Nuggets of Gold

• Troubleshooting Tips For diagnosing

vibration problems
• Balancing
• Alignment
• Motors
• Pumps
• Signal Processing Considerations
• DC Drives
 General Thought
WHEN HIGH VIBRATION
IS PRESENT
1: There may be a large force
2: The structure may be weak
3: Resonance amplification
may be present
 Balancing
1: When balancing, plot the original and trial vectors on all
bearings and orientations. When all vectors point towards the
same solution; you are in the right plane. When the vectors
do not point towards the same solution, try another plane. If
this still doesn't work, try a multi-plane solution.
Vectors Agree In weight adjustment

Vectors disagree in weight adjustment

 MOTOR GENERATOR SET ON DRAG LINE
SIX ROTORS WITH SEVEN SHARED BEARINGS SOLIDLY COUPLED ON A METAL DECK.

1 2 3 4 5 6
7 GENERATOR GENERATOR MOTOR GENERATOR GENERATOR GENERATOR
NO. 1 NO. 2 NO. 3 NO. 4 NO. 5

ORIGINAL READINGS
BRG. 1 BRG 2 BRG 3 BRG 4 BRG 5 BRG 6 BRG 10
H 7.8 45 DEG 5.4 45 DEG .8 80 DEG 1.5 220 DEG 1.7 253 DEG 2.1 245 DEG 2.6 260 DEG
V 10.3 50 9.1 50 9.7 45 8.6 47 5.6 63 2.5 107 5.1 180
BALANCE SHOT ADDED TO No. 1 GENERATOR 8.25 OZ at 310 DEGREES
H 5.9 37 3.9 35 .6 45 .8 270 .9 270 .8 270 1.1 45
V 11.0 45 9.6 45 10.1 36 8.8 40 5.4 45 1.1 117 5.9 200

After a weight was

NOTE !- LEVELS WENT DOWN IN THE HORIZONTAL DIRECTION, BUT UP IN THE VERTICAL DIRECTION

installed,Horizontals went
UNCOUPLED GENERATORS 1 & 2 FROM MOTOR
H 6.8 25 4.5 34 1.0 350 1.6 270 1.8 270 2.0 243 2.7 236
V 7.9 68 6.5 63 6.6 45 5.9 45 3.8 45 1.9 45 1.8 270

down Verticals went down.

NOTE THAT EVEN UNCOUPLED, GENERATOR NO. 1 IS OPERATING WITH ALMOST 8 MILS OF VIBRATION
PERFORMED RESONANCE TEST- STRUCTURE FOUND TO BE OPERATING NEAR RESONANCE

This is a strong sign that you

ADDED 50 OZ BALANCE WEIGHT TO MOTOR , ALL LEVELS DROPPED TO BELOW 3 MILS.

are adding weight to wrong

plane
2: When a rotor runs above its 1st
critical and there is an indication
of a bow, translate a portion of the
static balance component from
the ends to the center of the rotor.
If this is not done, the rotor will
run good on a balance machine
but bad in its own bearings at
high speed. This is due to
internal bending moments
produced by unbalance forces
acting on the axial distance
between the mass unbalance and
the correction weights.
BOWED FLEXIBLE ROTOR REQUIRING MID SPAN
SHOT
END PLANE SHOTS MID SPAN SHOT

CREATES INTERNAL BENDING IF SHOT IS DIRECTLY ACROSS

MOMENTS FROM BOW NO BENDING MOMENTS
ARE CREATED.
WHEN A ROTOR OPERATES BELOW ITS FIRST BENDING MODE, IT ACTS LIKE A
RIGID BODY AND THE BALANCE WEIGHT CAN BE ADDED ANYWHERE ALONG THE
ROTOR. IF, HOWEVER, THE ROTOR OPERATES ABOVE ITS BENDING MODE, IT
BECOMES A FLEXIBLE ROTOR AND THE WEIGHT NEEDS TO BE PLACED OPPOSITE
THE HEAVY SPOT TO PREVENT INTERNAL BENDING MOMENTS.

CASE HISTORY: A STATIC BALANCE SHOT IN THE END PLANES WAS INSTALLED
ON A HIGH PRESSURE TURBINE ROTOR WITH A BOW. 18 OZ WERE ADDED IN
EACH END, WITH ALMOST NO EFFECT. WHEN WEIGHT WAS ADDED IN THE MID
SPAN, THE ROTOR WAS EASILLY BALANCED.
3: When balancing 2 pole
motors which are above 1000
HP, beware of thermal
vectors. This class of rotor will
operate well uncoupled, but
will often have high levels of
vibration when pulling full
current. This is because the
rotor can bow as a function of
heating by the current flow.
Corpoven Refinery
Venezuela
4: During a startup, if a high speed compressor
has low response as it passes through its
critical, but the level increases steadily with
RPM, without much of a phase shift, then
suspect unbalance in the coupling. On a polar
plot, the response line will point straight outward
because the amplitude increases without any
shift in phase.
COMPRESSOR WITH UNBALANCE IN COUPLING
PROBE

45 315
COUPLING IS OVERHUNG

PHASE
90 270
UNBALANCE IN COUPLING AFTER BEFORE

MAGNITUDE
135 225

7000 RPM COMP.

180
BEFORE BAL.
UNBALANCE IN ROTOR 3.8 MILS 240 DEG
WT. ADDED
23.7 GM @ 50DEG
AFTER BALANCE
.6 MILS 203 DEG
5: When balancing a large machine
with multiple rotors, if there is no other
clear indication, then on the first trial
weight attempt, add weight in the rotor
with the largest inertia. Learned from
Art Crawford. Once that is done, refer
to balancing tip No. 1
 MOTOR GENERATOR SET ON DRAG LINE
SIX ROTORS WITH SEVEN SHARED BEARINGS SOLIDLY COUPLED ON A METAL DECK.

1 2 3 4 5 6
7 GENERATOR GENERATOR MOTOR GENERATOR GENERATOR GENERATOR
NO. 1 NO. 2 NO. 3 NO. 4 NO. 5

ORIGINAL READINGS
BRG. 1 BRG 2 BRG 3 BRG 4 BRG 5 BRG 6 BRG 10
H 7.8 45 DEG 5.4 45 DEG .8 80 DEG 1.5 220 DEG 1.7 253 DEG 2.1 245 DEG 2.6 260 DEG
V 10.3 50 9.1 50 9.7 45 8.6 47 5.6 63 2.5 107 5.1 180
BALANCE SHOT ADDED TO No. 1 GENERATOR 8.25 OZ at 310 DEGREES
H 5.9 37 3.9 35 .6 45 .8 270 .9 270 .8 270 1.1 45
V 11.0 45 9.6 45 10.1 36 8.8 40 5.4 45 1.1 117 5.9 200

