best to
deal with the natural frequency prior to attempting balancing. Isolators do deteriorate with time.
If the fan has recently shown high vibration, it may be resonant due to isolator property change.
If signs of looseness are shown in the data, figure 13.29, the isolators may be worn out. Excessive blade pass vibration can come from damper position, resonance, or ducting characteristics.
Fan case histories follow.

13.28

Table 13.2. Fan and Blower Fault Analysis

FAULT

FREQUENCY

SPECTRUM/TIME

CORRECTION

Critical Speeds

1x, BP*

Focused energy at frequency of critical

speed

Change natural frequency or operating

speed

Resonance, Structural

1x, BP

Focused energy at frequency of resonance

Change natural frequency or operating

speed

Resonance, Acoustic

1x, BP

Focused high vibration

at resonant frequency

Change ducting, fan

speed, or no blades

Aerodynamic

BP or
Random Noise

High amplitude BP or
high noise flow

Change damper position or redesign ducting

Isolators

1x and Orders

Impact like waveform

and spectrum or focused energy

Replace isolators if
worn out redesign if
resonant

High amplitude 1x

Replace impeller

Difficult to balance
new lower critical
speed, high amplitude
1x critical speed at x
non repeatable data

Repair wheel

Impeller Eccentricity
Impeller Cracks

1x
1x, 2x

Rubs

Fractional Frequency
1x and Orders

Subharmonics or orders of operating

speed

Change clearance or
reduce source of rub
excitation

Surge

Natural Frequencies

Low frequency pulse

with ring down

Change operation to
get on fan curve
avoid flow reversal

Belts

Belt Frequency
and Orders

Vertical and horizontal

in phase. Pulses in
time waveform when
defect passes pulley

Replace belts

Eccentric Pulleys

1x

Directional 1x phase in
horizontal

Replace pulley

Asymmetric Pedestal
Flexibility

1x

High amplitude in flexible direction

Balance using transducer with highest amplitude

Blower Pulsation

No. of Lobes
x RPM

High amplitude of pulsation frequency and

multiples

Add pulsation damper

in discharge pipe

Loose Fan Wheel

1x and
Multiples

Walking phase

Repair keys or shrink

fits

13.29

Figure 13.27. Fan Mass Unbalance with Induced Blade Pass Frequency

Figure
13.28. Fan Operation Off System Characteristic

13.30

Figure 13.29. Support Looseness

Cooling Unit Resonance

The cooling unit shown schematically in Figure 13.30 was experiencing numerous structural failures. The unit was well built from stainless steel and adequately balanced. However,
resonances were not considered in its design. The following forcing functions were present in
the system.
mass unbalance and misalignment

1x

misalignment

2x

impeller passing

3x

propeller passing

4x

Speed of rotating unit 1200 rpm

Thus forcing frequencies of approximately 20, 40, 60, and 80 Hz were available. The
structural resonance in the unit were eliminated by detuning using resonance testing for identification and stiffness for correction.

13.31

The critical speeds of the slender shaft/propeller unit involved more than simple resonance testing due to overhung propeller. Figure 13.31 shows an interference diagram for this
unit.
It shows the stiffening effect of the gyroscopic moments caused by the whirling propeller.
Bump tests showed the propeller/drive shaft natural frequency at 22 Hz very near the mass
unbalance frequency, 19.33 Hz. Note that the first mode is actually near 35 Hz due to the large
stiffening effect from the gyroscopic moments. This means that the first mode could be excited
by a 2x vibration excitation rather than the 1x of mass unbalance. For this reason it was important in this system to avoid any 2x vibration due to asymmetric support stiffness, heavy
mass unbalance induced nonlinearities, or excessive misalignment. This case shows the extreme dependence of the lateral natural frequencies on the rotor whirling spin effect (gyroscopic moments) when overhung wheels, propeller, etc. are present.

Figure

13.30.
Port-

able

Cool-

ing

Tow-

er

13.32

Figure 13.31. Overhung Fan Interference Diagram

Overhung Blower Balance Sensitivity and Resonance

A group of 200 hp, 3600 rpm motor blower units shown in Figure 13.32, were experiencing a balance sensitivity problem; i.e., they required endless small balance corrections to keep
them operating. The subject units consisted of a large overhung fan supported on two external
pillow block bearings which were mounted on a fabricated steel base. The blower shaft was
connected to the motor with a steelflex coupling. These units experienced continued bearing
failures

and

out-

ag-

es.

