You are on page 1of 161

Vibration Analysis on Rotating Equipment

The use of vibration analysis in the condition assessment of


rotating equipment

Prepared by: Ron Frend

COPYRIGHT RONALD FREND 2002

1
Vibration Analysis on Rotating Equipment

CONTENTS

VIBRATION ANALYSIS - AN INTRODUCTION -------------------------------------------9

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

How does it work? ------------------------------------------------------------------------------------------------- 22


Accelerometers -------------------------------------------------------------------------------------------------------24

Acceleration Amplitude Demodulation ------------------------------------------------------------------------ 25


Theory -----------------------------------------------------------------------------------------------------------------25
The Demodulation Process -----------------------------------------------------------------------------------------26
Resonance Sources---------------------------------------------------------------------------------------------------28
A.C. Motor Example.------------------------------------------------------------------------------------------------29

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

DC Control Comparitor Card --------------------------------------------------------------------------------------40


Example of a Comparitor Card Defect ---------------------------------------------------------------------------41
Importance of Exact RPM ------------------------------------------------------------------------------------------42
Example of Mechanical -v- Electrical Frequencies ------------------------------------------------------------43

Rotating Equipment ----------------------------------------------------------------------------------------------- 45


Imbalance--------------------------------------------------------------------------------------------------------------45
Vibration due to imbalance-----------------------------------------------------------------------------------------46
Misalignment ---------------------------------------------------------------------------------------------------------51
Looseness--------------------------------------------------------------------------------------------------------------54

Vibration due to aerodynamic forces--------------------------------------------------------------------------- 59


Aerodynamic cross coupling ---------------------------------------------------------------------------------------60
Surging-----------------------------------------------------------------------------------------------------------------60
Choking or Stone Walling ------------------------------------------------------------------------------------------61

Bearing Failures ---------------------------------------------------------------------------------------------------- 62


Elasto Hydrodynamic Lubrication --------------------------------------------------------------------------------62
First Stage of Bearing Failure--------------------------------------------------------------------------------------63
Second Stage of Bearing Failure ----------------------------------------------------------------------------------64
Third Stage of Bearing Failure-------------------------------------------------------------------------------------65
Fourth Stage of Bearing Failure -----------------------------------------------------------------------------------66
Bearing Defect Frequency Calculation ---------------------------------------------------------------------------67
Analysis of bearing defects -----------------------------------------------------------------------------------------70

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

SINGLE CHANNEL ANALYSIS -------------------------------------------------------------- 87

Taking measurements --------------------------------------------------------------------------------------------- 87

POTENTIAL FAILURE ANALYSIS ---------------------------------------------------------- 91

A methodology for objective set up ----------------------------------------------------------------------------- 91

Introduction --------------------------------------------------------------------------------------------------------- 91

The PFA Tree ------------------------------------------------------------------------------------------------------- 91


Base cause-------------------------------------------------------------------------------------------------------------91
Failure type------------------------------------------------------------------------------------------------------------91
External manifestation-----------------------------------------------------------------------------------------------92
Technology------------------------------------------------------------------------------------------------------------92
Parameter --------------------------------------------------------------------------------------------------------------92
Analysis----------------------------------------------------------------------------------------------------------------92
Interval-----------------------------------------------------------------------------------------------------------------92
Setup -------------------------------------------------------------------------------------------------------------------92

3
Vibration Analysis on Rotating Equipment

Developing a Potential Failure Analysis for Rolling Element Bearings --------------------------------- 93


Stage 1 -----------------------------------------------------------------------------------------------------------------93
Stage 2 -----------------------------------------------------------------------------------------------------------------94
Stage 3 -----------------------------------------------------------------------------------------------------------------94
Stage 4 -----------------------------------------------------------------------------------------------------------------94

Including the Component Failure in the PFA Tree. --------------------------------------------------------- 95

Conclusion ----------------------------------------------------------------------------------------------------------- 96
Measurement Windows ---------------------------------------------------------------------------------------------99
Averaging ----------------------------------------------------------------------------------------------------------- 100
Real Time Bandwidth and Overlap Processing --------------------------------------------------------------- 101
Octave Analysis ---------------------------------------------------------------------------------------------------- 102