NOTE !- LEVELS WENT DOWN IN THE HORIZONTAL DIRECTION, BUT UP IN THE VERTICAL DIRECTION
UNCOUPLED GENERATORS 1 & 2 FROM MOTOR
H 6.8 25 4.5 34 1.0 350 1.6 270 1.8 270 2.0 243 2.7 236
V 7.9 68 6.5 63 6.6 45 5.9 45 3.8 45 1.9 45 1.8 270
NOTE THAT EVEN UNCOUPLED, GENERATOR NO. 1 IS OPERATING WITH ALMOST 8 MILS OF VIBRATION
PERFORMED RESONANCE TEST- STRUCTURE FOUND TO BE OPERATING NEAR RESONANCE
ADDED 50 OZ BALANCE WEIGHT TO MOTOR , ALL LEVELS DROPPED TO BELOW 3 MILS.
6: Do not attempt to balance when the phase
is moving. This is a sign that there is a rub.
Machines that operate below a critical tend
to bow into the rub and the rub gets worse
with time. Machines that operate above a
critical can bow away from the rub causing
the phase angle to continually move against
rotation. Note that if a above critical
machine has a light rub, it can be a bad idea
to shut it down, because then it will have to
coast down through its critical speed.
7: When the horizontal and vertical
phase are the same or 180 degrees
out, then look for rocking or a loose
base. Another thing to consider is that
if a machine is operating between a
vertical and horizontal natural
frequency then his can also cause
unusual phase relationships between
the vertical and horizontal directions.
6 CASES OF LOOSENESS
Case History 1- The phase on a turbine bearing
was identical in the vertical and axial directions
with the axial vibration being very high. It was
discovered that one of the hold down bolts was
broken off beneath the surface of the concrete
allowing the bearing to rock.
Case 2- On a large fan, the horizontal and vertical vibration phase
angles were identical. The base bolts were loose allowing the bearing to
rock. The maintenance manager did not believe it so a cup of water was
poured on the base next to the bearing. When the water alternately shot
out from between the bearing housing and the base plate, he agreed to
have the bolts tightened. The horizontal vibration dropped from 12 mils
to less than 2 mils.
CASE 3- A power plant had spent $30000 on a mill motor trying to
get the vibration levels reduced. The rotor had been balanced
several times, but the amplitude was still high.
 APPARENT COUPLE UNBALANCE IN A
FAN THAT OPERATES BELOW 1st
NATURAL FREQUENCY

POSSIBLE CAUSES
DISSIMILAR PEDESTAL STIFFNESES
– WRONG PLACEMENT OF VIBRATION PICKUPS
– LOOSE BASE BOLTS
– PHASE REVERSAL WITHIN ONE PICKUP

• NOTE- IF THERE IS A LARGE COUPLE COMPONENT IN A

FAN THAT OPERATES WELL BELOW ITS 1ST CRITICAL, THEN
BE VERY SUSPICIOUS AND AVOID INSTALLING A COUPLE
SHOT
EXAMPLE OF A LARGE FAN WITH
APPARENT COUPLE UNBALANCE

Vectors showed
what appeared
to be a large
amount of
couple
unbalance
 SOLUTION
• The large couple component raised the level of
suspicion.
• The results of a previous balance person showed
the response to be highly non-linear
• The base bolt tightness was therefore checked
and all the bolts were all found to be significantly
loose.
• The bolts were tightened, the couple component
disappeared and the levels dropped to 1/4th their
original values.
• Following bolt tightening, the fan was then easily
balanced.
PLUNGER BOLT HOLDS BEARING TIGHT WITHIN HOUSING.

IF PLUNGER BOLT IS NOT TIGHT, THEN

BEARING WILL MOVE RELATIVE TO
HOUSING. THIS MOVEMENT RESULTS
IN A NON-LINEAR SYSTEM. BALANCING
IS VERY DIFFICULT, IF NOT IMPOSSIBLE
IN THIS SITUATION.
LOOSE BEARING IN HOUSING COMPLICATES
BALANCING
IF A BEARING IS LOOSE IN ITS
HOUSING, A NONLINEAR SYSTEM IS
PLUNGER BOLT
CREATED AND BALANCING IS
BEARING ALMOST IMPOSSIBLE.

2 MILS---- CASE HISTORY

SHAFT A LARGE ID FAN COULD NOT BE
BALANCED. THE CASING LEVELS
17 MILS WERE SLIGHTLY OVER 2 MILS,
HOWEVER, WHEN SHAFT STICK
LEVELS WERE MEASURED, THE
BEARING HOUSING AMPLITUDE WAS GREATER THAN 17
MILS. SINCE THE SHAFT TO INNER
BEARING CLEARANCE WAS ONLY 8
MILS, THIS MEANT THAT THE
BEARING HAD TO BE MOVING IN THE
HOUSING. WHEN THE PLUNGER
BOLT WAS TIGHTENED, THE CASING
LEVEL ROSE TO 21.5 MILS. AFTER
THE TIGHTENING, THE FAN WAS
EASILY BALANCED.
8: When balancing a rotor and the phase suddenly shifts 180 degrees,
then this is a sign that the rotor may be loose. Case History- While
balancing a large centrifuge, with a strobe light the vibration vector
changed from 3 mils at a given phase angle to approximately that amount
180 degrees out. The change was instantaneous. It was discovered
upon examination that the tapered fit of the shaft and bowl assembly was
loose.

Loose fit
between bowl
and drive shaft
 LARGE POWER
GENERATION GAS
TURBINES
• A different sort of an animal
• Balancing Cross effects are often
much larger than direct influence.
 DIRECT AND CROSS
EFFECT
WESTINGHOUSE 100
MW 501 GT
Exhaust End Shot on Exhaust End 18.6 oz/mil 25 Degree lag
Exhaust End Shot on Compressor End 4.7 oz/mil 46 degree lag

Compressor end shot on Compressor end 18 oz/mil 353 degree lag

Compressor end shot on Exhaust end 8.6 oz/mil 30 degree lag
 Weight added
here has 4
times as much
effect on
compressor
end.
ALIGNMENT
1: When aligning gear boxes with sleeve
bearings, beware of pop up pinions. Their
contribution to the shaft alignment does not
show up when you take hot alignment readings.
Their effect also does not show up on laser
alignment systems mounted on the cases or via
optical means. Source: Charlie Jackson

Case History: A speed increaser

Gearbox at a refinery had an input of
approximately 1200 RPM and an output of
9600 RPM. The amount of upward movement
of the pinion was greater than the tolerable
misalignment for the short high speed coupling.
The amount of shaft movement vertical and
horizontal was included in the alignment
settings. The unit operated for several years
without any problem or excessive coupling
wear.
 2: When aligning turbines setting
on condensers, beware of vacuum
draw down. It can be a much
greater effect than thermal growth.

1- A large turbine was experiencing oil whip. As the unit was

brought to speed, at almost exactly twice the critical speed, an
approximately ½ running speed component appeared as the speed
continued to increase, the frequency of the instability remained
locked at the critical speed frequency. A complex glycol proximity
probe alignment system was installed to measure the bearing
movement. When condenser vacuum was applied there was a
.016” difference in elevation between the No. 2 and No. 3 bearings.
 500 MW TURBINE OIL
WHIRL-WHIP
• When turbine would be shut down, if
it was not started up within 2-3
hours, it could not be started up for
two days because of excessive
vibration.