13.33

Figure 13.32. Schematic Diagram of an Overhung Blower

In a typical situation, one unit was balanced to a maximum of .05 in./sec. (vertical).
Several days later the unit had excessive radial vibration. It took only 3.75 oz.-in. to rebalance
the fan. These units required this type attention until the matter was given engineering attention.
Experimental analysis using ringing and coast down tests established the nonrotating
and rotating first natural frequency frequencies at 1750 cpm and 2200 cpm respectively. The
stiffening of the unit results from the rotation of the large overhung fan. It was obvious from
these tests that the unit was operating slightly under the second natural frequency. Figure
13.33 shows the interference diagram. Operation near the natural frequency causes a balance
sensitivity problem a large vibration is obtained for a small additional amount of mass unbalance (Figure 13.34). In this case slight flexing of the fan may have been adjusting the mass unbal-

ance

enough

to

in-

crease

the

vi-

bration.

13.34

Figure 13.33. Blower Interference Diagram

Figure 13.34. Blower Bod Plot

The solution to the problem was to raise the second natural frequency enough to eliminate the balance sensitivity problem but not raise the first natural frequency appreciably. Inspection of the unit revealed flexibility of the unit in the shaft, bearings, and pedestal. It was decided that the fabricated steel base would be the easiest element to change. To raise the natural frequency it had to be stiffened. The fabricated base was turned upside down steel rebars were welded in the base and it was filled with concrete. This action raised the natural frequencies enough to eliminate the balance sensitivity problem (Figure 13.34).
Critical Speeds and Balance Sensitivity of Overhung Fans
Many severe vibration problems with overhung fans are attributable to the fact that critical speeds or support resonances occur close to the desired operating speed of the fan. This
condition can be a result of improper mounting of either the fan on flexible structural members

13.35

or vibration isolators. In some cases the fan itself operates close to a critical speed due to bearing stiffness or the shaft span.
The fact that a fan operates close to a critical speed or a structural resonance creates a
balance sensitivity problem. Any degradation of balance due to wear, corrosion, or uneven
product buildup causes a severe vibration problem. Figure 13.35 shows the amplification of the
vibration response of a system if the operating speed is close to a natural frequency. Amplification factors of five or more are not uncommon if isolators or support structures are improperly
mounted.
A small change in condition can
therefore result in large changes in vibration
level. The problem can be solved by moving
a natural frequency of the system up or
down, thereby tuning it. Depending on the
structure and the nature of the fan, movement in either direction will be advantageous.
A support-oriented natural frequency
can be determined by bump or resonance
tests (Chapter 6). A natural frequency that
is shaft or bearing oriented, however, must
be determined by a coast-down test (Chapter 6).

In this case the natural frequency

usually varies with speed because of the

gyroscopic moments exerted by the overFigure 13.35. Vibration Response
to Mass Unbalance

hung fan. The gyroscopic effect tends to effectively stiffen the shaft and raises the natural frequency.

Errors in natural frequency as high as 15 percent can occur by bumping a stationary fan.
Figure 13.36 shows an interference chart of a fan operating close to its second natural frequency. Note that a bump test on the stationary fan would have identified 50 Hz as the natural frequency rather than 60 Hz. The true critical speed is 3600 rpm.

13.36

Figure 13.36. Fan Interference Diagram

Fan Mounting Resonance

The fan-motor unit shown schematically in Figure 13.37 was mounted on isolators
through a flexible steel skid. The 50 HP, two pole motor (3,594 RPM) drove a centrifugal fan
(3,180 RPM) with belts. The excessive vibration (Figure 13.38) of the skid and ducting varied
from .7 ips to 2.0 ips causing structural and bearing failures at regular intervals.
An analysis of the vibration indicated beating between the excessive fan and motor vibration (Figure 13.38) resonance was suspected. Impact tests were conducted to (a.) determine if resonance is the problem and (b.) determine mode shapes if resonance is the problem.
Figure 13.39 shows the measurement points on the skid. Figure 13.40 shows sample two
channel data taken at point 4 in the vertical direction. Figure 13.40a shows coherence (validation of the test) and frequency response mobility (ips/lb) while Figure 13.40b shows phase and
mobility at the same point in the vertical direction. Natural frequencies were obtained at 48 Hz
and 59.75 Hz. Thus motor speed is directly on a rocking natural frequency and the fan speed is

13.37

10% off a torsional natural frequency (Table 13.3). At other points, the torsional mode was
measured at 52 Hz, which is right on the fan speed. So resonance is the principal problem.