Analysis ------------------------------------------------------------------------------------------------------------ 103


Severity charts ------------------------------------------------------------------------------------------------------ 105

Two Channel Analysis------------------------------------------------------------------------------------------- 106


Two channel functions -------------------------------------------------------------------------------------------- 106

Advanced functions ---------------------------------------------------------------------------------------------- 108


Representation by complex numbers --------------------------------------------------------------------------- 108
Cascade & waterfall plots----------------------------------------------------------------------------------------- 109
Triggering ----------------------------------------------------------------------------------------------------------- 109
Bodé plots ----------------------------------------------------------------------------------------------------------- 110
Orbits----------------------------------------------------------------------------------------------------------------- 110

INTRODUCTION TO RESONANCE ------------------------------------------------------ 114

What is resonance? ---------------------------------------------------------------------------------------------- 115

Natural Frequency ----------------------------------------------------------------------------------------------- 115


Mechanical ---------------------------------------------------------------------------------------------------------- 115
Liquids & pumping systems-------------------------------------------------------------------------------------- 118
Air & gases---------------------------------------------------------------------------------------------------------- 120
Karman Vortices --------------------------------------------------------------------------------------------------- 122

Critical Speed (Balance Resonance) -------------------------------------------------------------------------- 123

IDENTIFYING RESONANCE IN MECHANICAL SYSTEMS ----------------------- 126

Mode Shape ------------------------------------------------------------------------------------------------------- 126

Phase---------------------------------------------------------------------------------------------------------------- 127

The bump test ----------------------------------------------------------------------------------------------------- 129


Running machine Bump Test ------------------------------------------------------------------------------------ 130
Reverse Bump or “Plucking” the Suspect Part. --------------------------------------------------------------- 130

Set up for FFT-type analyzers --------------------------------------------------------------------------------- 130


Impact hammer ----------------------------------------------------------------------------------------------------- 131

4
Vibration Analysis on Rotating Equipment

CALCULATING NATURAL FREQUENCY IN MECHANICAL SYSTEMS------ 135

Uniform Beams --------------------------------------------------------------------------------------------------- 135

Plates---------------------------------------------------------------------------------------------------------------- 137

DETUNING RESONANT STRUCTURES------------------------------------------------ 139

Vibration isolators ----------------------------------------------------------------------------------------------- 139


Springs --------------------------------------------------------------------------------------------------------------- 140
Rubber --------------------------------------------------------------------------------------------------------------- 142

Modifying the structure----------------------------------------------------------------------------------------- 145


Damping ------------------------------------------------------------------------------------------------------------- 145
Changing the Mass------------------------------------------------------------------------------------------------- 146
Changing the stiffness --------------------------------------------------------------------------------------------- 146

WHOLE BODY VIBRATION ---------------------------------------------------------------- 147

Sources of vibration---------------------------------------------------------------------------------------------- 147

Frequency ranges ------------------------------------------------------------------------------------------------ 147

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

Figure 54 Looseness slide 3 ____________________________________________________________ 55


