Escolar Documentos
Profissional Documentos
Cultura Documentos
1
Vibration Analysis on Rotating Equipment
CONTENTS
Vibration Examples------------------------------------------------------------------------------------------------- 9
Imbalance--------------------------------------------------------------------------------------------------------------- 9
Misalignment ---------------------------------------------------------------------------------------------------------10
Looseness--------------------------------------------------------------------------------------------------------------11
Rolling Element Bearing Defects----------------------------------------------------------------------------------13
VIBRATION THEORY--------------------------------------------------------------------------- 15
Simple Harmonic Motion-------------------------------------------------------------------------------------------15
RMS vs. PEAK-------------------------------------------------------------------------------------------------------18
Time Domain ---------------------------------------------------------------------------------------------------------19
The Frequency Domain ---------------------------------------------------------------------------------------------19
What is an FFT?------------------------------------------------------------------------------------------------------20
The FFT Analyzer ---------------------------------------------------------------------------------------------------20
Advantages of FFT Analyzers -------------------------------------------------------------------------------------21
Frequency Spans -----------------------------------------------------------------------------------------------------21
Measurement Basics -------------------------------------------------------------------------------------------------21
Spectrum---------------------------------------------------------------------------------------------------------------21
Parameter Selection------------------------------------------------------------------------------------------------ 22
Selecting displacement, velocity or acceleration----------------------------------------------------------------22
FAILURE MODES-------------------------------------------------------------------------------- 32
Induction Motors--------------------------------------------------------------------------------------------------- 32
Mechanical or Electrical Effects-----------------------------------------------------------------------------------32
Armature Related Problems ----------------------------------------------------------------------------------------32
Stator Related Problems --------------------------------------------------------------------------------------------33
Broken Rotor Bars ---------------------------------------------------------------------------------------------------34
DC Motors----------------------------------------------------------------------------------------------------------- 34
How DC Power Is Created. ------------------------------------------------------------------------------------34
DC Systems and Controls-------------------------------------------------------------------------------------------36
DC Control Firing Cards--------------------------------------------------------------------------------------------37
S.C.R. problems ------------------------------------------------------------------------------------------------------38
Example of a Firing Card Fault ------------------------------------------------------------------------------------38
2
Vibration Analysis on Rotating Equipment
Balancing ------------------------------------------------------------------------------------------------------------ 75
In-place Balancing ---------------------------------------------------------------------------------------------------75
Vibration Related to Imbalance------------------------------------------------------------------------------------78
How to Balance - Single Plane-------------------------------------------------------------------------------------79
Single Plane Vector Method of Balancing-----------------------------------------------------------------------80
Four-step Method of Balancing Single Plane -------------------------------------------------------------------82
Balancing in One Run -----------------------------------------------------------------------------------------------85
Introduction --------------------------------------------------------------------------------------------------------- 91
3
Vibration Analysis on Rotating Equipment
Conclusion ----------------------------------------------------------------------------------------------------------- 96
Measurement Windows ---------------------------------------------------------------------------------------------99
Averaging ----------------------------------------------------------------------------------------------------------- 100
Real Time Bandwidth and Overlap Processing --------------------------------------------------------------- 101
Octave Analysis ---------------------------------------------------------------------------------------------------- 102
Phase---------------------------------------------------------------------------------------------------------------- 127
4
Vibration Analysis on Rotating Equipment
Plates---------------------------------------------------------------------------------------------------------------- 137
5
Vibration Analysis on Rotating Equipment
List of illustrations
Figure 1 Velocity spectrum showing imbalance ______________________________________________ 9
Figure 2 - Velocity spectrum showing fan imbalance _________________________________________ 10
Figure 3 Velocity spectrum of misaligned fan - radial ________________________________________ 11
Figure 4 Velocity spectrum of misaligned fan - axial _________________________________________ 11
Figure 5 Velocity spectrum from a loose fan drive motor______________________________________ 12
Figure 6 Envelope spectrum of a fan drive motor with loose bearing ____________________________ 13
Figure 7 Enveloped acceleration spectrum of bearing - inner race defect _________________________ 14
Figure 8 Inner race spall_______________________________________________________________ 14
Figure 9 Simple Harmonic Vibration _____________________________________________________ 16
Figure 10 Integration from acceleration to velocity __________________________________________ 17
Figure 11 Integrating to displacement ____________________________________________________ 18
Figure 12 Peak -v- RMS _______________________________________________________________ 19
Figure 13 Compression mode accelerometer _______________________________________________ 24
Figure 14 Shear mode accelerometer _____________________________________________________ 24
Figure 15 Simple modulation example ____________________________________________________ 25
Figure 16 Bearing modulation example ___________________________________________________ 26
Figure 17 Demodulation process ________________________________________________________ 27
Figure 18 Enveloping process___________________________________________________________ 27
Figure 19 Fast Fourier Transform _______________________________________________________ 28
Figure 20 FFT - 3D view_______________________________________________________________ 28
Figure 21 Two channel time waveform - bearing defect _______________________________________ 29
Figure 22 High frequency waterfall ______________________________________________________ 30
Figure 23 Enveloped acceleration spectrum________________________________________________ 30
Figure 24 Comparison - velocity to envelope _______________________________________________ 31
Figure 25 The creation of DC power _____________________________________________________ 35
Figure 26 FFT spectrum of half wave rectification___________________________________________ 36
Figure 27 FFT spectrum of full wave rectification ___________________________________________ 36
Figure 28 Basic DC system circuit _______________________________________________________ 37
Figure 29 FFT spectrum full wave DC firing card frequencies _________________________________ 38
Figure 30 FFT spectrum after repair _____________________________________________________ 38
Figure 31 FFT spectrum showing half wave firing card frequencies _____________________________ 39
Figure 32 FFT spectrum of same motor (no load) ___________________________________________ 40
Figure 33 FFT spectrum showing comparitor card defect._____________________________________ 41
Figure 34 FFT after the comparitor card was replaced _______________________________________ 42
Figure 35 DC motor components ________________________________________________________ 42
Figure 36 FFT from a 5 HP motor - full wave rectified _______________________________________ 43
Figure 37 Same motor - speed lowered by 25% _____________________________________________ 44
Figure 38 Imbalance slide 1 ____________________________________________________________ 46
Figure 39 Imbalance slide 2 ____________________________________________________________ 46
Figure 40 Imbalance slide 3 ____________________________________________________________ 47
Figure 41 Imbalance slide 4 ____________________________________________________________ 47
Figure 42 Imbalance slide 5 ____________________________________________________________ 48
Figure 43 Imbalance slide 6 ____________________________________________________________ 48
Figure 44 Imbalance slide 7 ____________________________________________________________ 49
Figure 45 Imbalance slide 8 ____________________________________________________________ 49
Figure 46 Imbalance slide 9 ____________________________________________________________ 50
Figure 47 Misalignment slide 1__________________________________________________________ 51
Figure 48 Misalignment slide 2__________________________________________________________ 51
Figure 49 Misalignment slide 3__________________________________________________________ 52
Figure 50 Misalignment slide 4__________________________________________________________ 52
Figure 51 Misalignment slide 5__________________________________________________________ 53
Figure 52 Looseness slide 1 ____________________________________________________________ 54
Figure 53 Looseness slide 2 ____________________________________________________________ 54
6
Vibration Analysis on Rotating Equipment
7
Vibration Analysis on Rotating Equipment
List of Tables
Table 1 Speed of sound in liquids _______________________________________________________ 118
Table 3 Natural frequency calculation of uniform beams_____________________________________ 135
Table 4 Standard values for uniform beams _______________________________________________ 136
Table 5 Damping ranges of vibration isolators_____________________________________________ 139
Table 6 Whole body vibration (frequency ranges) __________________________________________ 148
List of Equations
Equation 3 Newton's 2nd law __________________________________________________________ 117
Equation 5 Differential equation of motion of a single-degree-of-freedom system _________________ 118
Equation 6 Velocity of sound in materials_________________________________________________ 119
Equation 7 Speed of sound in the ocean __________________________________________________ 119
Equation 8 General formula relating speed, wavelength & frequency ___________________________ 119
Equation 9 Newton-Laplace eq. for the speed of sound in a gas _______________________________ 121
Equation 10 Ratio of specific heats (gamma) ______________________________________________ 121
Equation 11 Speed of sound in a gas ____________________________________________________ 121
Equation 12 Karman vortices __________________________________________________________ 123
Equation 14 Amplitude magnification due to springs ________________________________________ 141
Equation 15 Natural frequency of a spring ________________________________________________ 142
Equation 16 Modulus of elasticity for rubber ______________________________________________ 144
8
Vibration Analysis on Rotating Equipment
Vibration Examples
When the novice analyst first carries out vibration analysis he will usually rush out and take a vibration
spectrum using the default parameters set up in the analyzer. We will carry on that noble tradition and look
at some spectra that have been collected from real machines and show typical examples of common defects.
Imbalance
Figure 1 shows a vibration spectrum that was taken at the sheave end of a centrifugal fan in the vertical
direction. The fan was driven from the AC motor via a V-belt and rotated at about 720 rpm. The AC drive
motor rotates at just under 1200 rpm.
The spectrum is a simply a graph of the vibration frequency on the bottom axis with the amplitude at that
frequency on the vertical axis. This spectrum is of velocity vibration so the amplitude units could be in
9
Vibration Analysis on Rotating Equipment
mm/s or ips (inches/second). The frequency is in cpm (cycles/minute) but it could have been displayed in
Hz (Hertz or cycles/second) or in orders (multiples of run speed). The spectrum was recorded from a
vibration transducer which was mounted radial to the shaft (vertical in this case).
Notice in the spectrum that there is one big spike which is labeled at 716.59 cpm and there are two much
smaller spikes just to the right. The first spike to the right is at 1187 cpm which equates to the run speed of
the motor and the second spike is at 1433.18 cpm which is exactly twice fan speed. Because the one spike
is so dominant that is the one that we are concerned about. A check with a stroboscope confirmed that the
fan was actually running at 717 rpm so the big spike of vibration is at exactly (within the precision of the
strobe) run speed. At this stage we are not concerned about the physics of why a vibration at run speed is
usually indicative of imbalance but we will look at our spectral explanation charts (see appendix 1) and
have a fair degree of confidence that the fan needs balancing.
Figure 2 shows a similar problem on a different fan but we see that the spectrum looks very similar with one
dominant spike at the run speed of the fan. Figure 1 amplitude was displayed with metric units and figure 2
with inch units but the shape of the spectrum is the same in both cases.
Misalignment
Probably 40% of all bearing and shaft failures are caused by misalignment of the components creating an
extra axial thrust on the bearings.
10
Vibration Analysis on Rotating Equipment
In figure 3 we see what initially looks like an imbalance condition of the fan, although the amplitudes are
relatively low.
However, in figure 4 we are now looking at the vibration taken axial to the shaft. If the problem was simple
imbalance of the fan we would expect all of the forces to be caused by centrifugal force and therefore acting
in a direction which was radial to the shaft. Again, looking at our spectrum explanation charts we see that,
on a belt driven train, a high axial velocity vibration relative to the radial vibration is almost always
indicative of component misalignment to the belt.
Looseness
Looseness exists when the component is not directly attached to the structure or rotating element and has a
relatively large clearance, allowing the component to rattle.
11
Vibration Analysis on Rotating Equipment
The above spectrum was recorded at the sheave end of the drive motor of an underground colliery main
ventilation fan. The motor was running at 590 rpm and immediately we see the large family of harmonics
of run speed. The amplitudes do not seem too high but the machine was massive and any vibratory forces
have to move the mass before we see a vibration. In this case the structure of the bedplate was cracked
causing parts of the structure to vibrate freely at the excitation frequency of the motor (speed).
Whenever we see multiples or sub-multiples of run speed vibration frequencies we immediately consider
the possibility of loose components.
12
Vibration Analysis on Rotating Equipment
The early stages of looseness can be detected in a similar manner, be looking for harmonics of run speed,
and using demodulated or enveloped acceleration readings. Figure 6 shows the early stages of looseness of
a bearing inside the fan drive motor. As the looseness deteriorates the envelope readings will decrease but
then the velocity readings will start to increase.
13
Vibration Analysis on Rotating Equipment
14
Vibration Analysis on Rotating Equipment
Vibration Theory
The following section is meant as a primer to help the newcomer to vibration analysis understand some of
the terms used and to develop an understanding of the concepts.
To understand the concept of vibration analysis, it is important to realize that the motion of the measured
surface varies with time. The transducer converts the movement into an electrical signal which is passed to
the spectrum analyzer which in turn converts that signal from the time domain into the frequency domain.
The time domain waveform is composed of a machine’s response to many individual forces such as
imbalance, misalignment, gear meshing forces, rotating electrical fields, and many other factors. When
viewing the time domain data it can be quite difficult to separate these components of vibration. However,
in the frequency domain it is much easier to separate these elements to determine the importance of each.
Vibration amplitude is measured using three different parameters, acceleration, velocity and displacement.
The purpose of this section is to describe the relationship between each of these and how they are used on
rotating machinery.
15
Vibration Analysis on Rotating Equipment
(48)
* FREQUENCY
* AMPLITUDE
... Displacement
... Velocity
... Acceleration
one cycle
C B
Figure 10 shows that at position B and C, the velocity is zero, and at position A the velocity is maximized,
first to the right, then to the left. Since the peak positive velocity occurs 1/4 cycle before the peak positive
displacement, velocity is said to lead displacement by 90°. The 90 degree phase lead is shown in the
diagram on figure 10. Velocity amplitude is expressed only in terms of zero to peak or zero to RMS.
The negative peak velocity differs only in direction, not magnitude. The rate of change of displacement is
the velocity, therefore if D is expressed in terms of inches, instead of the usual mils, then the product 2πfD
will be the velocity in inches per second which are the units used for velocity in vibration work.
16
Vibration Analysis on Rotating Equipment
(49)
A B A C A
Disp
Peak
Peak
to
Peak
C B
Disp
Vel
C B
Highest
Velocity
As velocity is continuously changing, an acceleration is also associated with the velocity, this acceleration is
also associated with the motion. Acceleration is the third way to express vibration amplitude. Figure 11
shows that at position B and C the acceleration is maximum. Just prior to point B, velocity is to the right
and just after it is to the left. At B therefore the rate of change of velocity, the acceleration, is maximum.
Conversely just prior to point A velocity is increasing and just after, it is decreasing. Therefore the rate of
change of velocity (the acceleration) must be zero at A. Note that acceleration reaches its maximum at
Points B and C just as displacement does, but at B acceleration is to the left whereas the displacement is to
the right. The maximum acceleration to the right occurs 1/2 a cycle before the maximum velocity to the
right and acceleration is said to lead displacement by 180°. Acceleration leads velocity by 90°.
The diagram in figure 11 shows these phase leads and also the acceleration amplitude relationship, A =
(2πf)2D. This says that for any given value of displacement, the acceleration is proportional to the square of
the frequency.
The unit of acceleration is the “g” which is equal to 9.81 m/sec2 and is derived from the acceleration due to
earth’s gravity.
17
Vibration Analysis on Rotating Equipment
(50)
A B A C A
Disp
Vel
C B
Accel
A
Acceleration Acceleration
Peak
Avg RMS
Peak
Amplitude
to
Peak
Time
18
Vibration Analysis on Rotating Equipment
(51)
Peak
Avg RMS
Peak
Amplitude
to
Peak
Time
Peak
Peak
to
Amplitude
RMS Peak
Time
Complex Waveform
Time Domain
The traditional way of observing signals is to view them in what is called the time domain. The time
domain is a record of what happened to a parameter compared to time- Typically the signal would be
displayed on an oscilloscope. With respect to machinery vibration, analysis of signals in the time domain
can be very difficult and is far easier in the frequency domain
19
Vibration Analysis on Rotating Equipment
What is an FFT?
The fast Fourier transform (FFT) is an algorithm for transforming data in the time domain to the frequency
domain. Most analyzers have an FFT processor, which performs this transformation automatically and then
stores the computed spectra into memory.
We cannot transform to the frequency domain in a continuous manner. We therefore must sample and
digitize the time domain input. The number of samples determines the resolution (number of lines) of
frequency.
Most analyzers offer resolutions of 100,200,400,800,1600,3200 or even 6400 Lines.
FFT Spectrum Analyzers take a time varying input signal, like you would see on an oscilloscope trace, and
compute its frequency spectrum.
Fourier's theorem states that any waveform in the time domain can be represented by the weighted sum of
sines and cosines. The FFT spectrum analyzer samples the input signal, computes the magnitude of its sine
and cosine components, and displays the spectrum of these measured frequency components.
Many of these measurements were once done using analog spectrum analyzers. In simple terms, an analog
filter was used to isolate frequencies of interest. The signal power, which passed through the filter, was
measured to determine the signal strength in certain frequency bands. By tuning the filters and repeating the
measurements, a spectrum could be obtained.
20
Vibration Analysis on Rotating Equipment
Everything below 2.5 Hz (for this example) is considered to be DC. The output spectrum thus represents
the frequency range from DC to 1000 Hz with points every 2.5 Hz.
Frequency Spans
Before we continue, let's clarify a couple of points about our frequency span. We just described how we
arrived at a DC to 1000 Hz frequency span using a 400 mS time record. Because the signal passes through
an anti-aliasing filter at the input, the entire frequency span is not useable. A typical filter has a flat
response from DC to 1000 Hz and then rolls off steeply from 1000 Hz to 2.56 kHz. The range between
1000 Hz and 2.56 kHz is therefore not useable and the actual displayed frequency span stops at 1000 Hz.
There is also a frequency bin labeled 0 Hz (or DC). This bin actually covers the range from 0 Hz to 2.5 Hz
(the lowest measurable frequency) and contains the signal components whose period is longer than the time
record (not only DC). So our final displayed spectrum contains 400 frequency bins. The first covers 0 - 2.5
Hz, the second 2.5 - 5 Hz, and the 400th covers 997.5 - 1000 Hz.
The length of the time record determines the frequency span and resolution of our spectrum. What happens
if we make the time record 800 mS or twice as long? Well, we ought to get 2048 time points (sampling at
2560 Hz) yielding a spectrum from DC to 1000 Hz with 1.25 Hz resolution containing 800 points. But the
analyzer places some limitations on this. One is memory. If we keep increasing the time record, then we
would need to store more and more points. (0.00125 Hz resolution would require 2,048,000 values.)
Another limitation is processing time. The more points you take, the longer the processing time.
Measurement Basics
An FFT spectrum is a complex quantity, This is because each frequency component has a phase relative to
the start of the time record. (Alternately, you may wish to think of the input signal being composed of sines
and cosines.) If there is no triggering, then the phase is random and we generally look at the magnitude of
the spectrum. If we use a synchronous trigger then each frequency component has a well-defined phase.
Spectrum
The spectrum is the basic measurement of an FFT analyzer. It is simply the complex FFT. Normally, the
magnitude of the spectrum is displayed. The magnitude is the square root of the FFT times its complex
conjugate. (Square root of the sum of the real (sine) part squared and the imaginary (cosine) part squared).
The magnitude is a real quantity and represents the total signal amplitude in each frequency bin,
independent of phase.
If there is phase information in the spectrum, i.e. the time record is triggered in phase with some component
of the signal, then the real (cosine) or imaginary (sine) part or the phase may be displayed. The phase is
simply the arc tangent of the ratio of the imaginary and real parts of each frequency component. For
vibration measurements phase is usually considered to be relative to the trigger pulse.