Figure 13.37. Schematic of Fan-Motor, Skid Isolator Mounting

13.38

Figure 13.38. Fan Base Resonance (Improper Isolators)

Figure 13.39.

Measurement
Points on Skid

13.39

13.40

Figure 13.40a. Impact Test Measurements Coherence Top and Mobility Bottom IPS/lb

Figure 13.40b. Impact Test Measurement Phase (Top) and Mobility (Bottom)

13.41

Table 13.3. Summary of Natural Frequencies

Measurement
Location
Hammer Vibration

Natural Frequencies, Hz
1

14

19.5

33.7

48

4V

4V

4V

3V

19.5

40

4V

2V

19.5

39

4V

1V

19.5

39

4H

3V

15

42

52

70

90

4H

1H

15

40

51.75

72

90

4H

2V

4H

2H

1A

2A

1A

3A

Mode Type

47

59.75

90

59.75

90

55
55

90

51
16

40
25

47
50.5

55

60

72

60

72

59.5
Horizontal

Horizontal

Vertical

Torsional

Rock

Rock

88
88

Structural

Structural

A skid mounted on isolators will have six (6) rigid body (frame does not deflect) modes
vertical, horizontal (2), rocking (2), and torsional (twisting). In this case, the skid had higher frequency deflection modes 72 Hz and 90 Hz. Some frame operating deflection shapes are
found in Figure 13.41.
Recommendations for a fix included
1. New isolators do tune the two resonances.
2. Replace isolators with steel blocks to hard mount the unit.
3. Replace isolators with steel spacers and a layer of damping material.
The owner elected to use number three (3) recommendation.

13.42

Figure 13.41. Frame Operating Deflection Shapes

13.43

Fan Induced Floor Resonance

A complaint was received from a high rise building that floor vibration was causing employee discomfort. It was noted that the air handling systems on the floor above the affected
area had 50 fans operating at different times. Floor impact tests (Figure 13.42) indicated a natural frequency of 7.125 Hz. After measuring many fans, the exciter fan (Figure 13.43) was
found to be operating at 427 RPM. Therefore, this fan was exciting the floor natural frequency
below. It was found that one of the original sheaves had been replaced with one of a differing
size thus changing the fan speed to be tuned to the floor natural frequencies. A new sheave
of original diameter was installed to correct the problem.

Fan Duct Acoustical Vibration

A large ID fan operating at 930 RPM was encountering high 1x vibration in the axial (discourage) direction Figure 13.44. The vibration (about 1 ips) was causing metal fatigue and
cracks in the discharge ducting (Figure 13.45). Structural impact tests shows no natural frequency at 930 cpm. It was noted on the coast down test that the vibration would disappear almost immediately after the power was out. Calculations on the discharge duct length indicated
a possible acoustical natural frequency at fan speed. The fan speed was reduced slightly by
changing the drive sheave thus eliminating the resonance.
Fan Loose Asymmetric Supports
The FD fan shown schematically in Figure 13.46 was experiencing high vibration at operating speed in the horizontal direction. Figures 13.47 and 13.48 show data that are RMS and
synchronous time averaged respectively. The data show no difference between RMS and synchronous time averaging.

This means that all vibration is mechanical related to operating

speed.
Figures 13.47 and 13.48 shows 1x vibration with a long series of orders indicating
pedestal looseness or excessive bearing clearance. Detailed analysis of the units support
structure confirmed excessive bearing clearance. Figure 13.49 shows 1x filtered data. The vertical vibration (top) is one-third the amplitude of the horizontal vibration (bottom). The 90 phase
difference between horizontal and vertical confirms that this portion (.2 ips) of the 1x (total .21
ips) horizontal vibration was mass unbalance excited. Analysis of the pedestals shows that the
fan was not resonant in the horizontal direction but was structurally flexible. The easiest solution is to eliminate the looseness and balance the fan to the horizontal plane.

13.44

Fig-

ure
13.42 Floor Impact Test Tape Speed 10x

Fig-

ure
13.43. Fan Vibration Tape Speed 10x

13.45

Figure 13.44. Axial Vibration on the Fan

Figure 13.45. Original Duct Configuration

13.46

Figure 13.46. Fan Configuration

Figure 13.47. RMS Averaged Data

13.47

Figure 13.48. Synchronous Time Averaged Data

Figure 13.49. Fan Inboard-Filtered, Horizontal and Vertical

13.48

DYNAMIC COMPRESSORS

Due to pressures involved, most centrifugal compressors have massive casings and
small lightweight rotors that make seismic measurements difficult. Compressor faults are similar in nature to those encountered in steam turbines and pumps occurring subsynchronous to
operating speed, at operating speed, or as multiples of operating speed. For fault analysis, see
Chapter 12 on rotor and bearing faults. Compressors have a minimum flow point called the
surge limit. The operation of the machine is unstable below the surge limit. The surge limit is a
function of compressor type, gas properties, inlet temperature, blade angle and speed.