Figure 55 Looseness slide 4 ____________________________________________________________ 55
Figure 56 Looseness slide 5 ____________________________________________________________ 56
Figure 57 Looseness slide 6 ____________________________________________________________ 56
Figure 58 Looseness slide 7 ____________________________________________________________ 57
Figure 59 Looseness slide 8 ____________________________________________________________ 57
Figure 60 Looseness slide 9 ____________________________________________________________ 58
Figure 61 Aerodynamic forces __________________________________________________________ 60
Figure 62 Elasto-hydrodynamic lubrication ________________________________________________ 62
Figure 63 Loss of Lubricant - Ball Bearing Inner Race Courtesy of the Barden Corporation__________ 63
Figure 64 Loss of Lubricant - Roller Bearing Courtesy of the Torrington Company _________________ 64
Figure 65 Waterfall plot from a damaged motor bearing ______________________________________ 65
Figure 66 Early Fatigue - Ball Bearing Courtesy of the Barden Corporation ______________________ 66
Figure 67 Developed Fatigue on Roller Bearing Courtesy of the Torrington Company ______________ 66
Figure 68 Ball Bearing Terminology _____________________________________________________ 68
Figure 69 Waterfall of early damage to a motor bearing collected every 1.5 hrs over 14 days _________ 69
Figure 70 Bearing damage severity assessment chart ________________________________________ 70
Figure 71 Demodulated acceleration spectrum from a dry bearing ______________________________ 70
Figure 72 Demodulated acceleration spectrum of a marked bearing_____________________________ 71
Figure 73 Demodulated acceleration spectrum from a slightly more heavily marked bearing _________ 72
Figure 74 Time waveform from a marked bearing.___________________________________________ 72
Figure 75 Time waveform from a heavily marked bearing _____________________________________ 73
Figure 76 Velocity spectrum from a spalled bearing _________________________________________ 73
Figure 77 Sources of imbalance _________________________________________________________ 75
Figure 78 Assembly tolerance stack up ____________________________________________________ 76
Figure 79 Heavy spot _________________________________________________________________ 76
Figure 80 Units of measure of imbalance __________________________________________________ 77
Figure 81 Mass centre displacement______________________________________________________ 77
Figure 82 Force due to imbalance _______________________________________________________ 78
Figure 83 The vector diagram___________________________________________________________ 80
Figure 84 Simplified vector diagram______________________________________________________ 81
Figure 85 Additional corrections ________________________________________________________ 82
Figure 86 Direction to shift the weight ____________________________________________________ 83
Figure 87 Sample problem vector diagram_________________________________________________ 84
Figure 88 Determining the flash angle ____________________________________________________ 86
Figure 89 Typical tap block for mounting an accelerometer ___________________________________ 87
Figure 90 Accelerometer mounting techniques a-d___________________________________________ 89
Figure 91 Accelerometer mounting techniques e-g___________________________________________ 90
Figure 92 Overview of accelerometer mounting techniques ____________________________________ 90
Figure 93 PFA development for rolling element bearings _____________________________________ 93
Figure 94 PFA for a main motor_________________________________________________________ 95
Figure 95 General severity chart for vibration _____________________________________________ 105
Figure 96 Vector addition of 2 vibrations _________________________________________________ 108
Figure 97 Cascade of fan over 20mS ____________________________________________________ 109
Figure 98 Bode plots _________________________________________________________________ 110
Figure 99 Orbit showing misalignment___________________________________________________ 111
Figure 102 Harmonic series for the tone C. _______________________________________________ 120
Figure 103 Sonic vibration in a tube_____________________________________________________ 122
Figure 104 Karman vortices ___________________________________________________________ 123
Figure 107 Campbell Diagram _________________________________________________________ 125
Figure 109 Mode shape readings _______________________________________________________ 127
Figure 110 Phase relationships_________________________________________________________ 128
Figure 111 Phase / frequency relationships @ resonance ____________________________________ 129
Figure 112 Impact hammer response ____________________________________________________ 132
Figure 113 Impact hammer specification sheet_____________________________________________ 132

7
Vibration Analysis on Rotating Equipment

Figure 114 Stress/strain diagram for steel ________________________________________________ 140


Figure 115 Stress/strain diagram for rubber ______________________________________________ 140
Figure 116 Resonance Curve __________________________________________________________ 141

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 Analysis - an introduction


The study of noise and vibration phenomena dates back centuries. The first recorded incidence of such
study was by Pythagoras in the sixth century B.C. who studied the origin of musical sounds and the
vibration of strings. In 1638 Galileo described the vibrations of pendulums, the phenomenon of resonance
and the factors influencing the vibration of strings. Euler in 1744 and Bernoulli in 1751 developed the
equation for the vibrations of beams and developed the normal modes for various boundary conditions. In
1882 Hertz developed the first successful theory for impact. So we can see that vibration analysis itself is
not new but some of the ways that we take the measurements and apply those measurements as machine
health diagnoses are very new.
In this section we will briefly take a look at some vibration examples of typical defects suffered by fans and
fan drives without delving too deeply into why!