21
Vibration Analysis on Rotating Equipment
Parameter Selection
F=m⋅⋅x a
22
Vibration Analysis on Rotating Equipment
F=m⋅⋅ϖ2⋅r
where ϖ = the rotational speed in radians/second
r = the radius at which the force is acting.
As we know that the acceleration is proportional to the force and we assume that the mass and radius of
force of the machine stay constant, then we may safely say that the acceleration is also proportional to the
square of the speed.
ϖ 2⋅ r
a=ϖ
The important point here is that the faster the machine goes, or the higher up the frequency range we go, the
acceleration amplitudes must increase for a given force even if there is nothing wrong with our machine.
However, we know that acceleration is simply the rate of change of velocity. So if we integrate our
acceleration reading with respect to time we will get a velocity reading. Integrating acceleration will
change our value from:
inches/second2
to
inches/second
effectively finding the square root of the acceleration (for time). We have already said that we have a
concern that the acceleration increases with frequency, so if we need a value that is independent of
frequency for severity analysis purposes we can use the velocity reading.
Back to our motor. If we plot the acceleration against time (time domain) we would see a sine wave which
is the result of simple harmonic motion. This is the signal that is passed along to the analyzer. The analyzer
will then convert this time domain signal into a frequency domain signal either as acceleration or as an
integration from acceleration into velocity. Either way, the out of balance condition will show itself in the
frequency domain as a single spike at a frequency which corresponds to the run speed of the machine. For
example, if the motor is running at 1,200 rev/min the spike will have a frequency at 1,200 cycles/minute
(cpm) or 20 Hertz (Hz).
23
Vibration Analysis on Rotating Equipment
Accelerometers
Looking at the figure above we see a schematic of an accelerometer. Modern accelerometers are available
as compression mode or shear mode. Generally speaking the shear mode accelerometer offers better axial
sensitivity with much better mechanical integrity. In other words the shear mode accelerometer is not as
affected by thermal transients and gives better accuracy for the axis in which it is mounted.
Many low cost industrial accelerometers are now shear mode. For off-line measurements the accelerometer
will probably be connected to a magnet and the magnet positioned at a pre-determined point every time a
reading is taken. However, the response from the accelerometer is better if it is permanently mounted.
Permanently mounting an accelerometer should be done with care. The way the accelerometer is mounted
will affect the resonant frequency and, hence, the useable frequency range. By far the best way to mount an
accelerometer is to spot face the subject surface and drill and tap it to accept the stud for the accelerometer.
However, on a motor it is usually not practical to drill into the motor frame for obvious reasons. The best
24
Vibration Analysis on Rotating Equipment
alternative to stud mounting is to have tap blocks made with a tapped hole that will accept the accelerometer
stud.
Theory
But before we look at any case histories using DEMODULATION we should be clear in our mind about
exactly what is MODULATION.
A signal may be said to be amplitude modulated if the amplitude of that signal is changing over a period of
time because of the influence of another signal. The example above was taken from a large steam turbine
running at 3600 rpm. The run speed signal is being MODULATED by a signal at 4 Hz which is probably a
foundation resonance. This type of modulation is commonly found in maintenance applications but
consider the example below.
25
Vibration Analysis on Rotating Equipment
Here we see a vibration at 2 kHz which has been modulated slightly more than three times within the time
period (50 mS which equates to 1 revolution of the inner race). The 2 kHz vibration is the resonance of the
bearing which is being excited by the bearing outer race frequency (3.07 x run speed). The excitation of the
2 kHz frequency by the bearing defect on the outer race causes the 2 kHz amplitude to be changed like the
“roller coaster” example above. In other words the bearing outer race frequency is modulating the bearing
resonance frequency. The demodulation process extracts the modulating frequency to produce a time
waveform which can be handled by the F.F.T. process.
When we DEMODULATE the above reading we are not interested in the 2 kHz frequency but we are
interested in the outer race defect frequency which is:
(1000/50*3.07) Hz = 61.4 Hz.
As can be seen from Figure 2, the modulation is at this frequency. In vibration terms, demodulation is a
way of extracting the rate of occurrence of high frequency resonances.
26
Vibration Analysis on Rotating Equipment
The demodulator circuit now passes the signal through a high pass filter to give the time waveform shown in
the lower section of the plot.
With the time domain signal in this format the F.F.T. conversion would give a single spike in the frequency
domain at the resonant frequency which we have earlier said is not what we want. To modify the signal so
as to be suitable for F.F.T. we must “envelope” (figure above) each parcel of energy by first rectifying and
then passing the signal through a smoothing R-C (resistance-capacitive) circuit.
27
Vibration Analysis on Rotating Equipment
The signal is NOW passed through the F.F.T. and we get a spike in the frequency domain at the bearing
defect frequency (figures above and below).
Resonance Sources
When taking a demodulated reading we must first decide on which filter setting to use that will allow the
carrier signal to pass without allowing the low frequency, high amplitude noise to pass. Conventional
thinking will tell you that the resonance frequency which we are using as the carrier wave is always the
resonant frequency of the bearing; while this is often the case it is not always so. For vibration readings, the
accelerometer which we will use to detect the signal will probably be sitting on top of a magnet which will
give a structural resonance in the 1.5 to 4 kHz range (typically). The bearing housing will have its own
28
Vibration Analysis on Rotating Equipment
resonance, the machine structure will have its own resonance. In short, the carrier wave signal resonance
could be coming from any part of the mechanical structure.
If we are taking a reading with a non-vibration parameter we will probably be utilizing a different carrier
signal so we may have to use a different high pass or band pass filter. Ultra-sound data are heterodyned to
the audible range so demodulating at 5 to 8 kHz gives acceptable results while A.C. electric current should
be demodulated from the A.C. frequency of 60 Hz or 50 Hz.
The 2 upper plots are the time domain signal in two planes over a period of 640 mS. The lower plots show
the time domain (left) and frequency domain (right) over a 50 mS period of the lower 640 mS plot. Note
that the frequency spectrum shows spikes at 2 kHz and 3 kHz while the time domain plots show an “angel
fish” pattern which is classic of a bearing defect. Note also that the lower left portion of the plot is a zoom
of the windowed part of the long time record. This shows a detail of the one “angel fish” and the amplitude
can be seen to be passing from positive to negative and back again many times during the life of a single
angel fish - i.e. a high frequency oscillation. This leads us to the conclusion that this is the frequency of 2
and/or 3 kHz seen in the spectrum and one or both of these frequencies are the result of impacts and
subsequent ring down and they are occurring at the resonant frequency of part of the mechanical structure.
29
Vibration Analysis on Rotating Equipment
The plot above shows a time/frequency cascade of the same time interval cropped below 0.001G. This
clearly shows the modulation of the 2 kHz frequency while the 3 kHz frequency is static. The modulation
has been calculated to be equal to the bearing outer race defect frequency of the motor inboard bearing.
Every time one of the bearing balls passes a defect on the outer race, the ball impacts on the defect causing
the 2 kHz vibration to suddenly rise and then ring down. The 2 kHz is the resonant frequency and the
bearing defect frequency (outer race) is the modulating frequency.
The figure below shows the demodulated spectrum on the left with waterfall plot on the right above a trend
of the defect frequency.
Note that the demodulated spectrum is clean and extremely easy to analyze. The spikes occur at the bearing
defect frequency (outer race) with multiple harmonics but there is no sign of the resonant frequency because
this high frequency has been removed during its use in the demodulation process. The frequency range of
30
Vibration Analysis on Rotating Equipment
the spectrum is such that the frequency of the impacts is clearly visible but we do not need to see the
resonant frequency. The last spectrum in the waterfall is lower than the previous spectrum due to greasing
of the motor bearings which lowered the amplitude at which the impacts caused the bearing to vibrate at
resonance.
This figure shows a similar defect on another machine but here the velocity spectrum (left) is displayed
alongside the demodulated spectrum (right). Note that the demodulated spectrum is much cleaner and
easier to analyze.
31
Vibration Analysis on Rotating Equipment
Failure Modes
Induction Motors
32
Vibration Analysis on Rotating Equipment
• Out of magnetic center This is almost always caused by improper fitting of the
bearings. This causes to motor to run out of magnetic center which gives a vibration at run
speed. There will be a phase difference in the vibration between energized and de-energized
33
Vibration Analysis on Rotating Equipment
DC Motors
34
Vibration Analysis on Rotating Equipment
A half-wave rectified DC drive system will then have a dominant electrically related frequency of 10,800
cpm and a full-wave rectified system will have a dominant electrically related frequency of 21,600 cpm
When these frequencies exist within their respective systems they should be considered normal unless
amplitudes greater than 0.1 in./sec. peak are detected. This usually means that there is an electrical control
problem.
35
Vibration Analysis on Rotating Equipment
36
Vibration Analysis on Rotating Equipment
desired rate of speed change to meet the needs of the driven system. The electrical problems associated
with these systems are:
• AC power supply
• AC to DC rectification components
• DC control components and
• DC motor component failure
The system operates by DC power being supplied to the motor which creates a magnetic field and causes
the motor armature to rotate at a speed proportional to the DC power supplied. As the motor turns, the
tachometer (tach.) also turns which creates a low voltage proportionate to its speed. The tach. low voltage is
compared to a constant or predetermined voltage variance by the comparitor card. The comparitor card,
based on the voltage differential between the constant voltage pot and the tach. low voltage, signals the
firing cards. There is a firing card or order for each SCR which controls the power flow through the SCR.
The SCRs fire or open to create and supply DC power to the motor which either speeds or slows the motor
to the appropriate speed determined by the pot. This is a very basic explanation of a DC motor circuit but
should be sufficient to give an understanding of the concept of how the various components interact.
37
Vibration Analysis on Rotating Equipment
1
S.C.R. problems
1
Thanks to Bill Rinehart for his permission to use this data
38
Vibration Analysis on Rotating Equipment
revealing the FFT spectrum in figure 26. The dominant frequencies are 7,200 cpm, 14,400 cpm and 21,600
cpm. These frequencies are related to DC ELECTRICAL problems; not bearings or mechanical defects
When an FFT appears with dominant 1/3 frequencies of the DC full pulse frequency (21,600 for a full
wave, six SCR system), then suspect firing card or SCR problems in the DC control.
This DC control system uses three firing cards to control six SCRs which is typical of many DC control
systems. When one firing card is not functioning then 1/3 of the power is lost. Dominant frequencies of
7,200 cpm and 14,400 cpm or frequencies equivalent to multiples of 1/3 of 21,600 cpm are representative
of this situation. The firing cards were examined and a loose connection on one card was found and
repaired. The FFT spectrum in figure 27 was taken after the repair. The 7,200 cpm and 14,400cpm
frequencies are now gone. The 21,600 cpm frequency is the normal frequency of a full wave system and
should be present.
Half-wave rectified AC power sources will tend to have 1/3 multiples of 10,800 cpm or frequency
separations of 3,600 cpm. Full-wave rectified AC power sources can also have frequency separation of
3,600 cpm if the system has:
• One firing card for each SCR, and one card is out
• A three card system and one card is partially disabled
• One SCR is not functioning
The FFT spectra directly above and below also show the difference between the firing card frequency
amplitudes of a motor that is uncoupled and running under no-load (below) and the same motor coupled and
running under a load (above). Although the frequencies are the same in each spectrum the amplitudes are
considerably load dependent. The amplitude at 7,200 cpm on the spectrum below is only 0.00847 in/sec
peak while the amplitude of the spectrum above at 7,200 cpm is 0.3037 in/sec peak, yet each spectrum
represents the same firing card malfunction.
Another possible reason for seeing these 1/3 multiples would be if one phase of the AC power source was
not present. This would affect one-third of the system power and virtually render one bank of SCRs
inactive. A simple voltage test of the three incoming AC phases should confirm this situation if present.
39
Vibration Analysis on Rotating Equipment
40
Vibration Analysis on Rotating Equipment
0.0872
5000 16,000
41
Vibration Analysis on Rotating Equipment
The commutator is the device which transfers the DC power to the motor armature. Brushes, usually made
of a carbon alloy, ride against the commutator and supply the DC power to the commutator. It has been
observed that as these brushes wear, readings at one times the motor RPM will rise in amplitude, When the
brushes arc it has been observed that these one times RPM readings will increase dramatically, sometimes
reaching 0.3 ins./sec. peak or even higher in extreme cases. Another frequency associated with the
commutator is the number of slots on the commutator times the motor RPM. Although the significance of
this frequency has not been specifically related to a problem, the brush wear would, once again, be suspect.
42
Vibration Analysis on Rotating Equipment
When determining a frequency relationship to a machine component, be sure that the resolution of the
spectrum is enough to see the difference between closely positioned frequencies. If necessary take multiple
spectra at different resolutions to capture and differentiate all possible frequencies. Remember,
mechanically related problems will follow the speed increases or decreases proportionately and electrical
frequencies will remain constant. This is a powerful analysis tool when determining a mechanical or
electrical frequency.
The following spectra will verify the importance of correct RPM and the ability to determine mechanical
frequencies by noticing how the frequencies change related to the RPM.
43
Vibration Analysis on Rotating Equipment
44
Vibration Analysis on Rotating Equipment
Rotating Equipment
Imbalance
Let us consider a fan impeller of 50 kg weight which is running at 1000 rpm. Let us imagine that this
impeller has an out of balance corresponding to 0.5 kg at 0.3 meter from the center (we will use S.I. units
here to make the math easier).
Newton says that:
F = (m ω2 r)/9.81
using S.I. units F is force in kg.F
m is the mass in kg
ω is the rotational speed in radians per second
and r is the radius at which the force (the out of balance) is acting
2
Data taken from TRW service catalog Form 382-14
3
Anti Friction Bearing Manufacturers Association
45
Vibration Analysis on Rotating Equipment
46
Vibration Analysis on Rotating Equipment
47
Vibration Analysis on Rotating Equipment
48
Vibration Analysis on Rotating Equipment
49
Vibration Analysis on Rotating Equipment
To confirm a suspected imbalance check the time waveform. An rotor imbalance will give a sine wave in
the velocity time signal with a period equal to the time it takes for one revolution of the rotor. Phase
readings will be steady .
Out of balance may occur in more than one plane. Most spectrum analyzers have a two plane balancing
function built into the software. Multiple plane balancing (more than two planes) is usually only necessary
on complex multiple disk rotors such as turbines which operate above their critical speed.
50
Vibration Analysis on Rotating Equipment
Misalignment
51
Vibration Analysis on Rotating Equipment
52
Vibration Analysis on Rotating Equipment
53
Vibration Analysis on Rotating Equipment
Looseness
54
Vibration Analysis on Rotating Equipment
55
Vibration Analysis on Rotating Equipment
56
Vibration Analysis on Rotating Equipment
57
Vibration Analysis on Rotating Equipment
58
Vibration Analysis on Rotating Equipment
59
Vibration Analysis on Rotating Equipment
Surging
Surging is a rather common problem encountered on high speed centrifugal and axial flow compressors and
occurs when the compressor is operated outside designed limits. Typically a compressor is designed to
deliver air or other gases over a specified mass flow range and at a specified pressure ratio. These
requirements are met by selecting rotor speed, number of compressor stages, blade configurations and other
factors. The manufacturer of the compressor can supply performance characteristics curves showing the
range of stable operation in terms of pressure ratio, mass flow and rpm. Attempting to operate the unit
outside the design range can result in excessive vibration and damage to the machine.
The problem of surge occurs when, for a particular operating speed, the delivery pressure to inlet pressure
ratio is too high or if mass flow is too low relative to design conditions. When this occurs, a reversal of gas
60
Vibration Analysis on Rotating Equipment
flow in the compressor will result. In the initial stages of surge, the flow reversal may only occur in the
boundary layers of the rotor blades (a rotating stall) however, at full surge the gas flow reverses its direction
and flows from the discharge to the inlet. Rotating stall shows itself as a vibration at approximately 35% of
run speed but is dependent on the physical configuration of the compressor.
The vibration characteristics resulting from compressor surging can vary depending on the extent of the
problem. In cases of mild surge, a noticeable increase in the vibration at blade passing frequency can
usually be detected. This frequency is the product of the number of rotor blades times the rpm of the rotor.
In other cases multiples of blade passing frequency may also be detected. When a full surge condition is
encountered the result may be a high amplitude of random, erratic vibration usually covering a rather broad
frequency range. This is caused by the turbulent flow within the compressor exciting the various natural
frequencies of the rotor wheels, rotor blades, diffuser blades, casing, shaft and other components. Of
course, if this condition is allowed to continue, extensive damage to the compressor c result.
61
Vibration Analysis on Rotating Equipment
Bearing Failures
Arguably the most common type of failure on rotating equipment are failures arising from the collapse of
the bearings.
Let us consider the lubrication of rolling element bearings.
We will discuss each stage of the failure and discover how each stage may be identified with predictive
maintenance techniques.
62
Vibration Analysis on Rotating Equipment
Figure 63 Loss of Lubricant - Ball Bearing Inner Race Courtesy of the Barden
Corporation
At this stage of the bearing failure the rolling elements have metal to metal contact onto the raceways
because the lubricant is no longer supporting the rolling element via a lubricant wedge. Because of this, the
metal to metal impacts excite the resonant frequencies of the bearing - just like hitting a bell with a hammer.
For a rolling element bearing these frequencies will be in the range of 1 to 4 kHz. The metal to metal
contacts also generate 4ultra sonic frequencies at between 30 and 50 kHz. Both of these frequency ranges
do not experience any modulation at this stage as the metal to metal contacts are irregular.
4
Ultra sonic - above the human audible range. Set at above 20 kHz for industrial applications.
63
Vibration Analysis on Rotating Equipment
64
Vibration Analysis on Rotating Equipment
65
Vibration Analysis on Rotating Equipment
66
Vibration Analysis on Rotating Equipment
67
Vibration Analysis on Rotating Equipment
68
Vibration Analysis on Rotating Equipment
Figure 69 Waterfall of early damage to a motor bearing collected every 1.5 hrs over
14 days
The plot above shows a waterfall display of acceleration vibration up to 2 kHz over a period of almost two
weeks on the drive end bearing of a drive motor. Notice that the vibration at about 1.5 kHz has risen
steadily. This frequency is typical of the bearing resonance. Notice also that there appear to be small
sidebands around the resonance frequency which are at the bearing outer race defect frequency. The
presence of a bearing defect frequency excited resonance does not give justification by itself to change the
motor bearings. We must also wait until the defect frequencies with sidebands and/or harmonics show up in
the velocity spectrum.
The severity chart in figure 69 is applicable for motors and fans running between 500 to 3600 rpm. The
envelope (or demodulation) amplitudes are quoted in dBG re 0.001G. As with any severity chart it is
important that the vibration analysts use their own judgment and experience when deciding whether or not
to change the bearing. For machines that are outside the speed ranges quoted the amplitudes will be lower
for slower machines and higher for faster machines. Velocity readings are heavily affected by the mass of
the machine so care should be taken when assessing very small or very large motors.
Don’t forget that roller bearings will stand more impactive forces than ball bearings. A roller bearing
should not be allowed to operate at above about 12 G (5true peak) and a ball bearing should not be allowed
to operate at above about 7 G (true peak).
5
True peak is defined as the peak seen in the waveform - not the derived peak seen in the frequency
spectrum.
69
Vibration Analysis on Rotating Equipment
The figure above was taken (as was most of these examples) from the drive end bearing of a drive motor.