• SOUNDS LIKE XENON POISONING

ON A NUCLEAR REACTOR
 MACHINE LAY OUT
HIGH VIBRATION ON THIS
BEARING WOULD TRIP
TURBINE ON HOT RESTART

HP TURBINE LP-1 TURBINE LP-2 TURBINE GENERATOR

CONDENSER
VIBRATION SPECTRUM
MAP PLOT

NOTE AS SPEED CHANGES WHIRL

FREQUENCY DOES NOT. IT IS LOCKED
ONTO ROTOR’S NATURAL FREQUENCY.
THIS IS CALLED OIL WHIP
ALIGNMENT TEST SYSTEM
DIAL INDICATOR MEASURES
DC MOVEMENT

READOUT SHOWS MOVEMENT

OF FLOAT WHICH IS COMPARED
TO DIAL INDICATOR READING.

SUPPLIES DC OFFSET AND AC

MOVEMENT
HEART OF SYSTEM
FLOAT WITH METAL TARGET MOVES
WITH CHANGE IN FLUID ELEVATION.
PROXIMITY PROBE MEASURES CHANGE
IN GAP DISTANCE.
SETUP ON TURBINE
REFERENCE PICKUP TO ACCOUNT
FOR FLUID EXPANSION OR LOSS
 TEST RESULTS
MILS MOVEMENT VERSUS VACUUM

16 MIL
DIFFERENTIAL
 DISCUSSION
VACUUM DRAW DOWN COMBINED WITH THERMAL
DIFFERENTIAL GROWTH UNLOADED BEARING CAUSING
IT TO GO UNSTABLE. WHEN BOTH WERE COLD,
IT WOULD BE STABLE. WHEN BOTH WERE HOT,
IT WOULD BE STABLE. THE PROBLEM OCCURRED
AFTER A TRIP, WHEN THE THINNER LP SECTION
WOULD COOL DOWN QUICKER THAN THE THICK
HP SECTION. THIS DIFFERENTIAL ADDED TO THE
VACUUM DRAW DOWN WAS TOO MUCH.
BEARING METAL TEMPERATURE
CONFIRMED THIS FINDING.
SOLUTION

TILT PAD
BEARING
FINAL RESULTS
. Case 2- Two boiler feed pumps were having vibration problems and
wearing out their gear couplings. When dynamic alignment was
performed between the turbines and pumps, it was discovered that
when vacuum was pulled on the turbines that they dropped .020”
relative to the pumps.
COUPLINGS BEING DESTROYED ON STEAM
GENERATOR FEED PUMP AT NUCLEAR
STATION
VIBRATION SPECTRUM

5.8 Mils of 2X
VIBRATION
ORBIT DISPLAY
 PROBLEM-ALIGNMENT
SPEC. WAS WRONG
• Vacuum draw down was 20 mils
• Even though pump was center
mounted, it was growing.
• Turbine was growing unevenly

PUMP NEEDED TO BE SET LOW RATHER THAN HIGH

CHILLER TURBINE WAS
DESTROYING BEARINGS
UPPER HALF OF
BEARINGS WERE
FAILING BY FATIGUE
 MACHINE SETUP

TURBINE

CHILLER
MONITORING OF
ALIGNMENT
 MACHINE LAYOUT
COMPRESSOR

TURBINE

EXPANSION JOINT
 TEST RESULTS

COMPRESSOR

TURBINE

EXPANSION JOINT
DIAL INDICATORS SHOWED THAT
THERE WAS NO MOVEMENT
ACROSS JOINT
 ACTUAL PROBLEM
LASER
MOUNTED ON
BEAM When compressor started up,
this pipe cooled down causing
LASER compressor to rock back. This
MOUNTED ON
made turbine appear to drop
COMPRESSOR
down relative to compressor.

TURBINE
3: If a machine operates well for a few weeks following an overhaul,
then 2X running speed shows up in the proximity probe spectrum,
suspect a locked coupling. The amount of misalignment may not have
changed. The problem is that either the coupling grease has broken
down or escaped from the coupling.
PUMPS
 LOMAKIN EFFECT

A LARGE PUMP WOULD OPERATE FOR A

YEAR OR SO, THEN THE VIBRATION WOULD
GET HIGH. THIS HAPPENED REPEATABLY.
AFTER THE OVERHAULS, THE PUMP WOULD OPERATE
OK FOR A FEW MONTHS. THE ROTOR WAS NEVER
FOUND TO BE OUT OF BALANCE.
 PUMPS
1: Pump seals can act like bearings and stiffen the shaft to the
point that the critical speed will be pushed out beyond the
operating speed. When the seals wear, the critical speed may
move back into the operating range. This seal stiffening
phenomenon is often referred to as the Lomakin effect.
Case History- A large feed pump had a history of operating well after it
was overhauled with new seals, but after time the running speed vibration
would increase. A system was set up to monitor the amplitude and phase
as the pump was brought to speed. It was discovered that after the seals
experienced wear that the pump was operating just below a rotor critical
speed.
 TEST RESULTS

PEAK HOLD PLOT

SHOWS THAT
VIBRATION DROPS
OFF RAPIDLY WITH
SPEED.
LOMAKIN EFFECT

WHEN SEALS WEAR,

ROTOR STIFFNESS DROPS
CAUSING NATURAL
FREQUENCY TO DROP INTO
OPERATING SPEED RANGE.
2: Pump seals can unload the bearings making them unstable
or reduce their damping. Running speed levels will be much
higher than normal and will be unstable. Due to the low
loading, even though the vibration response is high, the
machine can often operated indefinitely.