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 Velocity spectrum showing 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 - Velocity spectrum showing fan imbalance

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

Figure 3 Velocity spectrum of misaligned fan - radial

In figure 3 we see what initially looks like an imbalance condition of the fan, although the amplitudes are
relatively low.

Figure 4 Velocity spectrum of misaligned fan - axial

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

Figure 5 Velocity spectrum from a loose fan drive motor

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

Figure 6 Envelope spectrum of a fan drive motor with loose bearing

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.

Rolling Element Bearing Defects


The primary tool in assessing bearing condition is the use of enveloped acceleration readings.
Figure 7 shows the envelope spectrum from a bearing with a severe spall in the inner race. Notice that the
spike at about 8,772 cpm is marked BPIR which stands for Ball Pass Outer Race. We will study bearing
defects in detail later but notice that the main defects are not multiples of run speed. In other words they are
non-synchronous.

13
Vibration Analysis on Rotating Equipment

Figure 7 Enveloped acceleration spectrum of bearing - inner race defect

Figure 8 Inner race spall

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.

Simple Harmonic Motion


Simple harmonic motion can be visualized by many common examples such as a pendulum, a mass and
spring combination, a rotating mechanism or a diving board- Figure 9 uses a pendulum. If the pendulum
swings back and forth 100 times in one minute, then the frequency is 100 cycles per minute. Similarly if a
machine is rotating 100 times in one minute, its speed is 100 revolutions per minute or 100 RPM.
The frequency of vibration is often expressed in terms of cycles per second or HERTZ after the German
physicist Heinrich Hertz. However for predictive maintenance techniques where rotational speed is often
the key to vibration peaks, cycles per minute are used in preference to Hertz.
In addition to frequency the amplitude is the other necessary quantity that must be known in order to
characterize vibration. In figure 9 the points B and C represent the extreme position of the pendulum and
the distance between them is the peak to peak displacement-
Amplitude meters are often calibrated to give the peak to peak value because it is the displacement extremes
that are of interest. In vibration work, the displacement is often expressed in terms of mils or micron. One
Mil is equal to 0.001 inch and one micron (µm) is 0.001 mm.
Since the pendulum is continuously moving, it has a velocity associated with each position and, like
displacement, the velocity also varies between a positive peak and a negative peak.

15
Vibration Analysis on Rotating Equipment

(48)

* FREQUENCY

* AMPLITUDE

... Displacement

... Velocity

... Acceleration

Vibration is described by its frequency and amplitude.


The amplitude is expressed in units of either displacement,
velocity or acceleration.

one cycle

C B

An oscillating system will produce a certain number of


cycles per unit time, called the frequency. Frequency is usually
expressed in terms of cycles per second, or Hertz.

Figure 9 Simple Harmonic Vibration

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

Figure 3. The distance between the extremes of motion is the


peak-to-peak displacement. Displacement meters are often calibrated
in peak-to-peak units. The amplitude is one half of the peak-to-peak
value for a sine wave.
A B A C A

Disp
Vel

C B

Highest
Velocity

Figure 4. Velocity is highest where displacement is zero and is zero


o
where displacement is maximum. Therefore a 90 phase shift exists
between displacement and velocity. The velocity amplitude is directly
proportional to frequency for a given displacement.

Figure 10 Integration from acceleration to 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

Figure 5. At B, acceleration is maximum to the left and displacement


o
maximum to the right, a 180 phase shift. Acceleration amplitude varies
as the square of frequency for a given value of displacement.

Peak

Avg RMS
Peak
Amplitude

to
Peak
Time

Avg = 0.637 x Peak


RMS = 0.707 x Peak

Figure 6. The simple relationships which exist between average, RMS


and Peak amplitude values for sine waves are not valid for
combination or random waveforms.