Notice in the spectrum that there are no significant spikes but the spectrum is raised up from the floor - this
70
Vibration Analysis on Rotating Equipment
is a raised carpet level. A certain amount of rise from the floor is normal but when you find that your alarm
limits (based on the baseline) have been exceeded in the demodulated spectrum and there are no significant
spikes, the chances are that the bearing is starting to suffer lubrication problems.
In the case of a main drive motor the bearing is almost certainly grease lubricated. The grease lubricated
bearing may be of shielded (or sealed) construction or non-shielded construction. If the bearing is fitted
with seals then it is sometimes possible to force some grease pass the seal with a grease gun, but usually you
will just have to watch the bearing deteriorate and change the bearings before the damage gets so bad as to
cause secondary damage.
If the bearing is non-shielded then the bearing should be lubricated following the bearing and motor
manufacturers instructions. If a plug is fitted opposite the grease fitting make sure that you remove it.
Injecting too much grease into the bearing cavity will cause pressurization of the cavity and the grease will
force its way past the bearing into the motor windings
As you inject the grease into the bearing have a spectrum analyzer attached to an accelerometer on the
bearing housing and watch the vibration levels. A simple rule of thumb for bearing condition is that if the
vibration goes down and stays down, the bearing only had a lubrication problem and you have just fixed it.
If the vibration level goes down but rises again then the bearing is damaged - the sooner the level rises again
the worse condition the bearing is in (from a few of days to several minutes for a very bad bearing).
71
Vibration Analysis on Rotating Equipment
0.872
1.000
1.000
72
Vibration Analysis on Rotating Equipment
2.500
2.500
The two time waveform spectra above show the individual impacts caused by the rollers impacting onto the
damaged raceways. Generally, a ball bearing should not be allowed to run with impacts of more than 0-7 G
peak in the time domain and a cylindrical roller bearing should be running less than 12 G. If these figures
are exceeded then the bearing is almost certainly severely damaged.
Looking at the time interval between the peaks in the time domain, we can correlate this time difference
with the frequencies seen in the frequency spectrum.
0.0463
73
Vibration Analysis on Rotating Equipment
As the bearing deteriorates the bearing defect frequencies start to show up in the velocity spectrum. Figure
75 shows a velocity spectrum from a spalled bearing with multiple harmonics of the outer race defect
frequency. As a general rule of thumb, if you see the same defect frequencies in the demodulated
acceleration and the velocity spectra, irrespective of amplitude, then that bearing is spalled.
74
Vibration Analysis on Rotating Equipment
Balancing
Imbalance has been named the most common cause of vibration in our machines. Since this is
true it would be to our advantage to be able to correct this condition as easily as possible. Before
we can balance a part with the Vibration Analyzer, certain conditions must be satisfied. The
vibration must be due to imbalance, we must be able to make the weight corrections on the rotor,
and we must have a shaft angle reference (such as being able to observe the rotor with the strobe
light as used in this paper, or other type of key phasor) for phase measurements.
In-place Balancing
The conditions which must be satisfied so you can balance can almost always be met with the
imbalanced part mounted in its own bearings, operating as it normally does. The process of
balancing a part without taking it out of the machine is called In-place Balancing. The following
pages deal directly with this application. In-place Balancing eliminates the need to disassemble
the machine, transport the part to a balancing machine, balance under artificial conditions, and
assures smooth operation of the machine when you are done.
In-place Balancing is a straight forward process which involves following a few simple rules.
However, before we discuss balancing we should first understand imbalance, where it comes
from and what must be done to correct it.
75
Vibration Analysis on Rotating Equipment
76
Vibration Analysis on Rotating Equipment
77
Vibration Analysis on Rotating Equipment
F = 1/16(RPM/1000)2 gram-inches.
The amount of force generated by one ounce-inch of imbalance at high speeds is surprising and
explains why good balance becomes absolutely necessary for high speed machines. For
instance, at 3600 RPM nearly 23 pounds of force is generated for each ounce-inch of imbalance.
78
Vibration Analysis on Rotating Equipment
If an imbalance weight is added to a perfectly balanced part it will have a vibration at a frequency
equal to the rotating speed of the part. The part will vibrate a certain number of mils and the
reference mark will appear at some definite position when the strobe flashes. The amount of
vibration is 5 mils and the reference mark appears at 12: 00 o'clock. If we double the amount of
the imbalance weight without changing its position the vibration will increase to 10 mils and the
reference mark will still appear at 12: 00 o'clock when the strobe light flashes. This experiment
shows .it is correct to use the amount of vibration to tell us how much imbalance is in the part.
Note that the reference mark appeared at the same place each time.
What happens when the position of the heavy spot or imbalance weight is changed? If we move
the heavy spot 30 degrees clockwise from where it was before we will see the same amount of
vibration, 10 mils, but the reference mark will now appear at 11: 00 o'clock or 30 degrees
counterclockwise from where it was before. If we move the weight 30 degrees counterclockwise
from its first position we will still see the same amount of vibration, 10 mils, but the reference mark
will appear at 1: 00 o'clock or 30 degrees clockwise from it’s first position of 12: 00 o'clock.
The experiment tells us two things:
1. The amount of vibration is proportional to the amount of the imbalance.
2. The reference mark shifts in the opposite direction to a shift of the heavy spot. The angle the
reference mark shifts is equal to the angle of shift of the heavy spot.
How can we use this information to balance?
How to Balance - Single Plane
We said earlier that balancing is the process by which we learn the size and position of the heavy
spot so we can either add an equal amount to the opposite side of the rotor or remove weight at
the heavy spot. At the start of a balancing operation we do not have the least idea how large the
heavy spot is nor do we know where on the part it is located. The imbalance in the part at the start
is called the original imbalance and the vibration amount and phase which represents that
imbalance is called our original reading. We change the original imbalance by adding a trial weight
to the part. The new total imbalance in the part will be represented by a new amount and phase of
vibration. The change caused by the trial weight can be used to learn the size and location of the
original imbalance, or where the trial weight must be placed to be opposite the original imbalance
and how large the trial weight must be to be equal but opposite the original heavy spot.
Starting with the part out of balance, we see 1.8 mils of vibration and a reference mark position at
12:40 o'clock. What we must do is find a weight and position for that weight which would oppose
the original imbalance. Let's start by adding a trial weight. Three things can happen.
First, if we are lucky we might add the trial weight right on the heavy spot. If we do, the vibration
will in- crease and the reference mark will appear in the same position it did on the original run. To
balance the part all we have to do is move the trial weight directly opposite its first position and
adjust the amount until we reach a satisfactory balance.
The second thing that could happen is we could add the trial weight in exactly the right place
opposite the heavy spot. If the trial weight were smaller than the imbalance, then the heavy spot
would still be the heavy spot but we would see a decrease in vibration and the reference mark
would be in the same place. To balance all we would have to do would be to add more weight until
we reached a satisfactory vibration level. If the trial weight were larger than the imbalance then it’s
position would now be the heavy spot and the reference mark would shift 180° or directly opposite
where it was originally. To balance we would reduce the trial weight amount until we reached a
satisfactory vibration level.
The third thing that can happen is the usual one where the trial weight is added neither at the
heavy spot or opposite it. When this happens the reference mark shifts to a new position and the
vibration displacement changes. By what angle should the trial weight be shifted, in which
direction, and should its amount be in- creased or decreased to be equal but opposite the original
imbalance?
79
Vibration Analysis on Rotating Equipment
80
Vibration Analysis on Rotating Equipment
original position then 'we must more the trial weight counterclockwise or shift the trial weight
opposite the shift of the reference mark.
In the example shown the trial weight is larger than it needs to be. We know this because the trial
vector, line B, is longer than the “O'' vector. To determine the correct amount of the trial weight we
simply multiply the trial weight by the length of the original imbalance vector and then divide by the
length of the trial weight vector.
By inspecting our vector parallelogram we can see that the line A connecting the ends of the
original imbalance vector and the original plus trial weight imbalance vector is exactly the same
length as the trial weight vector line B. Furthermore, the angle between the original imbalance
vector and line A, is the same as the angle between the trial weight vector, line B, and its position
to get the desired effect to balance the part, the dotted line. Therefore, to solve the vector problem
we actually need only the three lines, O, O + T, and A.
81
Vibration Analysis on Rotating Equipment
82
Vibration Analysis on Rotating Equipment
Amount Phase
1. Original Reading 5.0 mils @ 3: 00 o'clock
First Trial Reading 3.0 mils @ 4: 00 o'clock
Answer.
The reference mark moved from 3: 00 to 4: 00, a clock- wise shift. The trial weight must be moved
counterclockwise from its position for the trial run.
83
Vibration Analysis on Rotating Equipment
The solution is to move the trial weight 32° counterclockwise from its position on the first trial run.
At the same time increase the size of the trial weight 1.75 times its original size.
Now, you work the rest.
Amount Phase
2. Original Reading 3.0 mils @ 7: 00 o'clock
First Trial Reading 4.0 mils @ 7: 00 o'clock
84
Vibration Analysis on Rotating Equipment
After the corrections indicated by the readings above have been made a new reading was taken
indicating additional correction is required.
Second Trial Reading 3.0 mils @ 4: 45 o'clock
What additional correction is required? Check your answers to these problems with the answers
supplied further in the manual. Then try some real problems on a simple system like the one
pictured here in the text.
Balancing in One Run
Do not let the title of this section fool you. To date no method has been devised to permit
balancing in only one run the first time a unit is balanced. However, once a unit has been
balanced by either of the methods described before it is possible to learn how much and where
weight must be added to balance the unit in one run. This information can be learned for any
system but applies only to the individual system. If the system is changed in any way, how much
and where weight must be added will also change. To show how this is done we will use the motor
and pulley system illustrated before.
We will assume that it is the pulley that is out of balance. With the pickup mounted in the vertical
direction and the filter tuned to rotating speed of the pulley we measure the vibration caused by
the imbalance. This is the original reading which, for the example is 5.1 mils with the reference
mark stopped at 2: 00 o'clock. After the part is balanced to a vibration of 0.1 mil we can learn how
much weight was added to correct for 5.0 mils of vibration. We will assume that 20 grams of
weight added at a radius of 2.0 inches was required. The imbalance corrected was therefore 40
gram-inches. Therefore, 40 gram-inches of balance correction was required for g 5.0 mils of
vibration or 8 gram-inches per mil.
The next time it is necessary to balance this particular unit or one just like it we will know 8 gram-
inches of correction is required for each mil of vibration. For example if we measure 3 mils, 24
gram-inches is required.
lf the weight is added at two inches 12 grams will be needed. This is true of course only for the
machine mounted the same way, operating at the same speed, and with the vibration pickup in
the same place.
Knowing how much weight to add is important, of course, but where is even more important.
Before a trial weight was added to the pulley the phase or position of the reference mark was 2:00
o'clock. The pulley's position with the reference mark at 2: 00 o'clock is the position of the pulley
when the strobe light flashed. With the reference mark at 2:00 o'clock we can see in the
illustration that the weight was added at 5:00 o'clock. The heavy spot or point of imbalance is 180°
away or at 11: 00 o'clock when the strobe light flashed.
85
Vibration Analysis on Rotating Equipment
86
Vibration Analysis on Rotating Equipment
Material:
Bright Steel or
Austenitic Stainless
Note:
One side to be machined smooth and
the other side to be machined to a
rough “gramaphone” finish.
1/4”
87
Vibration Analysis on Rotating Equipment
Particularly on motors, care should be taken to ground the shield wire at one end only - the end furthest
away from the motor - on to a good electrical ground. Failure to do this will result in high amplitude
multiples of electrical line frequency (60 Hz). Do not connect both ends of the shield to ground as you will
get beautiful ground loops that will really dominate your signal. Case isolated accelerometers are usually
less susceptible to picking up electrical frequencies. When running your accelerometer cable back to the
junction box, keep the cable as far away as possible from the motor power cable to avoid “cross-talk”.
A speed output should be installed for variable speed motors. This could be in the form of a T.T.L. pulse,
once per rev from an installed proximity switch, triggered by a key or from the installed tachometer channel
on the P.L.C. which gives a 4-20 mA output. To use this output the 4-20 mA must be dropped across a
resistor of about 6250 Ω to give a 2.5 volt drop for a 10 mA signal and 5 volt at full speed (20 mA). Before
installing this resistor check with the engineers responsible for the drive system to make sure the drive
system or the P.L.C. is not adversely affected by the resistor. Once installed the speed input will have to be
calibrated in volts / rpm and fed to a channel input as opposed to a trigger input as would be required for the
T.T.L. pulse..
6
V = I.R where V is volts, I is amps, R is ohms.
88
Vibration Analysis on Rotating Equipment
Thin film of
silicone grease
Steel stud
Max temp 1000°C (1800°F)
Thin film of
silicone grease
Mica washer
Steel stud
Thin layer of
bees wax
Methyl cyanoacrylate
cement (super glue)
Methyl cyanoacrylate
cement (super glue)
soft glue
Steel stud
89
Vibration Analysis on Rotating Equipment
H a n d D u a l R a il F la t M o u n t in g A d h e s iv e S tu d
P ro b e M a g n e t M a g n e t P a d M o u n t M o u n t
Sensitivity Deviation
+ 4 0
+ 3 0
+ 2 0
+ 1 0
0
-1 0
-2 0
1 .0 1 0 1 0 0 1 0 0 0 1 0 0 0 0 1 0 0 0 0 0
L o g F re q u e n c y (H z )
90
Vibration Analysis on Rotating Equipment
Introduction
The purpose of monitoring equipment in predictive maintenance (PdM) is to be able to assess the health and
condition of the machine relative to any potential failures. In order for us to carry out this assessment we
must be absolutely certain that we are taking the correct measurements at the right place and that we are
taking them often enough so that we do not miss a developing failure. The methodology that we will
employ to ensure that we are taking the right measurements with the correct parameters is the potential
failure analysis (PFA) tree.
Let us take each of these components and look at them in detail from the roots up.
Base cause
As the name suggests this is the root cause of any potential failure. Examples could be: lubrication
problems, misalignment, manufacturing defects and so on. The base cause often branches out to more than
one failure type, for example misalignment could cause a bearing failure or a shaft breakage.
Failure type
This is the failure that we could expect from the base cause. In other words this is a very short description
of the actual failure that the machine would suffer should the base cause carry on without remedial work
being carried out.
91
Vibration Analysis on Rotating Equipment
External manifestation
This is where we start getting technical. How will the problem show itself at the various stages of failure?
For illustration we will discuss the four stages of bearing failures and how they show themselves. An
example would be that a misalignment would show itself as an increase in vibration at the machine, whereas
a plugged heat exchanger would show itself as a rise in differential pressure and with a change in
temperature from optimum of the cooled fluid.
Technology
So now we know how the problem will show it’s ugly head but what is the best technology to detect it. We
have already said that misalignment shows itself as a rise in vibration so the technology to use would be a
vibration based technology. The heat exchanger temperature change may be detected by a mercury-in-glass
thermometer or alternatively by the use of infra-red technology. At this stage you need to have detailed
understanding of the failure modes and predictive maintenance applications.
Parameter
So the technology of choice for detection misalignment is vibration, but what type of vibration is best? In
this case the best vibration parameter is velocity. For the heat exchanger we may decide that the most
suitable parameter is a radiometric thermal image or maybe we will decide to use a simple point and shoot
infra-red thermometer.
Analysis
At this point we are giving the analyst an idea of what detail he should expect to see in the chosen parameter
when the subject machine has a developing defect. Our misalignment will show as an increase in velocity
vibration in the axial direction for a belt driven machine at run speed with two or three harmonics and our
thermal image will show a high temperature gradient across the cooler.
Interval
How often do we need to take the reading to ensure that we do not miss a developing problem. For example
if we are looking for rolling element bearing defects we should take the readings at least once a month but if
we are looking for a misalignment then once every three months would be ample.
Setup
To detect the misalignment we have said that we will use vibration technology with a velocity parameter
and that we are looking for two or three multiples of run speed in the frequency spectrum. If that is the only
defect that we are looking for then we can safely set the maximum frequency in the spectrum at about ten
time run speed with 400 lines of resolution as we are not looking for small changes in frequency such as slip
sidebands.
92
Vibration Analysis on Rotating Equipment
Consider figure 1, above. It shows the four stages of a rolling element bearing failure. Before you can
develop the PFA tree for the rolling element bearing you must have a thorough understanding of the
possible failure modes and extrapolate these modes to their logical conclusions.
Stage 1
Notice that the first stage of the defect is linked to lubrication problems. At this stage we can detect any
problems using high frequency vibration parameters such as demodulation or ultra sonics. To detect a
defect at this stage with vibration we need to use a high frequency technique such as envelope signal
processing (ESP) with the following parameters:
Fmax = BPFI x 8 (or thereabouts)
Envelope filter = 2.5 - 5 kHz (for electric motor bearings up to about 250 HP)
No lines = 400 or 800
Window = Hanning, Hamming or Kaiser
Averages = typically 4 with maybe 50% overlap
If your data collector does not support a high frequency function such as ESP or HFB then take an overall
acceleration reading.
93
Vibration Analysis on Rotating Equipment
Stage 2
The next stage of failure involves light marking of the bearing. Again the parameter of choice is a high
frequency technique but experience tells us that a spectral parameter is most suited to detecting marked
races. So the choice here would almost certainly be ESP. The setup for the ESP reading would be the same
as in stage 1. The onset of stage 2 can be also be detected by searching for a rise in amplitude in resonance
frequencies so if you do not have ESP you can take a velocity or acceleration spectrum with an Fmax above
resonance. If we expect resonance to occur at about 2 kHz then the Fmax should be about 3 kHz. Note that
if you are using seismic velocity transducers you will not be able to look this high in the frequency range.
Stage 3
The third stage of the bearing failure is when the bearing starts to spall. At this stage we start to see the
defect in velocity usually at the third or fifth harmonic of the bearing defect frequency. We may also see
sidebands of run speed or cage frequency around the defect frequencies or their harmonics. We should be
taking an ESP and a velocity reading. The ESP reading should be the same as that set up in stage 1. The
velocity reading should be set up as follows:
Fmax = BPFI x 8 (if also using ESP or similar reading)
or = 2.5 kHz (do not confuse AC motor frequencies with a brg defect)
No lines = 400 or 800
Window = Hanning, Hamming or Kaiser
Averages = typically 4 with maybe 50% overlap
We should also take a time waveform reading in G acceleration to look for the bearing impacts at all stages
of the failure. The time duration should be calculated to be enough to show about 3 revs of the shaft. We
also need enough resolution in the time domain to be able to differentiate the impacts and calculate the
frequencies of occurrence of the impacts. Usually 1,024 samples is adequate resolution. To calculate the
Fmax in the point setup screen we use the following formula:
Stage 4
The final stage of failure is when the bearing physically collapses. At this stage the bearing overheats
dramatically and very quickly. Temperature measurement using embedded thermo-couples have been used
94
Vibration Analysis on Rotating Equipment
successfully for protection systems but usually only for thrust bearings which develop the raised
temperature before standard radial support bearings.