Case History- A steam generator feed pump at a nuclear

power plant had above normal levels of vibration present on its
proximity probes, but there was never any damage to the
bearings. An analysis of the pump was performed and it was
determined that the seals were supporting the shaft to the
point that the bearings were only carrying a small fraction of
the rotor’s weight. Collaborated on this case with Malcolm
Leader who did the rotor analysis.
3: When analyzing pumps, measure the suction
pressure, the discharge pressure, calculate the total
developed head then look at the pump head curve
before doing anything else. Case History 1- Three
identical pumps were operating side by side. One of the
pumps was failing bearings every few weeks. A study
of the suction and discharge pressure showed that the
pump was operating at near shut off head conditions.
The problem was that during recirculation conditions,
the flow was way too small due to the presence of a
one inch orifice plate instead of the required three inch
orifice plate. Replacement of the orifice plate solved the
bearing failure problem.
FLOW IS BALANCED SO THERE SHOULD BE NO THRUST
 2- Several large circulating water pumps had high
levels of broad band vibration followed by failure of
the cases and impellers. It was found that the
discharge pressure was one third of the design point,
meaning that the pumps were operating in a run out
condition. The problem was that at times only one
pump was in operation instead of the two pumps that
were considered in the initial design.
4: If multiple pumps are
in the same header,
then if one pump is
dominate, then it will
force the weaker pump
back on its flow head
curve.
5: Vertical pumps have a high incidence of
resonance problems. Always test for
resonance in the direction in line with the
discharge line and in a direction 90 degrees
out from that orientation. Discharge lines can
stiffen the pump and the cutout that allows
the coupling to be removed can weaken the
structure. The combined effect of the
discharge line and access hole cutout
results in vastly different natural frequencies
in the two orthogonal directions. The result
can be 20 mils vibration in one direction and
2 mils in the other direction.
NATURAL FREQUENCY OF LARGE VERTICAL PUMP
IN LINE WITH DISCHARGE
90 DEGREES OUT FROM DISCHARGE
MODE SHAPE OF 480 CPM MODE
AMPLIFICATION FACTOR CALCULATION USING LOG
DEC APPROACH
6: One of the most important settings on a
vertical pump is the lift. The lift is the
distance between the impeller and the
stationary components. It is determined by
measuring the gap in the coupling before the bolts are
tightened and the impeller is lifted off the bottom of the
pump. If the lift is too much, then the pump will be inefficient
and will not produce the desired head or flow. If the lift is to
small, then the impeller can contact stationary components.
VERTICAL PUMP
7: At the Best Efficiency Point, the
angle of the fluid coming off the
impeller matches the angle of the
diffusers. Operating off of this
point will result in lower efficiency
and higher levels of vibration.
8: If the discharge valve of a large pump is not completely
closed, then this can cause the startup time to increase,
resulting in an over current trip.

Case history- A water company could not get a large

vertical pump to start. They brought in electrical experts,
but everything checked out properly. It was suggested by
the vibration analyst that the valve might be leaking.
When it was examined, it was discovered that the seal
was damaged. When the valve was repaired, the motor
started with no problem.
 Sleeve Bearings
Important Points
1: Improper bonding of the babbitt to the
base metal will cause sleeve bearings to run
hot because of poor heat transfer in the
areas where the bonding was not complete.
In many cases, you will see fans and air
movers blowing air across the bearing to
cool it, when the actual cause of the high
reading at the thermocouple is poor bonding.
An ultrasonic exam of the bearing will
quickly identify the problem.
2: A good rule of thumb is that a machine that
operates with shaft motion levels of less than
25% of clearance is running with acceptable
amplitudes. Above 30% is a low level alarm
and above 40% is a high level alarm. Levels
greater than 50% will cause rapid fatigue of
the babbitt. Source: Jim McHugh
3: When an axial resonance is suspected on a machine
with sleeve bearings, it is necessary to rotate the shaft
during the impact test. This gets the shaft up on the oil
film and decouples the end bell from the shaft, so that a
realistic natural frequency can be determined. Case
History- A large feed pump motor had .6 in/second of
axial vibration. Resonance was suspected. When a
static impact test was performed, there did not appear
to be a problem. However, when a strap wrench was
used to rotate the shaft and get the shaft up on the oil
film, a natural frequency appeared at almost exactly
running speed. This is a very common occurrence and
the number of times this has occurred are almost too
numerous to list.
RESPONSE TO IMPACTS IN AXIAL DIRECTION ON
LARGE MOTOR WITH SLEEVE BEARINGS
RESPONSE ON END BELL IN AXIAL DIREXCTION
WHILE TURNING SLOWLY AS COMPARED TO
SITTING STILL

6 TIMES AS MUCH RESPONSE WITH

SHAFT TURNING ON OIL FILM
RESPONSE TO IMPACTS IN AXIAL DIRECTION ON
MOTOR WITH SLEEVE BEARINGS WHILE SHAFT
IS STATIC AND TURNING
4: If excessive clearances
are suspected in a machine
with sleeve bearings, then
put a dial indicator on the
shaft and do a lift check.
SIGNAL PROCESSING
1: When performing resonance
tests, do not use the Hanning or
any similar windows, unless the
impact and response signals are
delayed to move them to the
center of the time block. Since
the response is maximum after
the impact, this data must be
moved away from the edge of
the time block where it would be
destroyed or severely attenuated
by the Weighting factor. Use of a
Uniform Window does not
require a delay, because it does
not attenuate the signal at the
beginning or end of the time
block.
 372 HZ IS CLOSE TO
360 HZ SCR FIRING
FREQUENCY

A customer had tested this

motor and did not find a
resonance near 360 HZ.
They had used a Hanning
window that destroyed the
data at the beginning of the
time block.
2: The FFT is a batch process. Impacts or
other transient processes which occur in time frames
which are short compared to the period of the analysis
time block result in significant amplitude errors
regarding peak values. Therefore a user should
always take a look at the time domain when
transients are present. THIS IS WHY IT IS
EXTREMELY IMPORTANT TO LOOK AT
ACCELERATION TIME WAVES WHEN IMPACTS
ARE PRESENT
3: A simple way to determine a mode shape is
the take a transfer function at equally spaced
points along the structure, then plot the
normalized amplitudes of the imaginary
components above the location of the test
points on a scaled plot. Note that this works for
acceleration and displacement data. If velocity
data is used, plot the normalized values of the
real part of the transfer function.
MODE SHAPE OF 480 CPM MODE
4: Beware of fat peaks- A fat peak
can be the result of a beat,
modulation, speed changes or a
resonance being excited.

Note that the amplitude of the peak

in the spectrum can be significantly
in error as compared to the actual
peak value. This is because the
energy is spread out over several
cells. Supplied by Jack Frarey.
5: When attempting to measure
the loc decrement to determine the
damping, if the desired frequency
is low, then if possible, use analog
integration so the time plot will be
in displacement. This technique
lowers the influence on the time
plot of the higher frequencies
which might dominate if an
accelerometer is being used.
 NATURAL
FREQUENCY
1770 CPM

N0=.0914, N10=.027
Time decay in g’s. Note
PSI=1/10*1/(2*PI)*ln(.0914/.027)=.0194
the presence of higher
Q=1/(2*.0194)= 25.7 frequency in time plot.
AMPLIFICATION FACTOR CALCULATION USING LOG
DEC APPROACH WITH ANALOG INTEGRATION

Time plot is in
displacement. Note that
no high frequencies are
present.
6: Long time samples are useful when low
frequency beats are present.

7: Long time samples are also useful when

rapid transients occur on a random basis.

8: Long time samples are bad when the

frequency is shifting.
Vibratory conveyors that were shaking houses ½
mile from foundry. Calculated Peak-Peak 4.1 mils.
Actual P-P over 7 mils. People feel much higher
levels than spectrum indicates. Long time block, in this
case 20 seconds,
DALT - DINING ROOM TABLE AT RHODES
allows time for all
P-P Displacement in Mils

DINING TBL-E-W EAST WEST

4.55
ROUTE SPECTRUM
3.5 components
23-JUN-99to phase

4.40
13:11:40
3.0
2.5
together so worse
OVRALL= 4.15 D-DG
P-P = 4.13
2.0 case can
LOAD =be
100.0seen.