INTRODUCTION
Dynamic compressors develop a pressure differential by the action of rotating blading
that imparts velocity and pressure to the flowing medium.
The compressor is used to increase the energy of a fluid in the form of pressure. Rotating units are divided into radial flow (centrifugal) and axial flow types depending on the flow path
and the design of the impeller wheels centrifugal and blading (axial). Compressors normally
can be direct driven by steam and gas turbines. Motors use a gearbox to attain efficient compressor speeds. The larger the volume of air to be handled, the larger must be the diameter of
the impeller. On account of centrifugal stresses, top speeds are limited to available materials.
The performance of a centrifugal compressor can be stated in terms of the volume of
gas, density relative to the air at the same temperature and pressure, and the ratio of the discharge pressure to the inlet pressure. To compare compressors operating under different conditions, the desired performance is converted to an equivalent performance under standard operating conditions inlet conditions 60 F and 14.4 PSIA. The discharge pressure under this
condition is called equivalent air pressure (EAP).
The flow through a centrifugal compressor is directly proportional to speed; the head is
proportional to speed squared; and horsepower, to speed cubed.
The specific speed is a parameter used to classify compressor impellers on the basis of
their performance and proportions. Specific speed is the speed in RPM at which the impeller
would rotate if reduced proportionately in size to deliver one cubic foot of gas per minute against
a total head of one foot.

13.49

Specific Speed = Ns

N Q
H3 / 4

N = operating speed, RPM

Q = design flow, CPM
H = design head, FT
Flow coefficient and head coefficient are commonly used nondimensional parameters used for
rating an impellers performance.
For a multi-stage compressor (compressed about 3-1 per stage), the gas passes
through a crossover and enters the vanes that guide the flow uniformly into the next impeller
inlet. Machines such as the cross section shown in Figure 13.50 are split horizontally at the
centerline for reasons of maintenance. For higher pressures, the casings (Figure 13.51) are
built as complete cylinders (barrel compressor) with the rotor removed axially. The flow in axial
compressors is axial and the compression cycle involves passing gas through alternating rows
of rotating and stationary blading (Figure 13.52).
PERFORMANCE CHARACTERISTICS
The relationship between the inlet volume, flow head, speed, efficiency, and power of a
dynamic compressor are characteristic curves Figures 13.53 and 13.54 for centrifugal and axial compressors respectively. These characteristics are developed from the laws of thermodynamics and properties of gases.
From Figure 13.53 it can be seen that at any constant speed when the system resistance increases or the flow is throttled, flow through the compressor is decreased. By following the constant speed curve with lower flows, a peak head is reached that is the surge point
(surge line for multi speeds). Flow instability takes place at this point on the curve causing the
flow to pulsate operation must be maintained with flow volumes above this level. A point of
maximum flow and minimum head occurs at the other end of the constant speed curve. This is
the choked flow where the compressor impeller cannot accept any more flow volume.

13.50

Figure 13.50. Cast Case Centrifugal Compressor

13.51

Figure 13.51. Cross Section of Barrel Compressor

Figure 13.52. Axial Compressor

13.52

Figure 13.53. Typical Multistage-Centrifugal-Compressor Characteristic

Figure 13.54. Typical Axial-Compressor Characteristic

13.53

DESIGN
Since dynamic compressors rotate at high speeds (5000 RPM to 12,000 RPM) to develop pressure by rotating blading that imparts energy to the flowing medium, rotor design is a significant aspect of the compressor. The rotor consists of a shaft, shrunk on impellers, a balance
drum, and a thrust collar (Figure 13.55). The rotor for an axial compressor can be a forging or
built-up assembly with individual blades mounted (Figure 13.56). Centrifugal compressor impellers (Figure 13.57) are usually shrink-fitted to the
Figure 13.55. Centrifugal Compressor Rotor
shaft and stack balanced in the process. Sometimes these
rotors are balanced in a high speed balance facility at operating speed.
Centrifugal

com-

pressors
normally use
tilt pad bearings

Figure 13.56. Typical

Axial-Compressor Rotor

13.58)

(Figure
be-

cause of their
high speed and relative light
compressors

can

have

any

Figure 13.57. Impeller

rotors.