Dry/Marked Bearing Failure High frequency Vibration Demodulation Spikes @ Brg defect On Line Fmax = BPFI x 8
bearings resonance @ bearing (Ultra Sound) frequencies (20 dBG) Radial
defect frequency intervals See sect. 2.5 G acceleration
Time
Waveform Peaks in waveform On Line = 3 revs
Spalled Bearing Failure Harmonics of bearing Vibration Velocity Spikes @ Brg defect On Line Fmax = BPFI x 8
bearings defect frequencies freqys. & harmonics (higher if not using
and/or sidebands. resonance parameter)
See sct. 2.5 Radial
Misalignment Bearing Failure/ High axial vibration Vibration Velocity Multiples of run speed On Line Fmax = 10 x rpm
Shaft Failure axial > radial Axial
Looseness Bearing Failure/ Run speed harmonics Vibration Velocity Multiples of run speed On Line Fmax = 10 x rpm
Shaft Failure/ & subharmonics up to 15x. Possible Vertical
Structural Failure sub-harmonics
Bad S.C.R.s Reduced Power Vibration @ SCR Frqys Vibration Velocity 240 Hz, 1/3 DC pulse, On Line Fmax = 120 kcpm
Motor Burnout s/band on DC(Sect.2.5) Drive End
Changes in SCR temps. Thermography Thermal Image Bad SCR is colder 6 month Compare SCR to SCR
Different current / Current Draw Amps Unbalanced supply 6 month Compare phase to phase
input phase
Winding Motor burnout Stator temp. rise/ Thermography Thermal image Hot spot on stator 6 month 1 baseline
defects Uneven temp. distribution (spot temp)
Earth current leakage Insulation test Megger Should be > 1.5 MΩ 6 month Motor de-energized
Rotor Rotor failure Line frequency + Motor load Motor current Sidebands of slip x No. 6 month Press idling
defects sidebands current analysis frequency of poles around 60 Hz
(Sect. 2.2)
Motor burnout Vibration @ rotor bar Vibration Velocity Sidebands of 1x On Line Fmax = RBPF x 3
pass frequency (Sect 2.2) Radial
c. Ron Frend - PreDiCon
95
Vibration Analysis on Rotating Equipment
Conclusion
The method outlined in this paper for setting up the type of readings in a predictive maintenance system will
result in quantifiable, repeatable and useful measurements if these factors are taken into consideration:
• The engineer or technician defining the measurements must be familiar with previous failures
encountered on machines similar to the machine under scrutiny
• The engineer or technician defining the measurements should be cognizant of all
• available predictive maintenance technologies.
• The engineer or technician defining the measurements must be aware of the limitations of the
technology which is used for the measurement.
Finally, don’t limit yourself to only taking vibration measurements - there are a host of other technologies
out there which complement vibration in a predictive maintenance application.
CHUTE ASSEMBLIES
DRIVE ASSEMBLY
96
Vibration Analysis on Rotating Equipment
BEARING COLLAPSE
PROBLEMS
OVER TEMPERATURE
ULTRA-LOW SPEED
FORIEGN BODY INGRESS BEARING DEFECTS
PARTIAL ROTATION
CORROSION
INNER BEARING MOUNTS
BRINNELLING
PROBLEM
FREQUENCY RESOLUTION NOT POSSIBLE
TIME-BASE READINGS REQ'D IN : ACCELERATION
STRESS WAVE
RADIATION
HEAD 2
97
Vibration Analysis on Rotating Equipment
Dry/Marked Bearing Failure High frequency Vibration Demodulation Spikes @ Brg defect On Line Fmax = BPFI x 8
bearings resonance @ bearing (Ultra Sound) frequencies (20 dBG) Radial
defect frequency intervals See sect. 3.5 G acceleration
Time
Waveform Peaks in waveform On Line = 3 revs
Spalled Bearing Failure Harmonics of bearing Vibration Velocity Spikes @ Brg defect On Line Fmax = BPFI x 8
bearings defect frequencies freqys. & harmonics (higher if not using
and/or sidebands. resonance parameter)
See sct. 3.5 Radial
Looseness Bearing Failure/ Run speed harmonics Vibration Velocity Multiples ofBPFO On Line Fmax = 10 x rpm
Shaft Failure/ & subharmonics or BPFI in velocity Vertical
Structural Failure Sect. 3.5 1 Baseline
98
Vibration Analysis on Rotating Equipment
Measurement Windows
Many people get confused with this topic. What is windowing? Let's go back to the time record. What
happens if a signal is not exactly periodic within the time record? We said that its amplitude is divided into
multiple adjacent frequency bins. This is true but it's actually a bit worse than that. If the time record does
not start and stop with the same data value, the signal can actually smear across the entire spectrum. This
smearing will also change wildly between records because the amount of mismatch between the starting
value and ending value changes with each record.
If a sine wave is passing through zero at the beginning and end of the time series, the resulting FFT
spectrum will consist of a single line with the correct amplitude and at the correct frequency. If, on the
other hand, the signal level is not at zero at one or both ends of the time series record, truncation of the
waveform will occur, resulting in a discontinuity in the sampled signal. This discontinuity causes problems
with the FFT process, and the result is a smearing of the spectrum from a single line into adjacent lines.
This is called "leakage"; energy in the signal "leaks" from its proper location into the adjacent lines.
Leakage could be avoided if the time series zero crossings were synchronized with the sampling times, but
this is impossible to achieve in practice. The shape of the "leaky" spectrum depends on the amount of
signal truncation, and is generally unpredictable for real signals.
In order to reduce the effect of leakage, it is necessary that the signal level is forced zero at the beginning
and end of the time series. This is done by multiplying the data samples by a "smoothing window" function,
which can have several different shapes. The difference between each smoothing window is the way in
which they transition from the low weights near the edges to the higher weights near the middle of the
sequence. If there is no windowing function used, this is called "Rectangular", "Flat", or "Uniform"
windowing.
While the smoothing window does a good job of forcing the ends to zero, it also adds distortion to the time
series which results in sidebands in the spectrum. These sidebands, or side lobes, effectively reduce the
frequency resolution of the analyzer; it is as if the spectral lines are wider. The measured amplitude of the
weighted signal is also incorrect because a portion of the signal level is removed by the weighting process.
To make up for this reduction in power, windowing algorithms give extra weight to the values near the
middle of the sequence.
Windows are functions defined across the time record which are periodic in the time record. They start and
stop at zero and are smooth functions in between. When the time record is windowed, its points are
multiplied by the window function, time bin by time bin, and the resulting time record is by definition
periodic. It may not be identical from record to record, but it will be periodic (zero at each end).
In the frequency domain, a window acts like a filter. The amplitude of each frequency bin is determined by
centering this filter on each bin and measuring how much of the signal falls within the filter. If the filter is
narrow, then only frequencies near the bin will contribute to the bin. A narrow filter is called a selective
window - it selects a small range of frequencies around each bin. However, since the filter is narrow, it falls
off from center rapidly. This means that even frequencies close to the bin may be attenuated somewhat. If
the filter is wide, then frequencies far from the bin will contribute to the bin amplitude but those close by
will not be attenuated significantly.
The net result of windowing is to reduce the amount of smearing in the spectrum from signals not exactly
periodic with the time record. The different types of windows trade off selectivity, amplitude accuracy, and
noise floor.
Several types of window functions are available including Uniform (none), Flattop, Hanning, Blackman-
Harris, and Kaiser.
Uniform
The uniform window is actually no window at all. The time record is used with no weighting. A signal will
appear as narrow as a single bin if its frequency is exactly equal to a frequency bin. (It is exactly periodic
99
Vibration Analysis on Rotating Equipment
within the time record). If its frequency is between bins, it will affect every bin of the spectrum. These two
cases also have a great deal of amplitude variation between them (up to 4 dB).
In general, this window is only useful when looking at transients which do not fill the entire time record.
Hanning
The Hanning window is the most commonly used window. It has an amplitude variation of about 1.5 dB
(for signals between bins) and provides reasonable selectivity. Its filter roll off is not particularly steep. As
a result, the Hanning window can limit the performance of the analyzer when looking at signals close
together in frequency and very different in amplitude.
Flattop
The Flattop window improves on the amplitude accuracy of the Hanning window. Its between-bin
amplitude variation is about 0.02 dB. However, the selectivity is a little worse. Unlike the Hanning, the
Flattop window has a wide pass band and very steep rolloff on either side. Thus, signals appear wide but do
not leak across the whole spectrum.
Blackman-Harris
The Blackman-Harris window is a very good window to use with the spectrum analyzer. It has better
amplitude accuracy (about 0.7 dB) than the Hanning, very good selectivity and the fastest filter rolloff. The
filter is steep and narrow and reaches a lower attenuation than the other windows. This allows signals close
together in frequency to be distinguished, even when their amplitudes are very different.
Kaiser
The Kaiser window, which is available on IRD analyzers, combines excellent selectivity and reasonable
accuracy (about 0.8 dB for signals between exact bins). The Kaiser window has the lowest side-lobes and
the least broadening for non-bin frequencies. Because of these properties, it is the best window to use for
measurements requiring a large dynamic range.
Averaging
In general, averaging many spectra together improves the accuracy and repeatability of
measurements.
RMS Averaging
RMS averaging computes the weighted mean of the sum of the squared magnitudes (FFT times its complex
conjugate). The weighting is either linear or exponential.
RMS averaging reduces fluctuations in the data but does not reduce the actual noise floor. With a sufficient
number of averages, a very good approximation of the actual random noise floor can be displayed.
Since RMS averaging involves magnitudes only, displaying the real or imaginary part or phase of an RMS
average has no meaning. The RMS average has no phase information.
100
Vibration Analysis on Rotating Equipment
Vector averaging requires a trigger. The signal of interest must be both periodic and phase synchronous
with the trigger. Otherwise, the real and imaginary parts of the signal will not add in phase and instead will
cancel randomly.
With vector averaging, the real and imaginary parts as well as phase displays are correctly averaged and
may be displayed. This is because the complex information is preserved.
Peak Hold
Peak Hold is not really averaging, instead, the new spectral magnitudes are compared to the previous data,
and if the new data is larger, then the new data is stored. This is done on a frequency bin by bin basis. The
resulting display shows the peak magnitudes which occurred in the previous group of spectra.
Peak Hold detects the peaks in the spectral magnitudes and only applies to Spectrum, PSD, and Octave
Analysis measurements. However, the peak magnitude values are stored in the original complex form. If
the real or imaginary part or phase is being displayed for spectrum measurements, the display shows the real
or imaginary part or phase of the complex peak value.
Linear Averaging
Linear averaging combines N (number of averages) spectra with equal weighting in either RMS, Vector or
Peak Hold fashion. This type of averaging is useful for eliminating transients.
Exponential Averaging
Exponential averaging weights new data more than old data. Averaging takes place according to the
formula,
New Average = (New Spectrum * 1/N) +(Old Average) * (N-l)/N
where N is the number of averages.
Exponential averages "grow" for approximately the first 5N spectra until the steady state values are reached.
Once in steady state, further changes in the spectra are detected only if they last sufficiently long. Make
sure that the number of averages is not so large as to eliminate the changes in the data that might be
important.
101
Vibration Analysis on Rotating Equipment
Overlap Percentage
The amount of overlap is specified as a percentage of the time record. 0% is no overlap and 99.8% is the
typical maximum. The maximum overlap is determined by the amount of time it takes to calculate an FFT
and the length of the time record and thus varies according to the span. For vibration analysis of rotating
machinery a good overlap is 50% as this ensures that no data is zeroed out by the smoothing windows, yet
sufficient samples are gathered for a valid analysis.
Octave Analysis
The magnitude of the normal spectrum measures the amplitudes within equally divided frequency bins.
Octave analysis computes the spectral amplitude in logarithmic frequency bands whose widths are
proportional to their center frequencies. The bands are arranged in octaves with either 1 or 3 bands per
octave (1/1 or 1/3 octave analysis). Octave analysis measures spectral power closer to the way people
perceive sound, that is, in octaves. For vibration measurements, octave analysis is now only used for sound
analysis or for quality acceptance checks at the end of an assembly line but is included here for the sake of
completeness.
102
Vibration Analysis on Rotating Equipment
Analysis
The following procedure gives an overview of the procedure to take in the analysis of
vibration frequency spectra.
Predictive Maintenance Procedure
Date:July 11, 2003
Procedure Name: Signature Analysis
Purpose: To provide a routine procedure for analysis of vibration in order to promote
understanding of the relationship between vibration frequencies and their causes.
Activities:
1) Select the first plot of the machine which will be at the driver outboard and will
be an enveloped acceleration reading. Identify the run speed accurately. If you
have frequency information for the machine ensure that the reference speed is
accurate - if not you must change the speed reference before continuing.
2) In the envelope spectrum see if any of the generated frequencies coincide with (or
are close to) any significant spikes. Remember that the bearing frequencies may
not be completely accurate if the bearing which has been nominated in the
frequency setup has been replaced with an equivalent.
• As a rule of thumb:
• Ball Pass Frequency Outer Race ≅ Run Speed x No. Of Rolling
Elements x 0.4
• Ball Pass Frequency Inner Race ≅ Run Speed x No. Of Rolling
Elements x 0.6
• Cage Frequency ≅ Run Speed x 0.4
• Note that this vibration is not necessarily direction specific.
3) Once a spike at a bearing frequency has been identified you should check the
baseline for this type of machine for the trend. If the trend is deteriorating then
further checks are necessary. Be careful that you do not confuse a run speed
harmonic or an electrical frequency with a bearing defect frequency. One
common bearing frequency is just over 3 x run speed for BPOR on a 8 element
bearing. The run speed of an electric motor cannot exceed the electrical speed so
harmonics of run speed cannot have a frequency even slightly more than 3600
cpm (for a 1200 rpm motor) or 5400 cpm (for a 1800 rpm motor) or 10,800 cpm
(for a 3600 rpm motor) - if the spike is even at a slightly higher frequency then it
is likely caused by a bearing defect otherwise it is likely a run speed harmonic or
an electrical frequency. Check the time domain signal for “angel fish” patterns.
103
Vibration Analysis on Rotating Equipment
4) In the single spectrum plot double click on the convert the display to dB(G)
(referenced to 0.001 G). Identify the carpet amplitude in dB(G) - this is the
average amplitude (excluding spikes). Identify the amplitude of the spikes above
the carpet level. The following rules of thumb apply to enveloped acceleration
levels in dB(G):
• > 20 dB(G) rise in carpet level - under lubrication
• 10-15 dB(G) spike above carpet - minor marking
• > 20 dB(G) spike above carpet - marked race.
5) If the spectrum shows multiples of run speed then there is an impact every rev of
the rotor, with possible looseness if there are many multiples. If the bearing
defect frequencies have sidebands of cage frequency then there is a FALSE
BRINELLING problem. If the bearing defect frequency and harmonics have
sidebands of run speed then there is probably a defect on the inner race. As the
defect deteriorates then the carpet level will rise and the sidebands and harmonics
will increase in amplitude up to a certain amplitude then stop. The carpet level
will continue to rise as random marking occurs around the bearing and may rise to
mask the spikes completely.
6) Move to the velocity spectrum for the same point but with amplitude set to linear.
Check to see if there are any spikes in velocity at the bearing defect, harmonics of
the defect and/or sidebands of the defect - look particularly for the third and fifth
harmonics. If any spike at these frequencies exist then there is physical spalling of
the race. If the amplitude of the spike reaches 1 mm/s then the spalling is severe.
7) In the velocity spectrum the following patterns indicate the associated defects:
Dominant Secondary Harmonics Sidebands Dominant Defect Suggested
Frequency Frequency of Direction Maximum
Dominant Amplitude
Frequency @
Dominant
Frequency
1x Nil Nil Nil Radial Imbalance 6 mm/s
1x 1/2 or 1/3 Multiple Nil Radial Looseness 3 mm/s
x
1x 3x 2 or 3 Nil Axial Misalignment 4 mm/s
BPFO / BPFI 1x Multiple 1x / Cage Radial Bearings 0.5 mm/s
7,200 cpm Rotor Bar 2 or 3 2xLF of RBF Any Electrical 5 mm/s
Any Any Nil Nil Any Resonance 7 mm/s
Gear Mesh 1x 3 or 4 1x Radial Gearing 1 mm/s
2x Belt 1x 2 or 3 N/A Radial Belts 5 mm/s
104
Vibration Analysis on Rotating Equipment
Severity charts
Figure 95 shows a general severity chart for vibration which is in widespread use. There are many versions
of these charts. The best use for these charts is for new or rebuilt equipment acceptance limits.
Dependency on these charts can be confusing, especially if the bedplate is flexible or on resilient mounts.
105
Vibration Analysis on Rotating Equipment
Transfer Function
The transfer function is the ratio of the spectrum of channel 2 to the spectrum of channel 1. For the transfer
function to be valid, the input spectrum must have amplitude at all frequencies over which the transfer
function is to be measured.
Cross Spectrum
The cross spectrum is defined as:
cross spectrum = FFT2 conj(FFT1)
The cross spectrum is a complex quantity which contains magnitude and phase information. The phase is
the relative phase between the two channels. The magnitude is simply the product of the magnitudes of the
two spectra. Frequencies where signals are present in both spectra will have large components in the cross
spectrum.
Orbit
The orbit is simply a two dimensional display of the time record of channel 1 vs. the time record of channel
2. The orbit display is similar to an oscilloscope displaying a "Lissajous" figure.
Coherence
Coherence measures the percentage of power in channel 2 which is caused by (phase coherent with) power
in the input channel. Coherence is a unit-less quantity which varies from 0 to l. If the coherence is 1, all the
power of the output signal is due to the input signal. If the coherence is 0, the input and output are
completely random with respect to one another. Coherence is related to signal to noise ratio (S/N) by the
formula:
S/N = γ2/(l -γ2)
where γ2 is the traditional notation for coherence.
Correlation
The two channel analyzer may also compute auto and cross correlation. Correlation is a time domain
measurement which is defined as follows:
Auto Correlation(τ) = ∫x*(t)x(t-τ)dt
Cross Correlation(τ) = ∫x*(t)y(t-τ)dt
where x and y are the channel 1 and channel 2 input signals and the integrals are over all time. It is clear
that the auto correlation at a time “t” is a measure of how much overlap a signal has with a “delayed-by-t”
version of itself, and the cross-correlation is a measure of how much overlap a signal has with a “delayed-
by-t” version of the other channel. Although correlation is a time domain measurement some analyzers use
frequency domain techniques to compute it in order to make the calculation faster.
106
Vibration Analysis on Rotating Equipment
Spectrum
Although the linear magnitude scale is used most often for displaying spectra, another way of displaying
amplitude is the Log Magnitude. The Log Mag display graphs the magnitude of the spectrum on a
logarithmic scale.
Why is the Log Mag display useful? Remember that the 16 bit analyzer has a dynamic range of about 90
dB. below full scale. Imagine what something 0.01% of full scale would look like on a linear scale. If we
wanted it to be 1 inch high on the graph, the top of the graph would be 833 feet above the bottom - It turns
out that the log display is both easy to understand and shows features which have very different amplitudes
clearly.
The real and imaginary parts are always displayed on a linear scale. This avoids the problem of taking the
log of negative voltages.
Phase
In general, phase measurements are only used when the analyzer is triggered. The phase is relative to the
pulse of the trigger.
The phase is displayed in degrees or radians on a linear scale, usually from -180 to +180 degrees.
The phase of a particular frequency bin is set to zero in most analyzers if neither the real nor imaginary part
of the FFT is greater than 0.012% of full scale (-78 dB below f.s.). This avoids the messy phase display
associated with the noise floor. (Remember, even if a signal is small, its phase extends over the full 360
degrees.)