4.71

4.94
RPM = 360.

5.07
1.5
RPS = 6.00
1.0
0.5
0
2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0 7.5
Frequency in Hz
Displacement in Mils

6
ROUTE WAVEFORM
4 23-JUN-99 13:11:40
P-P = 4.31
2
PK(+) = 3.93
0 PK(-) = 3.96
CRESTF= 2.60
-2

-4

-6
Time: 8.848
0 2 4 6 8 10 12 14 16 18 20 Ampl: 3.931
Time in Seconds
9: If a signal is clipped, then this effectively introduces a DC shift
into the data. This can prove disastrous if the data is then
integrated. Example:
A high output accelerometer was used to measure vibration on a
pump. The signal was integrated into velocity. Unfortunately, the
pump cavitated. The cavitation overloaded the accelerometer’s
electronics introducing a DC shift. The overall output of the pump
then fictitiously read several inches per second and all the alarms
went off. Source: Jack Frarey
10: When trying to separate two closely spaced
signals, remember that the frequency resolution is not
the number of lines divided by Fmax. It is not even
the number of lines divided by Fmax X the weighting
window factor. It is actually the number of lines
divided by Fmax X the widow factor X 2. If the factor
of 2 is not included, then the modulating effect of the
window function can create false sidebands that are
mistaken for actual frequencies. Source: Jack Frarey
A BEAT THAT COMPLETES ONE CYCLE DURING TIME BLOCK-
CENTERED IN BUFFER

Amplitude 1.46 units

Peak motion 1.69 Units from time

plot
A beat with one cycle in time buffer-Peak Amplitude at
beginning and end of buffer-minimum at center

Actual peak motion 1.69 units

Amplitude .6 units.
Note false resolution
caused by Hanning
Window.
11: When viewing a spectrum and time plot, there
are often times when things do not seem to add up.
For instance, the spectrum may show low levels
and generate no concern. On the other hand, the
time plot might show very high amplitudes. One
reason for this is the fact that the time based data
was sampled at a frequency that is equivalent to
2.56 Fmax. According to the Nyquist sampling law,
this is sufficient to pick up frequencies at 1.28
Fmax. This means that the time data can see
higher frequencies than are displayed in the
spectral plot. Example: When viewing data from a
motor, the spectrum showed nothing over .03
in/sec. In the acceleration time plot levels as high
as 8 g’s were observed. The maximum frequency
was set at 1000 HZ. The actual problem was that
the 1750 RPM motor had rotor bars generating a
signal just above 1000 HZ . The vibration was
visible in the time plot, but was beyond the 1000 Hz
Fmax, so it was not seen in the spectrum.
Solution: When this type of situation is
encountered, then increase the Fmax.
12: A similar discrepancy can occur when analog
overall levels are compared to spectral data or overall
values computed from the spectral data. The analog
overall value includes energy all the way out to say
20000 Hz. The digital overall value only includes those
components that are in the calculated spectrum. If there
is a discrepancy, then as stated in the previous topic,
increase the Fmax to determine what is causing the
difference.
 RESONANCE
1: When performing a resonance test,
if a peak shows up at the frequency of
interest, but the phase shift is small,
then this is an indication that there is
a resonant component that may be
located some distance from where the
test is being performed. For instance
if a section of pipe is tested for a
resonance and there is a peak, but a
small phase shift, then there might be
for example a nearby control valve
that is resonant. If the control valve
itself is tested, then the normal 180
degree phase shift would be present
when its natural frequency is excited.
RESONANCE TEST ON VERTICAL PUMP WHERE
PUMP IS RESONANT AT RUNNING SPEED AND
RESPONSE WAS MEASURED ON MOTOR

Note the presence

of a small peak at
3600 and small
phase shift
RESPONSE IS MEASURED ON PUMP

3600 CPM
RESONANCE
2: When anchor bolts or rebar break under the
surface of concrete, this reduction in stiffness
can alter the natural frequency and result in
resonance problems. Case History- Three
identical pumps were installed. On one pump,
the 120HZ vibration on the motor was several
times higher than on the other two. When the
motor positions were swapped, the problem
always occurred in the same location, indicating
that problem was location related instead of
motor related. A resonance test showed that
there was a resonance near 120 Hz at the
location with the high levels, but not at the other
two locations. When the foundation was broken
up, it was discovered that the re-bar was broken
in the foundation that had the problem.
3. When doing impact tests, beware of
trying to get too much resolution. For
instance, if you have a vertical pump that
has a suspected resonance at 10 HZ and
you choose a 100 HZ Fmax with 800 lines of
resolution, then the sample time is 8
seconds. If the response decays away in 1
second, then there will be 7 seconds of
noise present versus 1 second of good data.
The phase shift will look rough and the
coherence will be low. If on the other hand a
500 Hz Fmax and 400 lines of resolution are
chosen, then the sample time will be .8
seconds. In this instance, the data will be
much cleaner.
 4: The opposite situation could also be
true. If a lightly damped component is
excited then it may continue to ring clear to
the end of the time block. This can cause
problems when the FFT is performed
because a discontinuity is introduced. In
this case an exponential weighting factor
may need to be introduced to drive the
response to zero and eliminate the
discontinuity. It has to be noted that if a
log decrement calculation is made on data
that has been modified by an exponential
weighting factor that the answers will be
wrong.
Solution- Use exponential weighting factor
when viewing spectrum, but shut it off
when viewing the time domain.
5: A convenient way to locate support beams in a floor is to perform an
impact test and look for a reversal in the direction of the imaginary
components.
INDUCTION MOTOR CURRENT
TESTING
1: When taking spectrum of the current, measure the
ratio of the lower number of poles times slip frequency
side band in dB to the level of the line frequency current
in dB . If there are no other recommendations, then use
the 54-45dB rule. If the side band is more than 54 dB
below the line frequency signal, then the rotor is
probably OK. If the side band is less than 45 dB below
the line frequency, then the rotor is probably bad.
NOTE DATE IS
1982
2: Beware of cast aluminum
rotors. Cast aluminum rotors
will sometimes have voids
that will cause false positives
of the above described side
band test. When in doubt,
test the motor over several
starting cycles to determine if
the level is stable or getting
worse.
3: Pole modulation- If the number of spiders in
the rotor equals the number of poles, the current
will modulate and look just like a broken rotor bar
is present. The way to tell if this is the case is to vary the load
on the motor. If it is pole modulation, then the side band ratio will
be higher at low load. If there is an actual broken bar problem, the
opposite will be true. When a broken bar is present the degree of
modulation will increase at higher loads. Case History- A power
plant had 8 pole motors on its FD fans. Every year a current
spectrum test was performed to identify broken rotor bars. It was
noted both FD fan motors had indications of what appeared to be
broken rotor bars. The interesting thing was that the modulation
was less at high loads than at low loads. The cause of the
modulation turned out to be pole modulation. The motors ran for
many years and never had any problems, even though an expert
system program kept calling them out as having broken rotor
bars.
4: Mechanical Modulation- Beware of motor current testing, if
there is a speed reduction gearbox coupled to the motor. Low
speed mechanical modulation will sometimes cause the current
to modulate thereby mimicking a rotor problem. Always
determine the motor speed to within 1 RPM, then calculate the
number of poles times slip frequency side band frequency. If
there is any variation in the calculated versus the actual
frequency, then suspect mechanical modulation. Examples:

Case 1: Coal barge unloader. The rate at which the buckets dug
into the barge of coal was exactly the number of motor poles
times the slip frequency making it impossible to perform an
accurate rotor bar analysis.
Case 2- In large coal mills, the
rate at which the rolls pass over
the rotating table is very close to
the number of poles times the
slip frequency. This has resulted
in several mill motors being
falsely called out as having bad
rotor bars.
Case 3- A coal conveyor motor was
called out as having rotor problems. It
was discovered that the speed of the
output gear was close to the number of
poles times the slip frequency. The
problem was with the gear instead of with
the motor rotor. Very accurately
determining the speed of the motor
allowed a calculation to be made that
determined that the modulating
frequency was a match with the gear
instead of the number of poles times the
slip frequency.
5: Two pole and four pole motors with broken
rotor bars will often cause number of poles times
slip frequency side bands in both the current and
vibration spectra. Higher pole lower speed
motors, particularly those driving high inertia
loads will create number of poles times slip side
bands in the current spectra, but will in many
cases not cause them to appear at all in the
vibration spectra.
INDUCTION MOTORS
VIBRATION TESTING
1: Rotor eccentricity- An eccentric rotor will of course result in
unbalance. If the rotor is balanced, there can still be a problem of a
rotating deviation in the air gap. This causes unequal pull on the
rotor as the magnetic poles pass the rotating gap deviation. This
occurs at the number of poles times the slip frequency, which is the
same frequency that is generated by a broken rotor bar. Note that
in neither case will this low (usually less than 1.5 HZ) frequency
show up in the spectrum, but they can both appear as side bands of
running speed in the vibration spectrum. The way to tell the
difference between an eccentric rotor and broken rotor bars is to
obtain a current spectrum. A broken rotor bar will generate no.
poles times slip frequency around the line frequency in the current
spectrum where as the eccentric rotor will not.
 Current
sidebands are
over 60dB
down

ECCENTRIC ROTOR

Distinct No.
Poles X slip
sidebands in
vibration
spectrum
2: New high
efficiency
motors are
much more
susceptible to
soft foot than
older heavy
frame motors.
 3: Large 2 pole motors that have shorted laminations can have very high
levels of thermal vectors that cause the amount of unbalance to vary with
load. Case History- A 4000 Hp motor in a power plant was overhauled.
After the overhaul, the motor vibration would increase and the bearings
would be destroyed. The plant sent the motor back to the motor shop for
balancing, but upon return, it again wiped out the bearings.
The motor was then sent to the manufacturer to be balanced in a high
speed balance pit. Upon return, it again wiped out the bearings. Solution-
Proximity probes were installed on the motor and the amplitude and
phase were monitored as the motor was loaded. The motor had an 8 mil
thermal vector. The motor was compromise balanced and ran for several
years. It was discovered that the original motor shop that overhauled the
motor had dropped the rotor and damaged some of the laminations. The
eddy current heating in the shorted laminations had caused the rotor to
bow thereby causing the large thermal vector. If this condition is
suspected, induction heating the rotor then looking at it with an infrared
camera will allow the hot spots to be seen.
DC MOTORS
1: D.C. MOTORS- The spectrum of the current to
a D.C. Motor can be used to find problems with
SCRs or firing circuits. The rectifier input supply
frequency (50 or 60 HZ) times 6 for 3 phase full
wave rectifiers will normally be present in the
current spectrum. When 1/3 or 2/3 of the firing
frequency is present, it indicates failed SCRs or
firing circuits. It is much simpler to look at the
current spectrum or current waveform than to try to
see the problem with vibration. Vibration is a
secondary effect reflecting the problem which is
actually of electrical origin.
½ HALF WAVE RECTIFIER
 WHAT DOES THE
CURRENT PATTERN OF A
NORMAL DRIVER LOOK
LIKE
6 PULSES IN 1/60th OF A SECOND
 WHAT DOES THE
WAVEFORM OF A BAD
DRIVE LOOK LIKE ?
 BAD
BAD
SCR’S

120 HZ

GOOD
GOO
SCR’S
D
F E B R U A R Y 18, 1998
V IS Y F A N P U M P D R IV E
250 A M P LO A D

W H E N S C R ’S W E R E R E P L A C E D , T H E W A V E FO R M A N D SP E C T R U M R E T U R N E D T O
NORM AL.

1 2 0 H Z is
NOTE PRESENT
V A R IA T IO N IN A L O N G W IT H
H E IG H T O F SO M E 60H Z
PEAK S. COM PONENT

360 H Z SC R
F IR IN G
FREQUENCY
 2: D.C. MOTORS- The current spectrum from a D.C.
Motor can also be used to find tuning problems with
D.C. Drives. Improperly tuned drives will produce
frequencies at the oscillation rate of the instability.
These frequencies can also appear in the vibration
spectra and are very difficult to analyze since they do
not have a mechanical origin. These oscillation
frequencies are unpredictable. They are a result of the
interaction between the rotating inertias of the
mechanical components, the torsional stiffness
of the shafts and the tuning of the electrical control
system. If a completely unexplainable frequency
appears on a drive, then it may well be due to this
complex interaction.
 BAD TUNING, WHERE
SPEED IS CONSTANTLY
MOVING UP AND DOWN.
BADLY ADJUSTED NORMAL
DRIVE SPECTRUM

ONLY 360 HZ

OVERHEATING 180
DEG.
F
DRIVE INSTABILITY

AFTER TUNING
 SOMETIMES THE
PROCESS CAN CAUSE THE
DRIVE TO APPEAR
UNSTABLE
3: DC MOTORS- Unknown frequencies
in the spectrum of the current going to a
DC drive can originate from other
mechanical equipment in the drive train.
Case History- The current on a couch
roll of a paper machine had an unknown
component in its spectrum. It turned out
to be the vane pass frequency of the fan
pump located several yards away in the
basement. The fan pump was causing
pressure pulsations in the head box
that caused the paper to be deposited in
varying thicknesses. As the thicker
material passed over the vacuum rolls,
this caused the tension to increase
which changed the tangential force on
the couch roll which in turn caused the
current draw to the couch roll to
modulate at that rate.
 COUCH ROLL DRIVE

12.5 HZ VIBRATION DID NOT

MATCH ANYTHING
12.58 HZ SIGNAL MATCHES VANEPASS OF FAN PUMP
 5: DRIVE ROLL VIBRATES AT VANE PASS
DUE TO CHANGES IN TENSION AND MOTOR
CURRENT VARIES AT SAME RATE.
HEAD BOX

2: MATERIAL IS UNEVEN FROM PULSES

4: TENSION VARIES IN FELT

3: VACUUM ROLL PULLS DOWN ON

FELT.