Figure 11 Integrating to displacement

RMS vs. PEAK


The rms or root mean square value is calculated by breaking the waveform down into a number of points,
squaring the amplitude value of each point, calculating the mean of the squared values and then finding the
square root of the mean.
Using rms values can be compared to the use of rms in electrical circles i.e. stereo speakers power values
are measured in rms values. Electrical (AC) voltage is also measured in rms. This, like vibration signals, is
a continuously varying quantity, ranging from zero to a peak value. To measure only the peak value may be
misleading since the voltage is actually at a peak for only a small portion of the cycle. During most of the
cycle the value of the instantaneous voltage is somewhere between zero and peak.
RMS, then, is an attempt to apply a single quantitative value -which reflects the effective value of this
varying function. This same logic applies to vibration. Velocity is a quantitative measure of the effective
velocity and reflects the power or energy being used to vibrate the machine mass.
Peak value is the maximum amplitude seen during the measurement. When using FFT analyzers care
should be taken when evaluating peak or rms severity as the peak amplitude in the spectrum is derived from
a sine wave. True peak can be seen in the time waveform.

18
Vibration Analysis on Rotating Equipment

(51)

Peak

Avg RMS
Peak
Amplitude

to
Peak
Time

Simple Sine Wave

Peak
Peak
to
Amplitude

RMS Peak

Time

Complex Waveform

Figure 12 Peak -v- RMS

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

The Frequency Domain


If we now convert a time waveform to the frequency domain we will get a totally different picture. We now
have axes of amplitude v frequency instead of amplitude -v- time. Every sine wave separated out by the
FFT appears as a separate line. Its height represents its amplitude, its position represents its frequency.
The method most analyzers use to transform signals from the time domain to the frequency domain is called
:-

19
Vibration Analysis on Rotating Equipment

FFT (Fast Fourier Transform)

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.

The FFT Analyzer


An FFT spectrum analyzer works in an entirely different way. The input signal is digitized at a high
sampling rate, (2.56 x Fmax usually). 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 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. The resulting digital time record is then mathematically transformed into a frequency
spectrum using an algorithm known as the Fast Fourier Transform or FFT. The FFT is simply a clever set
of operations which implements Fourier's theorem. The resulting spectrum shows the frequency
components of the input signal.
Now here's the interesting part. 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 (remember Nyquist's
theorem). 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 Fmax to data sampling rate - 2.56. The lowest
frequency is just the Fmax divided by the number of lines:
Fmax = data sampling rate / 2.56
No. Of Lines = No samples / 2.56
Bin resolution = Fmax / No. of lines
= (2560 / 2.56) / (1024 / 2.56)
= 2.5 Hz (the same as the lowest measurable frequency)

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.

Advantages of FFT Analyzers


The advantage of this technique is its speed. Because FFT spectrum analyzers measure all frequency
components at the same time the technique offers the possibility of being hundreds of times faster than
traditional analog spectrum analyzers. In the case of a 1000 Hz span and 400 resolvable frequency bins, the
entire spectrum takes only 400 mS to measure. To measure the signal with higher resolution the time record
is increased, but again, all frequencies are examined simultaneously, providing an enormous speed
advantage.

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

Selecting displacement, velocity or acceleration


As previously discussed, displacement amplitude is higher at lower frequencies. Therefore when motions
are being measured a displacement measurement is in order because frequencies of interest on the shaft are
limited to 20 or so orders of rotation. For a 3600 rpm machine, 20 orders is a frequency of 1200 Hz. At
that frequency 0.5 in/sec is 0.13 mils pk-pk, very small but certainly a measurable value.
For higher frequencies however, significant vibration has a displacement value which -is too small to
conveniently measure and velocity or acceleration is more appropriate. Velocity measurements are
especially good for a number of reasons.
The most prominent advantage of a velocity measurement is that the value of rms velocity is related to the
potential for mechanical damage, regardless of the frequency. The many published vibration severity charts
are based on this principle. As an example suppose a displacement of 0.l mils is observed, is this severe?
At 6 Hz this is not severe at all; at 60 Hz this is rough but at 200 Hz this is very rough and should not be
permitted for machines up to the 100HP class. Now suppose a velocity of 0.6 in/sec (15mm/s) is observed.
Is this severe? The answer is “Yes, this is severe regardless of the frequency”.
Newton’s second law (F=ma) tells us that the acceleration of a body is directly proportional to the force
applied to the body. In other words the acceleration vibration gives a good indication of impactive forces
inside the machine such as bad bearings.
In summary, displacement measurements are good from 0 Hz to 500 Hz, velocity up to 1 kHz and
acceleration from 2 Hz to 20 kHz depending on the design of the accelerometer. In applying this to rotating
machinery displacement measurements are relative readings of the displacement of the shaft to a reference,
usually the bearing. Velocity and acceleration measurements are usually made on the bearing cap or on the
machine casing in way of a structural web to enhance the transmission of vibration to the pick-up point.