107
Vibration Analysis on Rotating Equipment
Advanced functions
a a+b
ϖt
0 φ
b
The parallelogram rotates at the vibration frequency ϖt so the two vibrations must be at the same frequency
for this representation to work. For numerical addition, however the vectors must be resolved
geometrically which loses almost all of the advantage of ease. There exists a simpler method of handling
the vectors numerically by employing imaginary numbers. A complex number can represented graphically
by a point in a plane where the real numbers 1,2,3 etc. are plotted horizontally and the imaginary numbers
are plotted vertically. With the notation
j = √-1
these imaginary numbers are j, j2, j3, etc. Harmonic motions are represented by rotating vectors. A
substitution of the variable angle ϖt for the fixed angle of the vector (α) leads to
a(cos ϖt + j sin ϖt)
representing a rotating vector, the horizontal projection of which is the harmonic motion. But this
horizontal projection is also the real part of the vector. So if we say that a “vector represents harmonic
motion” what we really mean is that the horizontal projection of the rotating vector represent that motion.
Similarly if we say that “a complex number represents harmonic motion” we imply that the real part of such
a number, written in the form “a(cos ϖt + j sin ϖt)” represents that motion. Almost all of the algorithms in
the analyzer which involve phase make reference to the imaginary number. We do not necessarily need to
make much use of this number but we do need to know where it comes from. The first use that we will
make of the imaginary number is the Nyquist chart which is useful for identifying resonances.
108
Vibration Analysis on Rotating Equipment
There is quite a lot of confusion over the terms cascade and waterfall plots in the vibration world. A
cascade plot is a 3-D representation of the amplitude against speed with respect to time and the waterfall is
the 3-D representation of the amplitude against frequency with respect to time such as the example shown
above. The advantage of these plots is that we can record data over a certain period of time and see how all
of the relevant frequencies are affected. This could be a very high speed machine over a short time duration
or a slow speed machine over a very long time period.
Whichever is the particular application, the collection of data must be considered very carefully before data
acquisition takes place. The OR25 series analyzer is limited to displaying data at the acquired Fmax or
1/10th or 1/100th. Ensure that the data collection rate is valid for the necessary analysis.
Triggering
As mentioned previously a trigger may be set up on a machine to control the acquisition of data. This
trigger is often a 5 volt, once per rev pulse or it could be from an encoder giving many pulses per rev. A
many pulse per rev trigger requires an external clock input to the collector which is programmed to the
number of pulses per rev. A third type of trigger is a single pulse or voltage change and is known as an
event trigger.
Encoder trigger
The encoder will give many pulses per rev and must be input to the external clock input. This input is
necessary for torsional vibration and should ideally be used for time synchronous averaging where the
speed may be expected to change from moment to moment.
109
Vibration Analysis on Rotating Equipment
Event trigger
This is probably the simplest trigger and may be setup as a channel reading or an external trigger. The
trigger level is set and the analyzer just sits there waiting for the trigger amplitude to reach the pre-
defined level. The analyzer is actually collecting data all of the time but is only storing as much as the
memory will allow. When the trigger is activated the data storage will commence. If the trigger is set
up as a pre-trigger, then some data collected before the trigger event will be collected as well as some
data after the trigger. A post-trigger will collect data only after the event. This can be very useful if the
subject machine is suffering from a transient fault.
Bodé plots
When the cross channel properties are calculated by the analyzer, they can be displayed as amplitude or
phase against speed (Bodé) plots or real-imaginary (Nyquist) plots. The Bodé plot is useful when assessing
the resonant condition of a machine or even bearing deterioration as in the example below.
Orbits
Displacement sensors measure the amount of shaft displacement, or rather, the total motion of
the shaft's orbit within the bearing. The display is set so that the output of one transducer is set to
one (X) axis while the output of the other sensor is on the other (Y) axis. The displacement
sensors are mounted orthogonally which is so that each sensor is 45° from the vertical. In the first
example we see an elliptical orbit which is typical of preload caused by misalignment.
110
Vibration Analysis on Rotating Equipment
BALANCE
Diagnosis of a degrading balance condition is performed by
concentrating on the synchronous amplitude which coincides with the
rotor speed. This can be accomplished by viewing the spectra from
any single Eddy Probe sensor. A similar diagnosis can be made by
viewing the filtered signals from two orthogonally mounted Eddy
Probes sensors as orbits. As the balance condition deteriorates the
size, and sometimes the shape, of the orbit will grow larger until the
peak-to- peak amplitude exceeds acceptable limits.
CRACKED SHAFT
A crack in a rotor, or shaft, can generate several different effects on
how the machine behaves: a change in the vibration level, a change in
the operating phase angle, and/or a change in the resonance frequency
as the machine starts or stops. Spectral analysis can be used to identify
this fault, but observing filtered, synchronous orbits with the phase
angle superimposed on the orbit allows rapid identification of this
condition.
The "possibility" must be emphasized and carefully analyzed because many other causes can
produce these changes, such as, a damaged or loose bearing support, foundation problems, loose
rotating parts...basically anything that can influence the "system" mass, damping, and/or stiffness.
111
Vibration Analysis on Rotating Equipment
mass, such as an impeller, which has come loose; it can rotate freely on the shaft independently.
As the loose part rotates it influences the balance condition of the rotor which appears as a cyclical
increase and decrease in the synchronous amplitude. This is observable using a spectrum analyzer,
but the changes may be too rapid for the sampling rate of the instrument. An oscilloscope set up to
observe a filtered orbit will sample continuously so that the changes can be seen. The phase
shifting can, also, be observed using an oscilloscope.
The inception of a loose part condition will produce a "nervous" filtered, synchronous orbit. The
orbit will appear to vibrate slightly as this condition is created; the part may be slipping and then
sticking on the shaft just prior to becoming a full fledged loose rotating part.
OIL WHIRL
Oil whirl and oil whip are sometimes listed as a single machine fault,
but closer observation of the vibration signals and the machine
conditions causing these signals will produce different, distinct signal
displays for each condition. This fault is caused by a condition which
prevents the rotor from creating a stable oil wedge on which ride. An
improperly designed bearing is the usual source for oil whirl
conditions, but a change in the fluid viscosity or machine alignment
state are other possibilities.
Generally, an oil whirl condition precedes an oil whip condition. Spectral and orbit analysis can be
used to identify either condition. This phenomenon creates an individual sub-synchronous
frequency which can occur within a frequency range from 35% to 48% of rotor speed, depending
upon the machine/bearing design or construction. As the machine accelerates the whirl frequency
will increase as machine speed increases.
Observing oil whirl as a filtered, synchronous orbit produces a distinctive display. The orbit will be
more or less round in shape with an amplitude that nearly approximates the bearing clearance, and
when the phase angle is superimposed upon the display, the orbit will appear to have two phase
marks on it. This characteristic is due to filtering at shaft speed and the fault being generated at a
sub-synchronous frequency. The two phase marks will not be displayed symmetrically on the orbit
because the whirl frequency is not at exactly ½ machine speed.
OIL WHIP
Oil whip occurs during the later stages of an oil whirl condition and it
has a distinctive orbital display. The display, with the phase input
superimposed on the display, appears to have several phase marks.
This display will be round in shape and the amplitude will greater that
the amplitude noted during oil whirl.
The size of the orbit be will larger because the shaft uses up the entire
bearing clearance as an oil wedge can no longer be established by the
rotor and the shaft is in direct metal-to-metal contact with the bearing. The orbit display will no
longer rotate because the oil whirl frequency has coincided with the first natural resonance, or
critical speed, and has "locked" onto this frequency. Oil whip is a dangerous condition because the
rotor uses up the entire bearing clearance and is in direct metal-to-metal contact that will wear
away the bearing rapidly and destroy the rotor if not corrected.
EXCESSIVE PRELOAD
All journal bearing machines have some amount of preload so that a
stable oil wedge can be established. The preload may be internally or
112
Vibration Analysis on Rotating Equipment
externally produced. Internal sources of preloads are from gear meshing or hydraulic loading
during pumping actions. External preloads may be from coupling misalignment or piping and
support system thermal changes. These sources of preload create an elliptical orbit that is flattened
in the direction of the preload vector.
As the preload increases the orbit is further flattened. As excessive preload increases further the
orbit begins to collapse to form a "banana" shape as the shaft tries to continue its normal rotation
pattern and direction.
After the orbit has been flattened into the "banana" shape a 2X frequency is present on spectra
displays. Heavy preloads further distort the orbit into a figure eight shape. As preload increases the
shaft centreline will shift in the direction of the preload vector.
RUB
A common problem in newly rebuilt or modified rotors is a slight rubbing condition as the rotor is
initially operated. Rotor rubs are not a phenomenon which continues
over an extended period; they usually increase the clearances until the
rub has been cleared or, if not corrected,
they will wear away the internal
clearances until the machine cannot be
operated. The shape of the orbit display
will differ depending upon the
relationship of the shaft speed to the first
natural frequency and the severity of the rub.
The severity of the rub will affect the shape of the orbit. A light rub will produce a "tear drop"
shaped orbit, with the point of the tear drop coinciding with the impact spot. As the rub gets
heavier the orbit will flattened and may appear as an excessive preload.
At higher machine speeds (above twice the first natural frequency) the unfiltered orbits will begin
to have internal loops with the fundamental rub frequency inversely proportional to the number of
internal loops. These internal loops will have their own phase marks displayed and the loops will
be located symmetrically on the display.
113
Vibration Analysis on Rotating Equipment
Introduction to resonance
The majority of vibration analysis programs concentrate on detecting dynamic machinery faults such as:
• Bearing defects
• Gearing defects
• Misalignment
• Out-of-balance.
Eventually, however, a vibration is encountered which does not fall into any of the above categories and the
vibration analyst is left scratching his head. He knows the complete structure of the machine. He knows the
bearing numbers, the speeds, the number of teeth on the gears, the belt and pulley dimensions and still none
of the pre-calculated defect frequencies line up with the big spike in the frequency spectrum. In many cases
the big spike is related to a resonance of a natural frequency or perhaps running a machine near a critical
speed.
Many consultants in the field of vibration analysis will tell you that 90% of all failures are directly caused
by vibration resonance. This is not the case. Most failures are due to material fatigue from normal
operation leading to mechanical failure. The vibration analyst running the average predictive maintenance
system knows that most of his machinery failures are from the “bread and butter” causes listed above. The
vibration consultant, on the other hand, is usually only called in when the in-house analyst is having trouble
identifying a troublesome vibration and these problems usually are caused by resonance.
The purpose of this course is to help the in-house vibration analyst identify and cure vibration resonance
problems without having to call in outside assistance.
By the end of this course you will be able to:
• use several techniques to identify resonant conditions
• recommend appropriate measures to cure the resonance
114
Vibration Analysis on Rotating Equipment
What is resonance?
Imagine a child on a swing. You are the adult who pulls the child back and releases the swing so that the
child swings back and forth. The rate at which the child swings is the NATURAL FREQUENCY of the
child/swing system. Every time that the child swings back to your position, you give the swing a small push
– you are the EXTERNAL EXCITING FORCE. You time the small pushes to coincide with the rate of
swing of the child – in other words the exciting force is at the same frequency as the natural frequency of
the system. Each small push (as long as it is large enough to overcome friction or damping) will add to the
amplitude of the swing. In this case the child/swing system is in RESONANCE with the exciting force.
Imagine what would happen if you pushed at a different frequency. Sometimes the pushes would add to the
amplitude of the swing but sometimes the pushes would reduce the amplitude of the swing. In this case the
system is not in resonance with the exciting force.
In the world of maintenance the exciting forces are from a myriad of sources with the strongest usually
being related to run speed. Such forces include imbalance, misalignment, gear mesh, bearing frequencies
etc. In fact, anything which can be measured as a vibration may be considered to be an exciting force.
The problem we have is that we can easily determine the exciting force frequencies but cannot always tell
which system has it’s natural frequency being excited. When we say a system, we mean any part or
component (or combination of either) of the machine and/or the structure. Looking at the average machine
and associated structure, we can see a host of interconnectivity which can give rise to a large number of
natural frequencies. Fortunately for us, even though there may many natural frequencies, most of the
systems have inherent damping. The damping acts like the friction in the child/swing system. If the
exciting force is not large enough to overcome the damping, then the amplitude at that frequency will
diminish. If, however, there is little damping then the interaction of an exciting force with a natural
frequency can give rise to very large amplitudes of vibration resulting in heavy load cycling and eventual
failure of the material.
Natural Frequency
The natural frequency of a system is that frequency at which the system will oscillate or vibrate when
excited by a single external exciting pulse. Any oscillating object has a natural frequency, which is the
frequency an oscillating object tends to settle into if it is not disturbed.
Mechanical
For example, the natural frequency of a pendulum 1 m (39 in) long is 0.5 Hz, which means the pendulum
swings back and forth once every 2 seconds. If the pendulum is struck lightly once every 2 seconds, the
amplitude of the swing increases gradually until the amplitude of oscillation is very large. The phenomenon
in which a relatively small, repeatedly applied force causes the amplitude of an oscillating system to
become very large is called resonance.
Let us consider the pendulum mentioned above. The natural period of oscillation of a pendulum is
calculated using the following formula:
115
Vibration Analysis on Rotating Equipment
π
2
L 1
4. . dφ =
g 2
x . sin( φ )
2
1
0
Notice that the time period of the swing is affected slightly by the angle of the pendulum but the effect is
only slight. The main influences on the natural time period are the length of the pendulum and gravity. To
calculate the natural frequency from the time period we simply divide the time period into one second.
In this case the time period is:
1
f Hz
Ts Equation 2 Calculating frequency from time period
1
=
Ts
In a conventional mechanical system, such as the one shown below, we have to consider all aspects of the
system that affect the natural frequency.
116
Vibration Analysis on Rotating Equipment
• Mass
• (Spring) Stiffness
• Damping
Let us look at the spring/mass system in a little more detail.
Think of a spring with a mass of a certain weight hanging off the end. If we pull the mass down and then
release it, the spring/mass system will move vertically up and down at a rate which is determined by the
spring rate, the weight and gravity.
Newton's second law states that:
Force = mass x acceleration
Equation 3 Newton's 2nd law
So at any instant we can determine the acceleration of the mass, and by applying Newton's laws of motion
we can determine the instantaneous velocity and displacement.
1.
u .t
2
s (a t) Equation 4 Newton's law of motion
2
Where s = displacement (x)
u = velocity (x2)
a = acceleration (x3)
t = time
In the single degree of freedom system such as the mass/spring system shown, if the spring force has
magnitude kx, and the extension "x" is zero then there is no force.
The sign of the spring force is negative as it is always acting in the opposite direction to the extension. So:
Spring force = "-kx"
The damping force is also negative, being "-cx2", because the damping is always in the opposite direction
to the velocity.
The forcing mechanism is defined harmonically as Po sin ωt, and this could be as simple as someone
pushing and pulling the mass with their hand.
Note: Po sin ωt is a combination of direct force "Po" and angular position "sin ωt"
where ω = rotational velocity &
t = time
117
Vibration Analysis on Rotating Equipment
or
mx3 + cx2 + kx = Po sin ωt
This equation is known as the "differential equation of motion of a single-degree-of-freedom system". You
can easily see the three components of mass, damping and stiffness directly affect the amplitudes of
oscillation in, acceleration velocity and displacement respectively.
In order to have a resonant condition we need to have the right combination of stiffness and mass to create
the natural frequency and a source vibration that matches the natural frequency. The damping controls the
amplitude response. The damping qualities of a structure control the resonant response. The higher the
damping effect the lower the amplitude of vibration will be as a result of a resonant condition. For example
a spring is a good example of a low damped system, a relatively small amplitude of input results in a large
amplitude response. A shock absorber is a good example of a highly damped system where a large input
results in a small response.
Pumping systems in general and hydraulic systems in particular are often associated with hydraulic
resonances. There are two different types of vibration in liquids:
1. Sonic vibration and
2. Pulse vibration
Sonic vibration
To think of a hydraulic resonance, consider a pipe filled with liquid. If we impact one end of the pipe, a
vibration will travel through the liquid in the pipe and the speed at which the vibration travels will be at the
speed of sound. Sound travels at different speeds in different liquids but in fresh water it travels at roughly
five times the speed of sound in air.
118
Vibration Analysis on Rotating Equipment
Mercury 1440
air (reference) 320
liquid helium 240
The sound waves created in a substance travel with a certain speed through the substance. The speed with
which the sound wave or sound energy travels depends on the interaction between neighboring molecules of
the substance. The speed of sound decreases as the mass of the molecules increases because the mass of the
molecules is related to the force per acceleration of the molecule. However, the speed of sound increases as
the strength of the interaction between the molecules increases.
The exact relation between the velocity of sound, V, and the properties of the substance through which the
sound travels is given by:
V = √(elastic property/inertial property).
Equation 6 Velocity of sound in materials
Now, if we know how fast the impulses travel through the liquid, we can determine the resonant frequency
of oscillation for a certain length of pipe:
c = f ⋅λ
Equation 8 General formula relating speed, wavelength & frequency
where c = speed of sound in the material
f = resonant frequency
λ = length of pipe (or direct sub-multiples)
If this frequency coincides with a run speed or vane pass frequency then a serious resonant situation could
occur.
119
Vibration Analysis on Rotating Equipment
Impulsive vibration
The other type of vibration in liquids is from the pulses generated by the system itself, for example:
• Vane pass
• Piston pulse
• Hydraulic valve operation
• Impeller eccentricity
As each hydraulic pulse enters the piping system, it pushes liquid in front of it because most liquids are
virtually incompressible. The shock wave from this pulse then travels right through the system giving
system-wide excitation at that impulsive frequency.
It has been the authors experience that heavy vibrations on pumping systems are usually caused by an inter-
reaction of these impulsive and sonic vibrations. The usual outcome is a very noisy operating environment,
but mechanical failure often results. The impulsive frequency can also excite any mechanical resonances in
the system from pipework or supporting structures, particularly if they are both at the same frequencies.
Before we get too deep into the vibratory aspects of air-borne sound for vibration analysis, let us look at the
historical use of air resonance – music. The subtle timbre of a particular instrument is caused by the
particular frequency in resonance and several harmonics and/or sub-harmonics of that frequency.
In musical instruments harmonics are a series of subsidiary vibrations that accompany a primary, or
fundamental, wave-motion vibration. They result when the vibrating body, for example, a stretched string
or an enclosed air column, vibrates simultaneously as a whole and in equal parts (halves, thirds, quarters,
and so on), producing wave frequencies that are in simple ratios with the fundamental frequency (2:1, 3:1,
4:1, and so on). In musical sound the full-length vibration produces the fundamental tone (or first harmonic
or first partial), which is usually perceived as the basic pitch of the musical sound. The subsidiary
vibrations produce faint overtones (second and higher harmonics or partials). As the series progresses, the
vibrating segments become smaller, the frequencies higher, and the musical pitches closer together. The
harmonic series for the tone C is given in the accompanying notation; black notes show pitches that do not
correspond exactly with the Western tuning system. Harmonics contribute to the ear's perception of the
quality, or timbre, of a sound: On a flute, certain harmonics of the series are most prominent; on a clarinet,
others.
In “over-blowing” a wind instrument, the player isolates and makes predominant one of the higher
harmonics, thus extending the range of the instrument upward. Unvalved brass instruments, such as the
bugle, produce only the tones of the harmonic series; the valves of the trumpet and the slide of a trombone
add extra tubing, giving the player a new fundamental wavelength with a new harmonic series; the
instrument can thus produce more tones. The harmonics of string players are flutelike tones produced when
they cause the string to vibrate solely in halves (or thirds, and so on).