1:FAN PUMP GENERATES PRESSURE PULSATIONS

AT RPM TIMES NO. OF VANES.
VARIABLE FREQUENCY DRIVES
CAN CAUSE PREMATURE
BEARING FAILURES
1: Line length is a factor in voltage spikes. Rapid switching of inverters causes
voltage spikes that can be amplified by longer line lengths
2: Solutions to minimize bearing failures that result from VFD problems.
A: Lower the firing frequency of the inverter The switching speed is a critical
factor in regards to VFD drive problems. “When VFD drives were first
introduced in the eighties, there were few field problems. The carrier
frequencies were generally below 2.5 kHz. As the switching frequencies
increased, the number of problems also went up
B: Keep the line length between the inverter and the motor as short as
possible.
C: Insulate bearings- Both bearings need to be insulated. In addition, the
coupling must also be insulated or the current can travel through the coupling
to the driven unit’s bearings and then to ground.
D: Shaft Grounding- Grounding the shaft with carbon brushes
allows the potential to travel to ground. The problem with this
approach is that brushes need to be maintained. If the
brushes wear out, then the current will again start flowing
through the bearings.
E: Conductive grease- Conductive grease allows the current
to drain off rather than building up to a destructive potential.
The downside to conductive grease is that it has been
reported that bearing life is not as long as with standard
grease.
F: Ceramic Bearings- Since ceramic bearings are
nonconductive, they are another method of achieving
electrical isolation between the rotor and the frame. Do not
forget to insulate the coupling.
G: Output filters- These devices filter out the unwanted high
order harmonics.
H: Isolation Transformers- “An isolation transformer with a
delta primary and a wye secondary will greatly reduce
common mode voltages within a drive and motor system.
GEARBOXES
1: Gear boxes which have common prime factors
between the teeth of intermeshing gears can
produce sub harmonics of the tooth mesh
frequency.

Case History- A large drag line gear box had 1

in/sec of exact ½ tooth mesh vibration. It was
discovered that there was a common prime factor
of two on the pinion and bull gear. The pinion
was worn badly . When the pinion was replaced,
the ½ tooth mesh vibration disappeared.
2: Side bands are often the result of
modulation by a defective component
in a gear box. However, beware of
making hasty conclusions when
analyzing a planetary gear box, where
modulation naturally occurs due to a
continually varying transmission path
caused by rotation of the planetary
gears. Jack Frarey
3: Non-linear
modulation resulting
from looseness can
generate families of
side bands. The
worse the looseness,
the greater the
number of side
bands.
 TORSIONAL VIBRATION
1: When making torsional measurements, pulse trains need to be placed on
anti-nodes and strain gauges installed on nodes.
2: As a synchronous motor comes up to speed, the torsional stimulation will
start out at 120 HZ and then drop off in frequency as the motor comes up to
speed.
3: It a torsional natural frequency needs to be altered, then the most likely
place to make a modification whether it be stiffness or damping is in the
coupling.
FANS
1: Always take an axial reading on the bearing that absorbs the thrust.
2: If a fan is mounted
on isolation springs,
then lock up the
isolators prior to
balancing.
3: On large sleeve bearing fans, excessive motion of the shaft can be
the result of the plunger hold down bolts being loose. This condition
can be picked up by either a shaft stick reading or a proximity probe.
The casing level may be moderate, but the shaft motion can be
severe. For instance, there may be three mils of vibration on a large
fan’s bearing housing and maybe 15 mils on the shaft. If the bearing
clearance is only .008” then it is a good bet that the bearing is moving
within the housing. Tightening up the plunger bolt will raise the casing
motion and reduce the absolute shaft motion. Once this is done, it
will be possible to balance the fan. In the case where the bearing can
move within the housing, the system is highly nonlinear and balancing
is almost impossible.
4: Large airfoil blade fans
have hollow blades. These
blades can fill with dust or
even worse with water
when they are pressure
washed.
OIL WHIRL
 Oil Whirl
1: If oil whirl occurs, check bearing
clearances, oil temperature and
alignment. Any of these can cause
a marginal system to whirl. If these
simple field fixes do not work, then call a bearing
expert.
There are too many things involved in bearing design
to try to do it yourself.
2: Oil whirl will remain oil whirl until the
machine reaches a speed of twice the
rotor’s natural frequency at that point it
can turn into oil whip. Even if the speed is
increased, the whip frequency will remain
locked in at the shaft critical speed. ( note
the critical speed will change slightly as
the speed increases due to oil film
stiffness and gyroscopics that change
with speed).
LOW STIFFNESS PROBLEMS
 TROUBLE BREWING
EXPANSION JOINTS PRESSURE PULSATIONS
1: Beware of expansion joints when dynamic pulses are
present. Also beware of long bolts holding expansion bolts
together. Expansion joints have very low stiffness, so small
pressure pulsations can result in large axial movements. Long
bolts can have low stiffness values. K=EA/L If L is big, then
K is small. The combination of a large expansion joint
restrained by long bolts can result in very high levels of
vibration if pressure pulsations are present. Case History - A
pipe in a refinery had over 2.3 inches per second of vibration at
4500 CPM on its end cap which was mounted past an
expansion joint. Pressure pulsation from throttling by a valve
caused high amounts of motion due to the large area (1075
square inches) on which the pulses acted combined with 20 ft.
long restraining bolts that had low K values.
WHY AM I IN THIS
BUSINESS?
 SOLUTION
• A 37” pipe has an area of 1075
square inches.
• Stiffness of bolts is EA/L
• Partially closed butterfly valve was
generating broad band noise that
caused pipe to resonate.
• Missing baffle allowed pressure
pulsations to move end cap.
MISSING COMPONENT
BAFFLE HAD BEEN
REMOVED FROM
THIS LOCATION
DURING OUTAGE
 IMPORTANT POINTS

• 1: Large areas can generate large

forces
• 2: Expansion joints allow those large
forces to generate high levels of
vibration
• 3: Long bolts are not very stiff
• 4: Do not remove components you
do not understand. Just because it is
full of holes does not mean that it is
not needed.
2: Beware of large thin surfaces that are subject to
pressure pulsations. Case History- A 6’ by 8’
window was vibrating excessively in a large
building. The thinness of the window meant that its
stiffness was very low. This in combination with its
large area combined to produce large amounts of
movement as a result of pressure pulsations from a
loose heat exchanger coil that was vibrating in a
duct feeding the atrium of the building.
 PUFFS OF AIR
AT AN OFFICE BUILDING WITH AN ATRIUM
WINDOWS IN THE LOWER OFFICES WERE VIBRATING

7 MILS OF VIBRATION
AT 240
CYCLES/MINUTE
 BUILDING LAYOUT
FAN ROOM FAN ROOM

OPEN ATRIUM
VIBRATING
WINDOW
 LOOSE COIL
FAN DISCHARGE TO ATRIUM

VIBRATING COIL ACTED LIKE

6X6 FOOT SPEAKER ELEMENT
FIVING OFF LOW FREQUENCY
PRESSURE PULSES

COIL WAS SUPPORTED BY ONLY SUPPLY AND

RETURN PIPE. NATURAL FREQUENCY 240 CPM.
3: Even worse- Beware of large thin surfaces that have a natural
frequency that equals that of a pressure pulsation. Case history-
A large circular window had a natural frequency of 33 Hz which
was the firing frequency of newly installed high efficiency boilers.
The exhaust stack from the boilers was located only a few feet
from the window. The window had extremely high levels of
movement which in turn generated a very high level of 33Hz
sound in the building which annoyed the people who were
occupants of this large sub-woofer of a building. To make
matters even worse, the stair well leading up to the window was
exactly one wave length of the 33 Hz signal. Solution- Exhaust
stacks were lengthened to a point well above window.
 SOUND PROBLEM FROM
FURNACE
At a university, the new alumni office was experiencing high
levels of both sound and vibration. The problem was traced to
new pulse furnaces that had been installed.