How does it work?


Consider a rotating machine (a motor) which has, for example, an out of balance condition on the rotor so
that for every revolution of that rotor the out of balance mass generates a centripetal (opp. to centrifugal)
force. We place our transducer on the drive end of the motor in the vertical direction, as near as possible to
the bearing and couple the transducer to a spectrum analyzer. The transducer sees the force once per rev. of
the rotor as a “simple harmonic motion”. That is to say that the machine surface will cause the transducer to
move in a downwards direction with the machine as the force itself is acting downwards and will cause the
transducer to move upwards when the machine is moving up etc. The output of the transducer will depend
on what type of transducer we are using.
• Displacement transducers will give an output proportional to the linear displacement
of the transducer in thousandths of a inch or micron.
• Velocity transducers will give an output proportional to the linear speed (velocity) of
the transducer in inches/second or millimeters/second.
• An accelerometer will give an output which is proportional to the acceleration of the
transducer in G’s or inches/second/second or meters/second/second.
For predictive maintenance purposes we use accelerometers almost exclusively so we will concentrate on
them for now. According to Newton’s Second Law

F=m⋅⋅x a

22
Vibration Analysis on Rotating Equipment

where F = the force


m = the mass
a = the acceleration
So immediately we see that the output from the accelerometer is directly proportional to the internal forces
acting on the machine. Newton also says that for a rotating body

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

Figure 13 Compression mode accelerometer

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.

Figure 14 Shear mode accelerometer

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.

Acceleration Amplitude Demodulation

Theory
But before we look at any case histories using DEMODULATION we should be clear in our mind about
exactly what is MODULATION.

Figure 15 Simple modulation example

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

Figure 16 Bearing modulation example

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.

The Demodulation Process


The time waveform of a machine with a bearing in the early stages of deterioration will look like the top
plot below. The bearing excitation resonance is shown as small, high frequency pulses sitting on top of the
high amplitude, low frequency vibration.

26
Vibration Analysis on Rotating Equipment

Figure 17 Demodulation process

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.

Figure 18 Enveloping process

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

Figure 19 Fast Fourier Transform

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).

Figure 20 FFT - 3D view

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.

A.C. Motor Example.


This plot shows the signal from the inboard bearing of a 35 H.P. A.C. motor operating a belt-driven fan.

Figure 21 Two channel time waveform - bearing defect

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

Figure 22 High frequency waterfall

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.

Figure 23 Enveloped acceleration spectrum

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.