As in liquids (described above), to calculate the resonant frequency in a pipe, for example, we must first
know the speed of sound in air.
120
Vibration Analysis on Rotating Equipment
where P =pressure
ρ = density,
and γ = ratio of the specific heat of the gas at a constant pressure over the specific heat at a constant
volume.
γ = cp over cv
Equation 10 Ratio of specific heats (gamma)
≈ 1.67 for monatomic gases,
1.40 for diatomic gases,
1.30 for triatomic gases,
and in the range of 1.2 to 1.1 for polyatomic gases.
Applying the Ideal Gas Law, the equation for the speed of sound in a gas becomes
v = √ (( γ k T) / m) = √ (( γ R T) / M)
Equation 11 Speed of sound in a gas
121
Vibration Analysis on Rotating Equipment
The antinodes (A) are displacement antinodes (pressure nodes) and the nodes (N) are displacement nodes
(pressure antinodes). If the frequency is fixed, as in the case of the tuning fork held over the open end of a
tube, the resonance lengths are v/4f, 3v/4f, 5v/4f and so on for a tube closed at one end, and v/2f, 2v/2f,
3v/2f for a tube open at both ends.
When resonance exists, the displacement is a minimum (node) at the closed end, but the antinode is not
exactly at the open end. It is actually a small distance beyond it. This extra distance beyond the end of the
tube is called the end correction . The acoustic length of the tube is equal to its physical length plus the end
correction.
In a typical laboratory experiment with a water reservoir-resonance tube apparatus about 1m long, and
tuning forks of frequency about 500 Hz, three positions of resonance can usually be found. If the positions
of resonance from the open end of the tube are L1, L2, and L3, the wavelength of the sound wave is equal
to 4(L1 + e) or 4/3(L2 + e) or 4/5(L3 + e), where e is the end correction. From these relations the
wavelength may be calculated from 2(L2 - L1) or 2(L3 - L2) or (L3 - L1), and the end correction may be
calculated from 1/2(L2 - 3L1) or 1/2(3L3 - 5L2) or 1/4(L3 - 5L1). Since the frequency of the tuning fork is
known, both the velocity of sound may be determined by using the relation c = f⋅⋅λ, and the value of the end
correction may be determined from the relations above.
Karman Vortices
Everyone knows the story of the Tacoma Narrows suspension bridge. The bridge began to sway violently
when the wind blew steadily at a certain velocity. The mechanism for the resonance-destruction of the
bridge was the same mechanism that causes a flag to flutter – Karman Vortices.
122
Vibration Analysis on Rotating Equipment
As the air passes over the object a vortex will form on one side of the object and down wind. The increased
pressure will cause a deflection of the air flow so that another vortex is formed on the other side of the
object just as the first vortex dissipates. The vortices are alternately clockwise and anti-clockwise and are
shed from the object in a regular manner resulting in an alternating force on the object, perpendicular to the
air flow. There is a distinct relationship between the frequency (f) of the force, the velocity of the wind (V)
and the width or diameter (D) of the object.
f * D = 0.22
V
Equation 12 Karman vortices
In the case of the Tacoma Narrows bridge, a steady wind blew along the valley onto the side of the bridge.
As the wind hit the side structure of the bridge it developed Karman vortices. The vortices for the wind at
this speed developed an alternating force which was exactly at one of the natural modes for the complex
structure of the bridge.
123
Vibration Analysis on Rotating Equipment
Rotor balance resonance speed is the shaft rotational speed (or speed range) which is equal to a lateral
natural frequency of the rotor system., When the speed increases or decreases, the observed vibration
characteristics due to rotor unbalance are:
1) a peak in the 1X amplitude and
2) a rapid change in the 1X vibration phase angle.
The resonance speed is at the point which is 90° phase away from the angle of the heavy spot for that
resonance mode. This may be slightly different than the rotational speed peak amplitude point. These
changes may not happen at the same frequency due to non-linearity, damping, and/or asymmetry in the
system stiffness.
Campbell Diagram.
A diagram used in rotating machinery design. A tool for selecting and checking shaft operational rotational
speeds and other possible forcing function frequencies against the spectrum of natural frequencies to avoid
resonances. The X axis represents the various possible excitation frequencies, i.e., rotational speed (1X),
oil whirl (.40X to .48X), blade or vane passing frequencies, gear mesh frequencies, etc. The Y axis
represents the rotor lateral and torsional natural frequencies. The term is sometimes used incorrectly to
describe the cascade plot and waterfall plot.
124
Vibration Analysis on Rotating Equipment
125
Vibration Analysis on Rotating Equipment
Whenever cracking of the machine or structure is reported, resonance should be suspected. Cracking almost
always occurs at one of the mode shape nodes. The material does not break due to excessive stresses but
due to the many millions of reversed stresses concentrated about the nodes, causing fatigue. The
phenomenon is similar to bending a wire back and forth until it breaks. The break has the characteristics of
a pure fatigue failure which is often mistakenly called “crystallization.” The break usually shows a
crystalline appearance and sharp edges.
This could account for so many familiar situations whereby pump shafts “mysteriously” break within a few
weeks to a few years, even though the calculated stresses indicate the shaft should last a lifetime without
breaking. Premature foundation or floor cracking, pipe weld breakage (often repetitively), gear cracking or
failure (usually in equal-spaced, radial lines), gear driven shafts broken with the characteristic 45° torsion
break, have all occurred as the result of metal or concrete fatigue at one of the nodes due to resonance.
Sometimes rotating machinery does not show excessive vibration at the usual points of measurement, such
as at the bearings. However, by feeling with fingers along the small diameter, lubrication, cooling water or
instrumentation piping, it can be determined if there are tell tale large “loops” (antinodes) and nodes,
characteristic of a resonant condition. Most often the vibration frequency is high enough and the tube or
pipe rigidity low enough to allow a higher resonance with several nodes. While it is unlikely that the
vibration on the tubing will cause vibration on the machine the problem should still be addressed to prevent
failure of the tube. It may take many months or even years for the tube or pipe to crack at one of the nodes,
but when it does, lost lubricant can wreck an expensive machine or a flammable product can be released.
The same phenomenon is very common for pressure gauges cantilevered on a pipe. If non-resonant, there is
no problem. But sometimes the pipe, with a gauge acting as a weight, is resonant to a specific frequency of
machine vibration. Eventually it will break, usually at its point of connection.
When hand feeling a component or structure reveals a possible resonant situation further confirmation
should be obtained by plotting out the mode shape on graph paper. This simple procedure can be performed
with the most rudimentary of instruments and is very effective in identifying the mode shapes. Remember to
126
Vibration Analysis on Rotating Equipment
plot both directions (horizontal and vertical). On piping sections the direction indicating the greatest “curl”
is the resonant direction.
The procedure for performing the mode shape plot is as follows:
1. Divide the component to be tested into a number of roughly equal parts and identify each location,
usually with a letter to avoid confusion.
2. Measure the vibration amplitude at the frequency of interest at each location. Displacement units are
preferred but velocity or acceleration can also be used. Overall readings can be used but filtered data is
preferred.
3. Divide each reading by 2 (for displacement).
4. On a horizontal zero line drawn on a sheet of graph paper mark the measurement locations spaced 1” or
25mm apart.
5. Plot the results above and below a horizontal zero line on a sheet of graph paper using a scale of ½ “ = 1
mil of vibration; 1 cm = 25 microns.
6. Join the dots “curve fit” if required.
Phase
If a force is slowly applied to a spring system, the force and its resulting deflection will move in the same
direction at the same time, or “in phase” with each other. If a vibrating force is applied to the same object or
“spring system,” and if the frequency of that vibrating force is below the critical speed range or resonance
range, then the force and resulting deflection will remain in phase. However, when the vibration frequency
enters the resonance frequency or the critical speed range, the deflection will time-lag the force. When the
vibration frequency reaches the actual critical speed or resonance frequency, the force will precede the
deflection by 90° . As the frequency increases further, the force increases its lead until the force finally
precedes the resultant deflection by 180° . The 180° relationship will remain the same for all frequencies
above the resonant frequency range, until the second resonance frequency range is approached. Then the
process starts all over again with another phase angle reversal, and so on. This phenomenon is shown in the
127
Vibration Analysis on Rotating Equipment
resonance diagram of a rotor’s vibration amplitude and phase relative to resonance of some part, such as the
rotor support system, part of a steel base or beam, or more rarely, resonance of the rotor itself.
128
Vibration Analysis on Rotating Equipment
Another simple yet effective way to confirm whether or not a part is vibrating in resonance is the “bump”
test. With the machine shut down and a vibration pickup held or attached to the machine, simply bump the
machine or structure with a force sufficient to cause it to vibrate. Since an object will undergo free
vibration at its natural frequency when bumped or struck, the natural frequency generated in this way will be
indicated on the analyzer. If the vibration diminishes very quickly it may be necessary to bump the machine
several times in succession in order to sustain free vibration long enough to register on the analyzer
frequency meter. Although care should be taken that the frequency at which the "bumps” are struck is not
read as a resonant frequency. The Real Time Spectrum analyzer providing instantaneous display of the
vibration amplitude versus frequency data is an ideal instrument for determining a natural frequencies this
way.
One method of determining the natural frequency of a structure is to calculate it. The calculations are fairly
straightforward for simple structures. For the more complex structures found in the plant the calculations
are far more difficult and therefore not usually as accurate as those obtained through simple bump or impact
tests. Bump tests, however, are almost always limited to determining the first critical frequency of the part
bumped. Also, if the part tested is connected to another part that is relatively flexible, then it is sometimes
difficult to be sure that the resulting vibrations are originating from the part being bumped rather than from
the part to which it is connected.
129
Vibration Analysis on Rotating Equipment
Bump tests are also ideal for rotors, rolls, fan blades, beams, columns, floors, covers, pedestals and welded
steel bases. They are also practical for relatively rigid parts, including cast iron or cast steel bearing
supports and brackets.
This test is based on the principle that when a “spring system” or part is bumped or deflected in some way,
it will vibrate for several cycles at its natural frequency. Its amplitude decreases with each cycle, but its
frequency remains the same. The low frequency of the repeated bumps is not measured, only the frequency
that is generated between bumps.
130
Vibration Analysis on Rotating Equipment
3) Set the instrument up for manual amplitude scaling (disable the autorange feature) and select an
appropriate amplitude scale. Start with a sensitive scale and adjust the scale as required to suit the
strength of the impact and the size of the structure. If autorange remains active the instrument will tend
to range to the initial bump rather than the “ring down” afterwards. The amplitude scale and strength
of impact should be set to respond to the “ring down” rather than the impact.
4) Set the resolution to an appropriate number such as 400. The more the lines of resolution, the longer it
takes for the instrument to acquire the data. Most modern data collectors have the capability to generate
a “live” time display. If this is used, then a setting of 100 lines may be more appropriate. Despite the
loss of accuracy due to the lower resolution, the results are usually acceptable for most situations. Take
care that the data collection rate will be high enough to differentiate between the oscillations in the
ring-down and not cause them to be smeared.
5) If the analyzer has the capability to display continuously updated (running) spectra using “peak-hold-
type” averaging, then its use is recommended for the best results. If “peak-hold” is not available, then
the number of averages should be minimized, such as 4 to prevent excessive data collection time.
Special windows and trigger settings are available on most data collectors today. They can be used but
are not necessary for this simplified test.
6) Take a spectrum and observe the background vibration data (if any). These peaks, representing
vibration coming from other machines will have to be mentally subtracted from the bump test results.
Some condition monitoring software programs will subtract spectra as well. If you have an analyzer
which can perform negative averaging, then use this capability to remove the background vibration
after the bump test.
7) The structure should now be impacted in the direction of transducer orientation using a soft hammer or
piece of wood. A single blow should cause a response on the instrument and the strength of the blow
should be adjusted to give an appropriate response on the instrument. If multiple blows are required
then random impacts are preferred. For expected resonant frequencies in the range of normal speed
machines, such as under 5000 rpm, bumping with a rubber mallet is preferred over harder materials
such as steel or plastic. Soft wood (such as a “2x4”) can also be used. For higher expected
frequencies, harder hammer heads such as plastic are usually recommended.
A typical spectrum obtained from a bump test will look similar to regular spectra obtained from vibrating
running machines, except that the various peaks will not necessarily be related to running speed. Instead,
the peaks will relate to the resonant frequencies of various machine parts that were displaced by the
bumping action. Typically, there is one peak that has a much higher amplitude than the rest. The largest
peak usually represents the resonant frequency of the part that receives the bump test hammer’s impact,
although not necessarily in all instances. When the hammer impacts one part, it causes a deflection and
release. However, the impact also causes some deflection and release on parts that are connected to the part
receiving the hammer blow. If the other part is more flexible or has much less damping, then that part may
cause the largest amplitude peak. A single bump test, using an FFT, may not always determine with
certainty which peak represents a specific part. If multiple peaks appear then multiple tests should be
performed on adjacent structures and in different directions in order to determine which peak relates to
which structure.
Impact hammer
A better way of determining natural frequencies is to use the impact hammer. This is simply a hammer with
a force transducer mounted which will send a trigger signal to start collecting data the instant the hammer
hits the structure. An accelerometer is mounted on the structure and the resultant output is generated by the
natural frequencies of the structure.
131
Vibration Analysis on Rotating Equipment
For specific information on the use of the impact hammer refer to the hammer operating manual. For this
course we will use medium sledge from IMI, model 086C41.
132
Vibration Analysis on Rotating Equipment
Care should be taken when defining the spectrum that sufficient resolution is used to stop the oscillations
from smearing into one another. If the sampling rate is not high enough to separate out the individual
oscillations then the time waveform will take on the form of a “pulse”. Consider table 2, above: Note that
if the individual oscillations are smeared into one pulse in the time domain, then the FFT cannot have
relevant frequency domain data. In some cases it is possible to get a quasi-triangular pulse which, when
133
Vibration Analysis on Rotating Equipment
passed through the Fourier transform, will give a frequency response which looks good but is, in fact, totally
meaningless.
134
Vibration Analysis on Rotating Equipment
Uniform Beams
One of the most common sources of resonant vibration is the structural supporting beam.
Table 3 Natural frequency calculation of uniform beams
135
Vibration Analysis on Rotating Equipment
The notation a1 through a5 is for each vibration mode. For example, calculating a1 for a clamped-clamped
beam will return the natural frequency of the first vibration mode, calculating a2 will return the frequency
for the second mode etc.
I
ωn a 1 . E. Equation 13 General formula - natural frequency of beams
µ .l
4
The difficulty in making this type of calculation is getting the values for E, I and µ.. Luckily, however,
standard text books such as the “Machinery’s Handbook” have all of the required information for different
types of beams.
• For low carbon steels E (Young’s Modulus) can be estimated at 30,000,000 psi.
• The weight per unit length is a part of the beam identifier. For example a wide-flange section beam
designated “W 18 x 64” has a nominal depth of 18” and nominal weight of 64 pounds per foot length.
This is µ in our calculation above.
• The moment of inertia I may be calculated but it is far easier to lift the value from tables as in the
example below:
Table 4 Standard values for uniform beams
136
Vibration Analysis on Rotating Equipment
Notice that there are two values if I. There is a value for each of the two non-axial directions (vertical and
horizontal) so we can calculate the natural frequencies for each of these directions. The way that the
example above is drawn, the section X-X will give us the moment of inertia horizontally and the section Y-
Y will give the moment of inertia vertically.
Let us calculate the natural frequency for a low carbon steel, “clamped-clamped” beam, 12 feet long,
designated W18x106.
I
ω n a 1 . E.
µ .l
4
E = 30,000,000
I (section Y-Y) = 220 in4
µ = 106 lb./ft = 8.833 lb. / in
a1 = 22.0 (from table 2)
l = 12 feet = 144 inches
Notice that we have been consistent with our units – we have only used pounds and inches in the equation.
Whenever you are given properties in other units, you must convert to the units for which your formula is
designed.
Plates
The procedure for calculating the natural frequency of any component is similar to that of uniform beams
except that the formula varies from case to case. For plates we should consider
• circular or rectangular plates
• point load or uniform load
• simply supported edges or clamped edges
In 7Hartog’s “Mechanical Vibrations” we have a number of formulae for different applications in the
appendix.
For a square plate, all edges clamped, length of side t, the fundamental mode is
D
36.
µ1
ωn
2
l
Where µ1 = mass per unit length
And D = the plate constant
7
Available in Borders bookstore. $12.95. (DOVER 0-486-64785-4)
137
Vibration Analysis on Rotating Equipment
3
t
D E.
2
12 1 µ
Where t = thickness (inches)
µ = Poisson’s ratio (≈ 0.3)
For a low carbon, steel plate, ¼” thick, the plate constant D will be 4.293 x 104
So that for a square plate, clamped at the edges, of side 6 feet we would have
ωn = 0.108 Hz
(Don’t forget to divide radians/second by 2π to convert to Hz)
For a plate of equal side = 3 feet
ωn = 0.611 Hz
If the same 3 foot plate were ½” thick:
ωn =1.221 Hz
If your plate is clamped at each end but free on the other two sides, then treat the plate as a
clamped/clamped beam.
138
Vibration Analysis on Rotating Equipment
Vibration isolators
This section explores some of the isolation materials that we can use.
The following 8table suggests some appropriate damping ranges for popular vibration isolation materials:
It should be noted that these ranges are for virgin materials and many manufacturers offer specialist
materials which cover much wider ranges than shown on this list. Vibration isolation is based on installing
machinery on springs or resilient material of uniform stiffness. The types of spring material which are used
most often are rubber and steel. Another alternative is air springs.
Cork was formerly used as a "spring" material and its elastic properties were based on compression and
expansion of the air locked inside the cork. However, the locked-in air is eventually pressed out, thus
impairing the isolation result and causing it to have no effect at all after a period of use. Cork does not
permit a large spring travel, which is why it is no longer used in more demanding applications. The same
applies to felt, which has more or less the same kind of properties.
Steel springs are normally used in the form of coil springs or leaf springs. The benefit of these is that they
permit relatively high deflections, but their disadvantage is that they provide very little damping. Because of
this, excessive movement occurs when running the machine through the resonance range, and more often
than not special devices have to be installed in order to limit the deflections.
8
Taken from information prepared by Larry Riley (3/6/96)
139
Vibration Analysis on Rotating Equipment
Compared with steel, rubber has high internal damping, it also absorbs noise, has good elasticity and
chemical resistance; and these are properties which make rubber particularly suitable as a spring material.
The stress-strain diagram illustrates clearly the difference in the strain properties of rubber and steel. Figure
1 illustrates in diagram form a tensile test on steel, and figure 2 illustrates the corresponding test on rubber.
The stress strain curve for steel is linear up to the proportionality limit, and in this range Hooke's law
applies. This states that stress is proportional to strain, and the proportionality constant is equal to the
modulus of elasticity E. Steel is elastic up to the elastic limit and restores to its original shape (this value is
less than 0.1 %). At the tensile yield limit, or yield point, which is just above the proportionality limit, the
material is extended without the tensile force increasing. The maximum tensile force is reached at the
ultimate strength.