Axial

bearing

from

plain to cylindrical with preload to pressure dam. The seals on compressors vary according to
design and purpose including labyrinths, carbon rings, brushings, and contact seals (Figure
13.59). The relative leakage for dry sealing is provided in Table 13.4.

13.54

Figure 13.58. Typical Tilting-Pad Bearing

Table 13.4. Seal Leakage
Seal

Leakage Index

Straight Pass labyrinth

100

Staggered Labyrinth

56

Segmented Carbon Rings

20

Dry contact

13.55

a. Axial Staggered Labyrinths

b. Segmental Carbon Ring

c. Bushing Seal

d. Carbon Face-Contact Seal

Figure 13.59. Seals

13.56

MEASUREMENT AND ANALYSIS

Due to the pressure involved, most centrifugal compressors have massive casings and
small lightweight rotors that make seismic measurements difficult. For this reason, it is almost
mandatory that x-y noncontacting displacment probes be mounted on the compressor near the
bearings. Common faults encountered in dynamic compressors are listed in Table 13.5. See
Chapter 12 for additional information.
Table 13.5. Centrifugal Compressor Faults
FAULT

SYMPTOM

Excessive Bearing
Clearance

High 1x vibration with orders cannot be balanced may go

unstable and/or change critical speed

Wiped Bearings

Same symptoms as above if severity generates excessive

clearance
Excessive noise on probe signal

Rough Journals
Bent or Bowed Rotor

High 1x vibration vibration dropout at discrete speed if phased

against mass unbalance

Build-up of Product
Deposits on Rotor or Casing

High 1x and/or rubs

Unbalanced Rotor

High 1x, fixed phase angle

Shaft Misalignment

High 2x, in one direction with 1x axial

Dry Gear Coupling

Same as misalignment

Worn or Damaged Coupling

High 1x

Liquid Slugging

Random vibration including natural frequencies

Operating in Surge Range:

Insufficient Flow

Pressure impact on rotor rings at natural frequencies

Oil Whirl/Whip

Excessive vibration at subsynchronous frequencies

Subharmonic Resonance

High vibration at exact subharmonic

Gas Flow Buffeting in

Shrouded Impellers

Random noise

Piping Pulsation

Check piping for resonance between compressor vane pass frequency and piping natural frequency

The procedures used to measure the shaft vibration of a centrifugal compressor are relatively similar to those used with a steam turbine. Active thrust is toward the active thrust bearings; some compressor manufacturers counterbalance the aerodynamic thrust to place the rotor
initially on the inactive side of the float zone. As labyrinths and other seals wear, the rotor eventually wanders across to the active position.

13.57

The balance piston pressure should be measured either statically in the chamber or actively in the balance line. Additions to API 617 (1979) allow a flow measurement in the balance
line to permit better assessment of deterioration of the thrust balance. Such deterioration occurs when the balance piston labyrinths wear and more flow is required in the balancing line.
Surge detection devices determine reverse flow and surge and are used to control aerodynamic performance. They affect mechanical performance and can cause large increases in
vibration levels. The response to surge on an axial compressor causes much more damage
than that which occurs on a centrifugal compressor.

High Speed Centrifugal Compressors

High speed centrifugal compressors provide special challenges in measrement and
analysis. These units (Figure 13.60) are usually motor driven through a speed increaser with
speeds as high as 70,000 RPM. With light speed rotors and heavy casings this yields special
measurement and analysis problems. Units could have rolling element and/or fluid film bearings. Many of these machines have permanently mounted proximity probes. The noncontacting displacement probes can be used to evaluate rotor-bearing faults and condition: however,
accelerometers must be used to evaluate gear mesh frequencies which could be as high as 25K
Hz.
The accelerometer should be located as close to each stage bearing as possible in the
axial direction. Since the acceleration range could approach 50 kHz, care must be taken to select appropriate accelerometers, which have a calibrated frequency range that will cover two to
three times gear meshing frequency and ten (10) times the ball pass frequency of the inner
race.
Table 13.6. High Speed Centrifugal Compressor Faults
Mass unbalance
Gear meshing
Gear breakage (cracked, broken, chipped)
Distortion
Rolling element faults
Fluid film bearing faults
Blading/diffusers*
Alignment
Oil pump hydraulics

13.58

* Blade Pass Frequency

Blade Rate Frequency (BRF)

=
=

Nb x RPM
Nb x Nd
x RPM
K

where
Nb

number of vanes on impeller

Nd

number of diffuser vanes

highest common factor of Nb and Nd

BRF quantifies pulse rate due to compression of air between rotating and stationary
vanes.