Stacks from pulse furnaces

VIBRATION AND SOUND LEVELS IN ROOMS

NEXT TO WATCHTOWER WERE HIGH.
LAYOUT OF BUILDING
 BAD LUCK

1.0
TEST - I.U. FOUNDATION BUILDING
I.U. BLDG -P01 POINT 1
Route Spectrum
FIRING
11-JAN-00 09:29:40

FREQUENCY OF

37.94
OVRALL= .9600 V-DG
0.8 PK = .9550
LOAD = 100.0

FURNACE IS 33
RPM = 3600.
PK Velocity in In/Sec RPS = 60.00

0.6

0.4
HZ
0.2

0
Freq: 37.81
0 60 120 180 240 300 Ordr: .630
Frequency
TEST in Hz
- I.U. FOUNDATION BUILDING Spec: .751
I.U. BLDG -P01 POINT 1
0.40

0.35
Analyze Spectrum
11-JAN-00 14:00:30

PK = .4147
Response of 3 foot
diameter window in
LOAD = 100.0
0.30 RPM = 3600.
RPS = 60.00
PK Velocity in In/Sec

0.25

0.20
watch tower
0.15
to impact 33 hz.
0.10

0.05

0
Freq: 32.50
0 100 200 300 400 500 600 Ordr: .542
Frequency in Hz Spec: .273
More Bad Luck
Natural frequency of large window
equals pulse frequency creating a
gigantic subwoofer 3 feet in
diameter.

Height of tower closely

matched wavelength
of problem frequency
SOLUTION
Raised height of stacks to well
above window

INSTALLED
BETTER
MUFFLERS
TEST TECHNIQUES
1: Don’t forget to use a strobe light- In the past Strobe lights were
used frequently due to fact that they were utilized to obtain the phase
readings. Their use will almost certainly remain relevant as long as
there is rotating equipment. Strobe lights are still the best way to
looks at belts or to see if the elements in a coupling have cracked.
They can also be used to determine the size of a key or locate a
keway or look at balance weights. Strobes that are triggered by the
vibration signal are useful to freeze the rotating element that is
causing the vibration. Case History- A paper mill was going to
remove a roll to get it balanced. The frequency of the vibration
matched the frequency that the group of wire rolls was
vibrating at. This was determined by a calculation of the RPM
of the roll base upon the roll diameter and the product’s speed.
A strobe fired by the vibration was used to verify that the roll
was the source. To everyone’s surprise, the strobe froze a
different roll. It turned out that the drawings had incorrect
diameters. The strobe didn’t lie or care about drawings, it just froze
the correct component.
2. Phase locked loop strobes with a phase delay
are very useful for providing a once per
revolution output that can be used for balancing.
The use of a phase lock strobe can make it
unnecessary to stop a machine so that photo-
reflective tape can be installed. This can save
hours of time on a balance job, particularly when
a machine is limited to the number of starts that
is can go through.
3: Don’t forget shaft sticks- While it
would be great to have proximity probes
on every machine to measure shaft
motion, in the real world this is just not
the case. A shaft stick can measure the
absolute motion of a shaft. Note, before
using a shaft stick, use the strobe
mentioned in point 1 to make sure there
is not a keyway where the shaft stick is
to be placed against the shaft.
 4: It is handy to have an analog integration box for certain tests that
allow the time waveform from an accelerometer to be viewed in
displacement. Case 1- Two large vertical pumps had resonant
frequencies near their operating speed. One of the calculations that
were needed was to determine the damping so the amplification
factor could be obtained. By using an analog integrator, the time
waveform of the low frequency response could be directly measured
and used for the log decrement calculation. Case 2- Foundries have
low frequency vibratory conveyors that move the castings throughout
the plant. These often cause vibration problems beyond the plant
boundaries that result in complaints by neighbors. These conveyors
which operate around 5 HZ move in and out of phase with one
another causing the vibration levels to vary significantly with time. If a
spectrum is taken of one snapshot in time, the overall value is hard to
obtain. It is much easier to look at the motion in the time domain over
a period of several seconds. The maximum peak to peak motions
that people offsite are feeling can them be easily determined.
If analog integration is available, then a long term time plot of the
motion in displacement is very useful in determining the maximum
levels of motion that are being experienced.
Vibratory conveyors that were shaking houses ½
mile from foundry. Calculated Peak-Peak 4.1 mils.
Actual P-P over 7 mils. People feel much higher
levels than spectrum indicates.
DALT - DINING ROOM TABLE AT RHODES
P-P Displacement in Mils

DINING TBL-E-W EAST WEST

4.55
3.5 ROUTE SPECTRUM

4.40
23-JUN-99 13:11:40
3.0 OVRALL= 4.15 D-DG
2.5 P-P = 4.13
2.0 LOAD = 100.0

4.71

4.94
RPM = 360.

5.07
1.5
RPS = 6.00
1.0
0.5
0
2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0 7.5
Frequency in Hz
Displacement in Mils

6
ROUTE WAVEFORM
4 23-JUN-99 13:11:40
P-P = 4.31
2
PK(+) = 3.93
0 PK(-) = 3.96
CRESTF= 2.60
-2

-4

-6
Time: 8.848
0 2 4 6 8 10 12 14 16 18 20 Ampl: 3.931
Time in Seconds
5: A microphone with an analog output that can be used to
supply a signal to a spectrum analyzer can be very useful
in the analysis of vibrations that are transmitted by the air
rather than through solid material. Note that most
microphones have a pretty severe roll of below 20Hz, so
the true pressure pulsation amplitude may not be present
at a lower frequency. The spectral data can still however
be used to identify the problem.
6: If a temporary shaft rider is needed, then Ebelon rod
works well. Ebelon is graphite impregnated Teflon and it
will last a significant amount of time in contact with a
reasonably smooth shaft.
 Some Final Thoughts
A vibration analyst must understand the
basic laws of physics (F=ma and F=kx and
that dynamic stiffness is different than
static stiffness). They must also
understand signal processing so they do
not get bad data. They must have an
appreciation for human nature so they can
get the truth out of mechanics and
operations personnel. They need to
understand how fans motors, gear boxes,
compressors, pumps and turbines work.
But most of all they must be able to put all
these things together under adverse
conditions and then be able to think clearly
and arrive at a logical conclusion.