The tensile sequence for rubber is quite different. Rubber achieves high strain levels even at very low
tensile stress. The increase in load is neither linear; in other words, rubber does not "stretch" proportionally
to the load. Consequently, the modulus of elasticity is not constant, and Hooke's law does not apply in this
case apart from approximately in a small strain range of the magnitude of 5-10 %. There is no yield point,
but the stress increases continuously until there is abrupt failure.
Springs
Springs have long been a favorite with fan manufacturers for separating fan vibration from the floor.
The principle relating to vibration-isolating with springs is that they are placed between the machine and the
base or plinth. To ensure effective isolation, the springs must be calculated very accurately, otherwise the
result could be impaired performance. In favorable cases, the force can be reduced to only 2 or 3% of the
force of a rigidly mounted machine. In such cases, you could say that the vibrations are practically
140
Vibration Analysis on Rotating Equipment
eliminated. The relative magnitude of the transmission of force depends entirely on the tuning ratio Z. If Z
is high, the force transmission percentage will be small.
The magnification factor B, that is, the change in force acting on the base of a machine supported on springs
compared with a machine which is rigidly mounted, conforms to the formula:
B=
Equation 14 Amplitude magnification due to springs
Where D = damping
Z = ratio of actual frequency to resonance frequency
B = magnification factor
The factor D depends on the internal damping of the spring material. In rubber D has the value 0.04 - 0.1,
depending on the hardness of the rubber. The term “4D2 Z2” can generally be neglected completely except
in the resonance range, that is, when Z=1. If Z=1, that is, the machine speed (rpm) = the natural vibrations
of the system, it is said that there is resonance, and the vibrations will be infinitely large if there is no
damping. Here, then, a rubber spring has a distinct advantage over a steel spring, which has minor internal
damping and in which the amplitude, in theory, grows to a very high value in the resonance point.
As can be seen in figure 10, B (at Z=D√2) has dropped to 100% and when Z is further increased, B drops
rapidly. Vibration isolation is therefore of significance first when the operating frequency considerably
exceeds the natural frequency. For practical applications, Z should be between 3 and 5, which means that 88
and 96 % of interference forces are eliminated.
As a rule you usually know the operating speed of a machine (interference frequency). If the system's
natural vibration coefficient can be altered in any way, and thereby influence Z, then it is possible to change
the force transmitted. This is exactly what happens when vibration isolation is achieved i.e. by means of
springs. The low elasticity and shear moduli of rubber are used to achieve a low natural frequency.
To sum up, transmission of vibration forces can be affected in three ways:
1) Rigidly mounted machines transmit vibration forces in unchanged form to the base, which is therefore
forced to be a part of the movement of the machine. The magnification factor can be regarded as being
100%.
2) In the case of an unsuitable spring system, the magnification factor can increase considerably and may
amount to several hundred per cent.
3) The force transmission percentage is reduced substantially by correct calculation and suitable
mountings being installed between the machine and base. Typical reductions can be from 100 down to
10 %, but in favorable circumstances can be as low as 2 %.
All machines have more than one resonance point as, through many interacting movements, they can vibrate
in many different ways. The resonance points can be determined, but the methods of calculation are often
141
Vibration Analysis on Rotating Equipment
very difficult. Experience has shown that all resonance frequencies that can arise do not need to be
clarified. It is usually quite sufficient to calculate the most significant ones which can be determined easily.
The desired level of isolation and the interference frequency determine where the resonance frequency shall
be.
To calculate the natural frequency of a spring:
Equation 15 Natural frequency of a spring
where G = torsional modulus (typically 11,200,000 psi for ASTM A229 oil tempered spring steel)
d = spring wire diameter
n = number of turns
D = coil diameter
Let us calculate the response of a motor/fan system sitting on a frame which, in turn, is supported by by four
springs. To make the calculation easy we will assume that the weight of the unit is equally distributed over
the four springs.
Calculate the natural frequencies of the unit using the following assumptions:
Weight of motor and fan = 500 lbs.
Spring coil diameter = 1”
Wire diameter = 0.192”
Number of turns (spring) = 6
so:
k = (11,200,000 * 0.1924)/(8*6*12)
= 317.089 lbs./inch
ωn = (317.089*4/500)0.5/2π
= 0.2535 Hz
Rubber
Compared with other engineering materials, rubber is very ductile. In some cases, the strain may be higher
than 1000 %, and by far the highest proportion of this strain is elastic. Metals, on the other hand, have very
small strains below the elastic limit. Compared with metals, the tensile strength of rubber is rather low,
however. The maximum level that can be achieved with rubber is 25-30 MPa. However, because of the
high strain, rubber has a very large work absorption capacity compared with the best grade of steel. If a
material is subjected to a load below the elastic limit, the deformation will, according to Hooke's law, be
proportional to the load. This does not apply to rubber under tension or compression. This means that
rubber does not have any constant tensile or compression module of elasticity. Metals will normally be
softer towards the end of a tensile test, while the opposite is often the case with rubber. Rubber does not
have any yield point, and the modulus is increased until there is abrupt failure.
142
Vibration Analysis on Rotating Equipment
High elastic ductility is therefore the most pronounced feature of rubber. Just how easy it is to deform
rubber is shown by the fact that the modulus of elasticity of compression for rubber within the normal
hardness range is between 2 and 12 MPa; while the modulus of elasticity of steel is 210,000 MPa. This
means that soft rubber is about 100 000 times softer than steel.
Damping capacity is another important feature of compounded rubber. This is of particular importance
when operating a machine that is supported on springs through the resonance range. The resonance
deflection with rubber springs is only 1/5 to 1/50 compared with the deflection when using steel springs
with the same stiffness, see figure 17.
With a spring made of natural rubber working under compression or shear load, the direct loss of energy is
between 6 and 30% depending on the hardness of the rubber. This energy loss is such that it is possible in
many cases to use rubber springs as dampers. But care must be taken when it comes to damping in a rubber
element. If the element works with high amplitudes, a substantial amount of energy is converted into heat,
and the heat which is generated may lead to the rubber element being destroyed.
In the case of single impact, the vibrations sequence will be as shown in figure 19. The left-hand curve
represents a steel spring, while the right-hand curve represents a rubber spring. These two curves clearly
show just how quickly the vibrations degenerate in the rubber, while in steel springs they diminish slowly.
As sound-insulating material, rubber is one of the very best. The effect of sound isolation increases with the
thickness of the rubber. Rubber is an excellent absorber of impact sound, which occurs in foundations,
floors, buildings, etc.
High chemical resistance is yet another valuable feature when using rubber. Steel, on the other hand, can
rust easily when exposed to air and acids leading to fatigue problems. Rubber is fully resistant to moisture
and common acids, and at normal temperatures, does not give rise to ageing problems. The highest
temperature to which natural rubber should be continuously exposed is about +75°C (167°F). At extremely
143
Vibration Analysis on Rotating Equipment
low temperatures, -30°C (-22°F) to -40°C (-40°F), rubber stiffens and becomes rigid. At even lower
temperatures it becomes hard, brittle and non elastic (though this is reversible).
Rubber mountings may sometimes have to be installed at places where they can be exposed to the risk of
coming into contact with oil. The harmful effects of oil can often be avoided through the design of the
mounting or by fitting a mechanical shield. An alternative would be to use oil-resistant rubber material, but
compared to natural rubber it has inferior damping properties.
When calculating compression characteristics of rubber, it should be noted that the deflection is not directly
proportional to the load, as the modulus of elasticity in compression increases with the degree of stress.
The modulus of shear, however, remains constant for normal stresses. The factor with the most effect on
stiffness calculation is the ratio between loaded and free surface area of rubber - this is the so-called shape
factor (often designated S). With thin rubber sections, a very high modulus of elasticity can be achieved.
Equation 16 Modulus of elasticity for rubber
E8 = E0 (1+2.const.S2)
In other respects, the stiffness of a rubber spring is determined by the dimensions and the hardness of the
rubber.
Figure 20 illustrates the relationship between rubber hardness and shear modulus. Figure 21 shows the
dependence of the bulk modulus on the shape factor. The latter curve applies at 10% deformation.
Rubber at a shape factor of 0.25 is about 6-8 times softer when in shear than when in compression (for the
same rubber hardness). Since only 3-4 times the stress value in compression can be considered, it may be
said that rubber is best used in shear to achieve large deflections and good isolation properties, particularly
at low interference frequencies. The given modulus values apply at a static compression test where the end
surfaces can not slide, but when rubber is subjected to rapid vibrations it appears to be stiffer. Thus a
rubber mounting displays static and dynamic spring constants. The dynamic spring constant is dependent
144
Vibration Analysis on Rotating Equipment
on the frequency and amplitude. The effect of frequency, however, is small within the ranges in which
modern vibration mountings are most effective: from about 13 to 400 Hz.
The effect of amplitude is such that rubber will be stiffer the lower the amplitude. The ratio between
dynamic and static stiffness is dependent on the material and the geometrical shape of the rubber body in
the mounting.
One important question is how large a continuous stress can rubber be subjected to? When it comes to
shear, normally 0.25 MPa is considered to be a normal stress factor, but in some instances it can go as high
as 0.5 MPa without problems. Higher shear stresses are not suitable as they change to tensile stress.
Experience has shown that tensile loads are detrimental to rubber bonded designs. If subjected to high
shear stresses, rubber should be pre-compressed. When it comes to compression, 1 MPa is a normal stress
for the rubber bonded to metal with the restriction that the deformation will not exceed 20-25% of the
rubber thickness.
With hard rubber, we can except the stress to increase slightly. The fact that the element is made of rubber
and bonded to metal does not represent any restriction in the use of the rubber's strength properties; quite
the reverse, in fact. The bond between the rubber and the metal is often stronger than the rubber material
itself. If the area under load is large in relation to the free area (high shape factor), the modulus of elasticity
will be large and the deformation small. Higher compression stresses can then be utilized. Examples of
applications are bearings for bridges and elements for vibration-isolation of buildings. The surface pressure
here may be as high as 15-20 MPa.
One of the reasons why the aforementioned relatively low stresses for normal rubber mountings must be
taken into consideration is that the rubber element will be permanently deformed if it is subjected to too
high a load. This is called "compression set". The most important factor for reducing compression set is
that the rubber is of a high quality and correctly vulcanized. Moreover, it is essential that the rubber spring
is not exposed to high temperatures.
Damping
Looking at the formulae above we see that changing the damping will have zero effect on the natural
frequency of the system. In the real world, damping is usually associated with friction, either at the
molecular level or because of friction between machine components. The damping qualities of a structure
control the resonant response. The higher the damping effect the lower the amplitude of vibration will be as
a result of a resonant condition. For example a spring is a good example of a low damped system, a
relatively small amplitude of input results in a large amplitude response. A shock absorber is a good
example of a highly damped system where a large input results in a small response.
In order to have a resonant condition we need to have the right combination of stiffness and mass to create
the natural frequency and a source vibration that matches the natural frequency. The damping controls the
145
Vibration Analysis on Rotating Equipment
amplitude response. In small systems it is sometimes feasible to add a damper such as a shock absorber, but
usually we cannot change the damping of the system significantly.
I
ωn a 1 . E.
µ .l
4
Remember that µ is the mass per unit length and l is the length. So increasing the mass of the structure will
lower the natural frequency of the structure. Adding mass to a structure is often a very easy way of
checking your analysis of the problem. Just by adding sand bags to a structure we can check how much
weight we need to add to the structure to remove the natural frequency away from the exciting force
(without adding stiffness).
Take care when adding the finished mass (often in the form of steel) that you do not increase the stiffness of
the structure.
I
ωn a 1 . E.
µ .l
4
The combination of E and I may be considered to be the stiffness of the beam. Examining the formula it is
obvious that increasing the stiffness will increase the natural frequency of the beam. But the ratio of
increase is not linear. In other words, increasing the stiffness by 10% will increase the natural frequency by
only 3.16% (√10).
By increasing the value for I we can increase the natural frequency of the system. If stiffness is added to the
machine, great care should be exercised. In the first place, only stiffen in the direction which has a problem
with the natural frequency being co-incident with the exciting force.
• If you stiffen in places where it is not required, you could end up with a totally new set of
resonance problems.
• As you stiffen the structure, you will add mass. But additional mass reduces the natural frequency.
It is very possible that the additional stiffeners will increase the natural frequency because of their
stiffening properties but reduce the natural frequency due to the extra mass. In other words you
have done a lot of work for no change in the vibration characteristics of the machine.
If you think that adding stiffness is the correct solution for the resonance problem, be very careful about the
mass that is added. Additional stiffness can often be added by using turn-buckles at a very small mass
penalty.
146
Vibration Analysis on Rotating Equipment
The initial effects of whole body vibration are discomfort and irritation. This can contribute to fatigue and a
reduced ability to perform certain tasks. Prolonged or extreme exposure to whole body vibration has been
associated with gastrointestinal disorders, neural disorders, low back pain and spinal degeneration. In many
cases, vibration-related syndromes are aggravated by other occupational factors (work history, skill level,
work environment, posture) and non-occupational factors (age, health, fitness).
Frequency ranges
The specific effects also depend on the vibration frequencies to which the worker is exposed. Very low
frequency vibration (less than one Hz or cycle per second) can result in motion sickness; frequencies
between four and eight Hz are likely to affect the gastrointestinal and spinal systems; higher frequency
vibration (15 to 40 Hz) can interfere with vision. Whole body vibration usually originates from a moving
vehicle or from the operation of heavy machinery. Much can be gained by routine maintenance of the
147
Vibration Analysis on Rotating Equipment
vibration source and existing isolation systems. For more persistent problems, vibration measurements will
likely be necessary to make an educated selection of appropriate vibration control strategies.
15 – 40 Hz Vision difficulties
Whole body vibration transmitted to vehicle operators may be reduced by proper maintenance of roadways,
vehicle maintenance and control of vehicle speed. Tire pressure, track and suspension systems, as well as
seating and engine mounts should be routinely checked and serviced when required. Vehicle cab isolation is
available in some vehicles and can effectively reduce vibration exposure. If vibration control available
through vehicle and roadway maintenance is insufficient, improved seating selection may provide further
protection for the worker.
Generally, two types of seating systems are considered when addressing a vibration problem: conventional
foam seats or suspension seats. Conventional foam seats tend to amplify vibration frequencies in the range
of greatest susceptibility for the human body. Anti-vibration suspension seats include a damper and a spring
in the seat system. The spring may be constructed of steel or it may be a column of air. There are large
differences between seats and the performance of an individual seat may change during its lifetime. It is
important to know the vibration levels and frequencies in the vehicle in order to select an appropriate
suspension seat.
Control room vibration may result from machinery operation or heavy vehicles traveling in the vicinity.
This vibration may be addressed in a number of ways:
1. reduce vibration at the source (in the case of heavy machinery this may be accomplished by balancing
moving parts)
2. modify a vibrating system to reduce the likelihood of excessive vibration due to resonance (in the case
of a cover panel, this may be accomplished by •
2.1. stiffening the system by welding a steel section to it);
2.1.1. increase the damping in the system (damping materials may be coated on the system);
or •
2.1.2. isolate the system from the vibration source (mount the system on anti-vibration
mounts).
148
Vibration Analysis on Rotating Equipment
Glossary
A
Accelerometer
Transducer for measuring vibration in the form of acceleration. It is one of the most rugged transducers
for vibration measurements and also has the widest frequency range.
AC
Literally Alternating Current but often used to imply a time-fluctuating signal.
Acceleration
The rate of change of velocity in inches/second/second or G’s (acceleration due to gravity is 1 G). This
is a good indicator of the forces inside a machine since F = m.a (Newton’s 2nd law).
ADC
Acronym for Analog to Digital conversion
Alarm
Alarms are used to identity specific operating conditions or to define the boundaries between safe and
unsafe conditions. When an amplitude reaches or exceeds the alarm amplitude then the software
automatically generates an exception report.
Anti Aliasing
Nyquist's theorem says that as long as the sampling rate is greater than twice the highest frequency
component of the signal, then the sampled data will accurately represent the input signal. Certain
analyzers pass the input signal passes through an analog filter which attenuates all frequency
components above Fmax by 90 dB to make sure that Nyquist's theorem is satisfied. This is the anti-
aliasing filter.
ASCII
American Standards Code for Information Interchange.
Attenuation
The weakening of a signal by distance from the source or a mechanical interface.
Averaging
In general, averaging many spectra together improves the accuracy and repeatability of measurements.
B
Band Pass Filter
A measurement filter that removes data below the low cutoff frequency and above the high cutoff
frequency. The band pass filter only passes the data between the cutoff frequencies.
Bin
See “resolution”
Blackman-Harris Window
149
Vibration Analysis on Rotating Equipment
The Blackman-Harris window is a very good window to use with the spectrum analyzer. It has
better amplitude accuracy (about 0.7 dB) than the Hanning, very good selectivity and the fastest
filter rolloff. The filter is steep and narrow and reaches a lower attenuation than the other
windows. This allows signals close together in frequency to be distinguished, even when their
amplitudes are very different.
C
Calibration
The process of multiplying or dividing the voltage signal from a transducer by a factor that
represents a specific engineering quantity. e.g. 100 mV/G for an accelerometer.
Coherence
Coherence measures the percentage of power in channel 2 which is caused by (phase coherent
with) power in the input channel. Coherence is a unitless quantity which varies from 0 to l. If the
coherence is 1, all the power of the output signal is due to the input signal. If the coherence is 0,
the input and output are completely random with respect to one another. Coherence is related to
signal to noise ratio (S/N) by the formula:
S/N = γ2/(l -γ2)
where γ2 is the traditional notation for coherence.
Correlation
The two channel analyzer may also compute auto and cross correlation. Correlation is a time
domain measurement which is defined as follows:
Auto Correlation(τ) = ∫x*(t)x(t-τ)dt
Cross Correlation(τ) = ∫x*(t)y(t-τ)dt
where x and y are the channel 1 and channel 2 input signals and the integrals are over all time. It is
clear that the auto correlation at a time t is a measure of how much overlap a signal has with a
delayed-by-t version of itself, and the cross-correlation is a measure of how much overlap a signal
has with a delayed-by-t version of the other channel. Although correlation is a time domain
measurement the some analyzers use frequency domain techniques to compute it in order to make
the calculation faster.
Cross Spectrum
The cross spectrum is defined as:
cross spectrum = FFT2 conj(FFT1)
The cross spectrum is a complex quantity which contains magnitude and phase information. The
phase is the relative phase between the two channels. The magnitude is simply the product of the
magnitudes of the two spectra. Frequencies where signals are present in both spectra will have
large components in the cross spectrum.
CPM
Abbreviation for Cycles per Minute - the most common format for displaying frequency in
vibration analysis.
150
Vibration Analysis on Rotating Equipment
D
Database
A collection of information files that are ties together by a common topic. A database allows rapid
access of the database files.
Data Collector
Hardware device for collecting vibration data off line.
Decibel (dB)
A convenient unit for displaying logarithmic data.
dB = 10 log (X/Xο)
where X or Xο is
a power or squared quantity
or dB = 20 log (Y/Yο)
where Y or Yο is a linear quantity.
Decibels are usually referenced to a standard. e.g. 0.1 GdB re 0.001G.
DC
Literally Direct Current. In PdM terms this may be used to check the accelerometer or for a
process measurement.