Figure 13.60. Compressor-Gearbox Assembly

13.59

Monitoring and Analysis

Because of the high speeds of these units, monitoring overall levels is not recommended except for the permanently mounted proximity probes. Common faults are provided in Table
13.6.
It is recommended that spectrum and time waveforms be examined on a regular basis.
For added sensitivity, it may be prudent to use a log display in the spectrum so the emerging
sidebands can be identified early. Figures 13.61 and 13.62 compare linear and log plots from a
relatively new gearbox (gearmesh = 742.5 Hz, input speed = 1,790 RPM). For particular compressors, the dB range between the center frequency (gearmesh or rolling element bearing
fault), should relate to the condition. For example, if the gearmesh is 2 gs and the pinion speed
sideband is .005 gs, the dB range would be

dB 20 log

.005
52
2

This is well within the capabilities of data collectors and FFT analyzers if the dynamic
range is adjusted properly. For instance, if a 16 bit instrument is being used, the range is 2 15 or
32,768 to 1 or dB = 20 log 32,768 = 90.

If the range on the analyzer is set a 10 gs for this measurement, then the base value is

90 20 log

10
X
90

10
or
10 20
X
x .0003 gs

13.60

Fig-

ure
13.61. Gearbox Vibration Displayed on a Linear Scale

Figure 13.62. Gearbox Vibration Displayed on a Log Scale

13.61

REFERENCES
13.1.

Karassik, IlJ., Krutzsch, W.C., Fraser, W.H., Messina, J.P., Pump Handbook, 2nd Ed.,
McGraw Hill, Inc., NY, 1986.

13.2.

Baxter, N., Machinery Vibration Analysis III Notes, Volume II, Vibration Institute.

13.3.

Bolletar, U., Blade Passage Tones of Centrifugal Pumps, Vibrations, Vol. 4, No. 3, Sept
1988, pp 8-13.

13.4.

Guy, K., Case Histories: Power Industry, Vibration Institute, Willowbrook, IL, 1993.

13.5.

Wachel, B., Szenasi, F. et al, Vibrations in Reciprocating Machinery and Piping Systems, Engineering Dynamics Inc., San Antonio, TX, 2002.

BIBLIOGRAPHY
Welch, Harry J., Transamerica Delaval Engineering Handbook, 4th Ed., McGraw-Hill Inc., NY,
1983.
Block, Heinz P. and Geitner, Fred K., Machinery Failure Analysis and Troubleshooting, 3rd Ed.,
Gulf Publishing Co., Houston, TX , 1997.
Block Heinz P. and Geitner, Fred K., Machinery Component Maintenance and Repair, Gulf Publishing Co., Houston, TX, 1985.
acoustic resonance, 25
asymmetric supports, 43
atmospheric pressure, 5
average fluid velocity, 5
axial pump, 2
back pressure, 7
balance sensitivity, 36
best efficiency design point, 6
blade pass, 28
blade rate frequency (BRF), 57
Bolletar, U., 60
bypass orifice, 9
casing, 5
casing structural resonance, 14
cavitation, 5
centrifugal compressors, 48
centrifugal fans, 27
centrifugal machines, 1
choked flow, 49
compressor, 1
damper position, 28
diffuser, 2
diffuser pump, 2

13.62

diffuser vanes, 2
duct acoustical vibration, 43
dynamic head, 6
dynamic vibration absorber, 19, 20
firewater pump, 14
floor resonance, 43
flow instability, 49
foreign objects, 23
gage suction pressure, 5
gaps, 25
head curve, 9
impeller, 5
impeller eccentricity, 28
impeller-diffuser, 25
improper pump assembly, 13
Karassik, IlJ., 60
liquid vapor pressure, 5
model, 18
multistage pumps, 2
operating deflection shapes, 16
overhung blower, 33
poor inlet conditions, 25
pulse width modulation (PWM) motor excitation, 22
pump, 1
pump shaft deflection, 20
pumps, 2
recirculation, 9
shaft resonance, 20
specific speed, 5
static head, 5
stiffeners, 19
torsional vibrations, 22
total pump head, 4, 5
turbine multistage pump, 3
vapor pressure, 5
vertical pumps, 14
volute, 2
volute pump, 2
Wachel, B., 60
wear ring clearance, 24

13.63