Differentiate
Differentiation changes displacement to velocity and velocity to acceleration relative to time using
the formulae:
Acceleration = Velocity /(2.π.f)
Velocity = Displacement /(2.π.f)
or Acceleration = Displacement /(2.π.f)2
where f = frequency
n.b. remember 1G = 386 in/s2 or 9.806 m/S2
E
Envelope Measurements
A.k.a. demodulation (see appendix)
Exception
To have an alarm condition. Data is outside the defined “safe” area.
Exponential Averaging
Exponential averaging weights new data more than old data. Averaging takes place according to
the formula,
New Average = (New Spectrum - I/N) +(Old Average) - (N-l)/N
where N is the number of averages.
151
Vibration Analysis on Rotating Equipment
Exponential averages "grow" for approximately the first 5N spectra until the steady state values are
reached. Once in steady state, further changes in the spectra are detected only if they last
sufficiently long. Make sure that the number of averages is not so large as to eliminate the changes
in the data that might be important.
F
Flattop Window
The Flattop window improves on the amplitude accuracy of the Hanning window. Its between-bin
amplitude variation is about 0.02 dB. However, the selectivity is a little worse. Unlike the
Hanning, the Flattop window has a wide pass band and very steep rolloff on either side. Thus,
signals appear wide but do not leak across the whole spectrum.
FFT Spectrum Analyzer
FFT Spectrum Analyzers take a time varying input signal, like you would see on an oscilloscope
trace, and compute its frequency spectrum.
Fourier or FFT
Fourier's theorem states that any waveform in the time domain can be represented by the weighted
sum of sines and cosines. The FFT spectrum analyzer samples the input signal, computes the
magnitude of its sine and cosine components, and displays the spectrum of these measured
frequency components.
Frequency
The rate at which periodic events happen. Typical units are Hertz (Hz), Cycles per Minute (cpm)
or orders (multiples of run speed).
Fundamental Frequency
Primary frequency, such as operating speed, to which other frequencies may be referred back.
H
Hanning Window
The Hanning window is the most commonly used window. It has an amplitude variation of about
1.5 dB (for signals between bins) and provides reasonable selectivity. Its filter rolloff is not
particularly steep. As a result, the Hanning window can limit the performance of the analyzer
when looking at signals close together in frequency and very different in amplitude.
Harmonic
Frequencies at direct multiples of a fundamental frequency. The fundamental is not necessarily the
run speed.
Hertz (Hz)
Common frequency units in cycles per second. Named after, Heinrich Rudolf Hertz (1857-94).
German physicist born in Hamburg and educated at the University of Berlin. From 1885 to 1889
he was a professor of physics at the technical school in Karlsruhe and after 1889 a professor of
physics at the university in Bonn. Hertz clarified and expanded the electromagnetic theory of light
that had been put forth by the British physicist James Clerk Maxwell in 1884. Hertz proved that
electricity can be transmitted in electromagnetic waves, which travel at the speed of light and
which possess many other properties of light. His experiments with these electromagnetic waves
led to the development of the wireless telegraph and the radio. The unit of frequency that is
measured in cycles per second was renamed the hertz; it is commonly abbreviated Hz.
152
Vibration Analysis on Rotating Equipment
A measurement filter that removes data below its low cutoff frequency.
I
Input Couple
Measurements may be AC or DC coupled. Use DC coupling for process measurements and use
AC for vibration measurements. DC coupling includes both DC and AC signals whereas AC
coupling does not include any DC offsets.
Integrate
Integration is the opposite to differentiation and changes acceleration to velocity and velocity to
displacement where A = V x (2.π.f)
K
Kaiser
The Kaiser window, which is available on IRD analyzers, combines excellent selectivity and
reasonable accuracy (about 0.8 dB for signals between exact bins). The Kaiser window has the
lowest side-lobes and the least broadening for non-bin frequencies. Because of these properties, it
is the best window to use for measurements requiring a large dynamic range.
L
Leakage
Errors resulting from Fourier transforming nonperiodic time domain data. This effect is reduced
by using windows such as Hanning or Flat Top.
Linear
Linear scaling displays all data on an equal basis. Small signals may be hard to detect with linear
scaling but become visible with logarithmic scaling.
Linear Averaging
Linear averaging combines N (number of averages) spectra with equal weighting in either RMS,
Vector or Peak Hold fashion. This type of averaging is useful for eliminating transients.
Line of Resolution
see “resolution”
Logarithmic
Although the linear magnitude scale is used most often for displaying spectra, another way of
displaying amplitude is the Log Magnitude. The Log Mag display graphs the magnitude of the
spectrum on a logarithmic scale using dBEU (Engineering Units) as units.
The 16 bit analyzer has a dynamic range of about 90 dB. below full scale. Imagine what
something 0.01% of full scale would look like on a linear scale. If we wanted it to be 1 inch high
on the graph, the top of the graph would be 833 feet above the bottom - It turns out that the log
display is both easy to understand and shows features which have very different amplitudes clearly.
The real and imaginary parts are always displayed on a linear scale. This avoids the problem of
taking the log of negative voltages.
153
Vibration Analysis on Rotating Equipment
O
Octave Analysis
The magnitude of the normal spectrum measures the amplitudes within equally divided frequency bins.
Octave analysis computes the spectral amplitude in logarithmic frequency bands whose widths are
proportional to their center frequencies. The bands are arranged in octaves with either 1 or 3 bands per
octave (1/1 or 1/3 octave analysis). Octave analysis measures spectral power closer to the way people
perceive sound, that is, in octaves.
The center frequency of each band should be calculated according to ANSI standard S1.11 (1986).
Typically the shape of each band is a third-order Butterworth filter whose bandwidth is either a full, 1/3,
or 1/12 octave. The full octave bands have band centers at:
Center Freq: = 1 kHz x 2n
The 1/3 octave bands have center frequencies given by:
Center Freq: = 1 kHz x 2(n-30/3)
Operating System
A form of software that controls and supervises how the computer operates. It loads programs, handles
input and output operations and accepts and executes commands issued by the user.
Orbit
The orbit is simply a two dimensional display of the time record of channel 1 vs. the time record of
channel 2. The orbit display is similar to an oscilloscope displaying a "Lissajous" figure.
Orders
A frequency axis scale which is useful for viewing data as a function of the operating speed. The first
order corresponds to the operating speed, the second order is two times the operating speed and so on.
Overall
A single value representing the vibration or some other measurement parameter. For vibration
measurements this value includes vibration at all frequencies.
Overlap Processing
What about narrow spans where the time record is long compared to the processing time which is what
we normally see when taking vibration measurements? The analyzer computes one FFT per time record
and can wait until the next time record is complete before computing the next FFT. The update rate
would be no faster than one spectra per time record. With narrow spans, this could be quite slow.
And what is the processor doing while it waits? Nothing. With overlap processing, the analyzer does
not wait for the next complete time record before computing the next FFT. Instead it uses data from the
previous time record as well as data from the current time record to compute the next FFT. This speeds
up the processing rate. Remember, most window functions are zero at the start and end of the time
record. Thus, the points at the ends of the time record do not contribute much to the FFT. With
overlap, these points are “re-used" and appear as middle points in other time records. This is why
overlap effectively speeds up averaging and smoothes out window variations.
Typically, time records with 50% overlap provide almost as much noise reduction as non-overlapping
time records when RMS averaging is used. When RMS averaging narrow spans, this can reduce the
measurement time by a factor of two.
The amount of overlap is specified as a percentage of the time record. 0% is no overlap and 99.8% is the
maximum (511 out of 512 samples re-used). The maximum overlap is determined by the amount of
time it takes to calculate an FFT and the length of the time record and thus varies according to the span.
154
Vibration Analysis on Rotating Equipment
Peak
The maximum value as seen in the time domain data. For a sine wave
peak = RMS x 1.414
Peak Hold Averaging
Peak Hold is not really averaging, instead, the new spectral magnitudes are compared to the
previous data, and if the new data is larger, then the new data is stored. This is done on a
frequency bin by bin basis. The resulting display shows the peak magnitudes which occurred in
the previous group of spectra.
Peak Hold detects the peaks in the spectral magnitudes and only applies to Spectrum, PSD, and
Octave Analysis measurements. However, the peak magnitude values are stored in the original
complex form. If the real or imaginary part or phase is being displayed for spectrum
measurements, the display shows the real or imaginary part or phase of the complex peak value.
Phase
In general, phase measurements are only used when the analyzer is triggered. The phase is relative
to the pulse of the trigger. The phase is displayed in degrees or radians on a linear scale, usually
from -180 to +180 degrees. The phase of a particular frequency bin is set to zero in most analyzers
if neither the real nor imaginary part of the FFT is greater than about 0.012% of full scale (-78 dB
below f.s.). This avoids the messy phase display associated with the noise floor. (Remember, even
if a signal is small, its phase extends over the full 360 degrees.)
Power Spectral Density (PSD)
The PSD is simply the magnitude of the spectrum normalized to a 1 Hz bandwidth. This
measurement approximates what the spectrum would look like if each frequency component were
really a 1 Hz wide piece of the spectrum at each frequency bin.
What good is this? When measuring broad band signals such as noise, the amplitude of the
spectrum changes with the frequency span. This is because the line width changes so the
frequency bins have a different noise bandwidth. The PSD, on the other hand, normalizes all
measurements to a 1 Hz bandwidth and the noise spectrum becomes independent of the span. This
allows measurements with different spans to be compared. If the noise is Gaussian in nature, then
the amount of noise amplitude in other bandwidths may be approximated by scaling the PSD
measurement by the square root of the bandwidth. Thus the PSD is displayed in units of V/√Hz or
dBV/√Hz.
Since the PSD uses the magnitude of the spectrum, the PSD is a real quantity. There is no real or
imaginary part or phase.
R
Rayleigh’s Principle
This principle states that ∆f is the lowest measurable frequency for a time record length T:
∆f = 1/T
155
Vibration Analysis on Rotating Equipment
The rectangular windows applies an even weighting (no window) over the time period. It does not
reduce leakage errors and should only be used for impulsive or transient data that dies out within the
time sample period.
Resolution
The accuracy of a reading based on the number of discrete values used to define it. For a frequency
domain spectrum this refers to the number of lines (or bins) of resolution that are combined to display
the spectral data. For example a spectrum of 500 Hz Fmax with 400 bins or lines would have a line
resolution of 1.25 Hz.
RMS (Root Mean Square)
The square root of the average of a set of squared values. For a sine wave RMS = Peak x 0.7071
RMS Averaging
RMS averaging computes the weighted mean of the sum of the squared magnitudes (FFT times its
complex conjugate). The weighting is either linear or exponential.
RMS averaging reduces fluctuations in the data but does not reduce the actual noise floor. With a
sufficient number of averages, a very good approximation of the actual random noise floor can be
displayed.
Since RMS averaging involves magnitudes only, displaying the real or imaginary part or phase of an
RMS average has no meaning. The RMS average has no phase information.
Route
For off-line data collectors this is a an ordered list of points containing the sequence for collecting data.
S
Sampling Rate
The original digital time record comes from discrete samples taken at the sampling rate. The
corresponding FFT yields a spectrum with discrete frequency samples. In fact, the spectrum has less
than half as many frequency points as there are time points. Suppose that you take 1024 samples at
2560 Hz. It takes 0.4 Seconds to take this time record. The FFT of this record yields 400 frequency
points or lines, but over what frequency range? The highest frequency will be determined by the in-built
ratio of F-max to data sampling rate - 2.56. The lowest frequency is just the F-max divided by the
number of lines:
F-max = data sampling rate / 2.56
No. Of Lines = No samples / 2.56
Bin resolution = F-max / No. of lines
= (2560 / 2.56) / (1024 / 2.56)
= 2.5 Hz (the same as the lowest measurable frequency)
Everything below 2.5 Hz is considered to be DC. The output spectrum thus represents the frequency
range from DC to 1000 Hz with points every 2.5 Hz.
Sideband
A frequency which occurs either side of a fundamental frequency. Sidebands occur because of a
modulation of the fundamental by another frequency.
Spectrum
156
Vibration Analysis on Rotating Equipment
The spectrum is the basic measurement of an FFT analyzer. It is simply the complex FFT. Normally,
the magnitude of the spectrum is displayed. The magnitude is the square root of the FFT times its
complex conjugate. (Square root of the sum of the real (sine) part squared and the imaginary (cosine)
part squared). The magnitude is a real quantity and represents the total signal amplitude in each
frequency bin, independent of phase.
If there is phase information in the spectrum, i.e. the time record is triggered in phase with some
component of the signal, then the real (cosine) or imaginary (sine) part or the phase may be displayed.
The phase is simply the arc tangent of the ratio of the imaginary and real parts of each frequency
component. The phase is always relative to the start of the triggered time record.
T
Tachometer
A device for measuring the speed of rotation.
Time Record
The time record measurement displays the filtered data points before the FFT is taken For baseband
spans (spans that start at DC), the time record is a real quantity. For non-baseband spans (zooms) the
heterodyning discussed earlier transforms the time record into a complex quantity which can be
somewhat difficult to interpret.
Time Synchronous Averaging
see “vector averaging”
Transfer Function
The transfer function is the ratio of the spectrum of channel 2 to the spectrum of channel 1. For the
transfer function to be valid, the input spectrum must have amplitude at all frequencies over which the
transfer function is to be measured.
Two-Channel Measurements
Two-channel analyzers offer additional measurements such as transfer function, cross-spectrum,
coherence and orbit.
U
Uniform Window
The uniform window is actually no window at all. The time record is used with no weighting. A
signal will appear as narrow as a single bin if its frequency is exactly equal to a frequency bin. (It is
exactly- periodic within the time record). If its frequency is between bins, it will affect every bin of
the spectrum. These two cases also have a great deal of amplitude variation between them (up to 4
dB).
In general, this window is only useful when looking at transients which do not fill the entire time
record.
157
Vibration Analysis on Rotating Equipment
V
Vector (Synchronous Time) Averaging
Vector averaging averages the complex FFT spectrum. (The real part is averaged separately from the
imaginary part.) This can reduce the noise floor for random signals since they are not phase coherent
from time record to time record.
Vector averaging requires a trigger. The signal of interest must be both periodic and phase
synchronous with the trigger. Otherwise, the real and imaginary parts of the signal will not add in
phase and instead will cancel randomly.
With vector averaging, the real and imaginary parts as well as phase displays are correctly averaged
and displayed. This is because the complex information is preserved.
W
Windowing
What is windowing? Let's go back to the time record. What happens if a signal is not exactly
periodic within the time record? We said that its amplitude is divided into multiple adjacent
frequency bins. This is true but it's actually a bit worse than that. If the time record does not start
and stop with the same data value, the signal can actually smear across the entire spectrum. This
smearing will also change wildly between records because the amount of mismatch between the
starting value and ending value changes with each record.
Windows are functions defined across the time record which are periodic in the time record. They
start and stop at zero and are smooth functions in between. When the time record is windowed, its
points are multiplied by the window function, time bin by time bin, and the resulting time record is
by definition periodic. It may not be identical from record to record, but it will be periodic (zero at
each end).
In the frequency domain, a window acts like a filter. The amplitude of each frequency bin is
determined by centering this filter on each bin and measuring how much of the signal falls within
the filter. If the filter is narrow, then only frequencies near the bin will contribute to the bin. A
narrow filter is called a selective window - it selects a small range of frequencies around each bin.
However, since the filter is narrow, it falls off from center rapidly. This means that even
frequencies close to the bin may be attenuated somewhat. If the filter is wide, then frequencies far
from the bin will contribute to the bin amplitude but those close by will not be attenuated
significantly.
The net result of windowing is to reduce the amount of smearing in the spectrum from signals not
exactly periodic with the time record. The different types of windows trade off selectivity,
amplitude accuracy, and noise floor.
Several types of window functions are available including Uniform (none), Flattop, Hanning,
BlackmanHarris, and Kaiser.
158
Vibration Analysis on Rotating Equipment
Index
Decibel, 152
Demodulation, 2, 6, 25, 26, 27
—A— Developed Fatigue, 7, 67
AC, 154 Differentiate, 152
acceleration, 2, 6, 7, 13, 14, 15, 17, 22, 23, 30, 69, 70, displacement, 2, 6, 7, 15, 16, 17, 18, 22, 77, 79, 80,
71, 72, 74, 93, 94, 103, 104, 117, 118, 119, 127, 110, 117, 118, 121, 122, 127, 152, 154
150, 152, 154 Displacement, 127
accelerometer, 6, 7, 22, 23, 24, 28, 71, 87, 88, 90,
131, 151, 152 —E—
Accelerometer, 7, 41, 89, 90, 150
Accelerometers, 2, 24 eccentric rotor, 32
Aerodynamic cross coupling, 3, 60 Elasto Hydrodynamic Lubrication, 3, 62
aerodynamic forces, 3, 59 envelope, 6, 13, 27, 31, 69, 93, 103
Air, 120 Envelope, 6, 13, 93, 152
Analysis, 2, 3, 4, 9, 70, 87, 91, 92, 93, 101, 102, 103, Enveloping, 6, 27
106, 155, 156 External manifestation, 3, 92
Analyzer, 153
Armature, 2, 32
Average, 152
—F—
Averaging, 4, 100, 101, 150, 152, 154, 156, 157, 158, Failure type, 3, 91
159 fan, 6, 7, 9, 10, 11, 12, 13, 29, 45, 59, 109, 130, 140,
142
—B— fans, 9, 59, 60, 69
FFT, 2, 5, 6, 18, 19, 20, 21, 28, 36, 38, 39, 40, 41, 42,
Balancing, 3, 75, 78, 80, 82, 85 43, 99, 100, 101, 102, 107, 130, 131, 133, 153,
Base cause, 3, 91 155, 156, 157, 158, 159
bearing, 3, 6, 7, 10, 13, 14, 22, 26, 28, 29, 30, 38, 41, Filter, 154, 155
43, 45, 60, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, Firing Cards, 3, 37
72, 73, 74, 78, 91, 92, 93, 94, 95, 103, 104, 110, Flattop, 99, 100, 153, 159
111, 112, 113, 114, 115, 130 Fourier, 6, 20, 28, 134, 153, 154
Bearing, 2, 3, 6, 7, 13, 26, 45, 62, 63, 64, 65, 66, 67, Frequency, 2, 3, 4, 5, 19, 21, 34, 67, 103, 104, 115,
68, 70, 114 135, 148, 149, 153
Benzene, 119 Fundamental, 153
Bin, 157
Blackman-Harris, 99, 100, 151
Bode, 132
—G—
Boltzman, 121 Gas, 121
bowed rotor, 32 Glycerin, 119
BPIR, 13
Bump test, 129, 130
—H—
—C— Hanning, 93, 94, 99, 100, 151, 153, 154, 159
Harmonic, 108, 153
Campbell, 124 Harmonics, 120
Choking, 3, 61 Hertz, 153
Coherence, 106, 151 Hydraulic valve, 120
Comparitor Card, 3, 40, 41 hysteresis whirl, 60
Compression mode, 6, 24 Hz, 21, 26, 153, 156, 157, 158
Correlation, 106, 151
CPM, 151
Cracking, 126 —I—
Critical speed, 123 Imbalance, 2, 3, 6, 9, 45, 46, 47, 48, 49, 50, 75, 78,
104
—D— Imbalanced phases, 33
Impact hammer, 131, 132
Damping, 117, 139, 143, 145, 146 Impeller, 120
DC, 2, 3, 6, 21, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, Induction Motors, 2, 32
152, 154, 158 Integrate, 154
159
Vibration Analysis on Rotating Equipment
160
Vibration Analysis on Rotating Equipment
161