Figure 24 Comparison - velocity to envelope

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

Mechanical or Electrical Effects


Vibration of electrical motors can be either mechanical or electrical in origin. Mechanical problems may
include imbalance, misalignment, defective bearings and looseness. Vibration caused by electrical
problems are normally the result of unequal magnetic forces acting on the rotor or stator. The unequal
magnetic forces may be due to open or shorted windings, broken rotor bars, unbalanced phases, unequal air
gap and other similar problems. Generally, the largest component frequency of vibration resulting from
these electrical problems will be 1 x RPM and, this will appear similar to imbalance. A common way to
check for electrical vibration is to observe the change in vibration amplitude the instant electrical power is
disconnected from the unit.
If the vibration disappears the instant the power is shut off, the vibration is likely to be due to electrical
problems. If this is the case conventional electrical testing procedures can be carried out to pinpoint the
true cause of vibration. On the other hand, if the vibration amplitude decreases only gradually after power
is disconnected, the problem is more likely to be mechanical in nature. Perhaps an even better indication of
the contribution of electrical problems is by observing the time waveform of the vibration as power is
disconnected.
Electrical problems with induction motors will often cause the motor load current ammeter to swing or
pulsate in a cyclic manner. If phase readings are taken, or if the motor has a strobe light flashed at run
speed, it will be seen that the phase is erratic and instead of the strobe “freezing” the rotor, the rotor will
appear to swing back and forth. This pulsating vibration common with induction motors will either be a
single frequency whose amplitude is being modulated or it will be a beat between two frequencies of
vibration which are very close together. If the nature of the pulsating vibration can be determined, this can
help significantly to identify the specific problem as discussed in the following paragraphs.

Armature Related Problems


Typical problems associated with the rotor or armature of an induction motor which cause electrical
vibration include:
• Broken rotor bars This type of defect is best discovered by the use of motor
current analysis and comparing the height of the slip * No. of poles sidebands around line
frequency to previous levels.
• Wound rotor windings Defects in the rotor will cause a modulation of rotor bar pass
or stator slot passing frequencies at run speed. In other words these frequencies will have
sidebands of 1x. If the machine has a “beat” at slip frequency this is usually due to a defect in
one rotor winding phase such as a broken conductor or bad brush.
• A bowed rotor This defect usually occurs on very large, horizontal motors
where the motor has sat idle for an extended period and the weight of the rotor causes a sag in
the middle of the rotor. In vibration readings this will look very like imbalance with a small
axial component. If bowing of the rotor is suspected then the condition can usually be
corrected by “slow rolling” the motor for up to two days to reset the sag.
• An eccentric rotor The variable air gap this produces between the rotor and stator
give a vibration at 2 x line frequency with sidebands at pole pass frequency as well as
sidebands around run speed of pole passing frequency.

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

Stator Related Problems


Electrical problems in the stator of an induction motor can also result in vibration with a pulsating
amplitude. However, in this case the pulsation is the result of a beat between two separate frequencies of
vibration which are very close together. Common stator related problems which can be expected include:
• Stator windings Phasing problems such as a loose connector in the stator
windings or supply will cause a modulation or sideband of the reciprocal of the number of
pairs of poles x line frequency (e.g. 60 x 1/3 for a 1200 rpm motor) around 2 x line frequency.
A defect in the windings themselves will also cause an unequal magnetic flux around the
motor although is usually very small except in severe cases.
• Windings insulation Gradual degradation of the insulation of the windings will result in a
fall off of the insulation to ground resistance which should be at least 1.5 MΩ for a main drive
motor.
• Imbalanced phases A difference in the supply voltage or power factor of the three
phases will cause a vibration of 2 x line frequency (120 Hz) around rotor bar passing
frequency. If demodulated spectra are used this will show up as very high spikes at 120 Hz
(100 Hz in Europe) in the frequency domain.
In the case of a stator related problem, to produce a vibration whose amplitude pulsates in a cyclic fashion it
is necessary that two frequencies of vibration be present. One of these vibration frequencies may be the
result of some imbalance or misalignment occurring at the running speed of the armature. The other
vibration will probably be an electrical vibration which occurs at the rotating speed of the magnetic field
powering the motor. If any of the above stator problems are encountered a mechanical vibration will occur
at the rotating speed of the magnetic field. Since the mechanical and electrical vibrations are relatively
close in frequency their amplitudes will alternately add together and subtract at a rate equal to the difference
between their frequencies. The result will be noticeable steady pulsation or “beat” of the vibration
amplitude.
Observing the pulsating vibration in time waveform on an oscilloscope or spectrum analyzer can be useful
in identifying the beat frequency characteristics of stator related induction motor problems. The phase
relationship between the two individual vibration frequencies is constantly changing producing a resultant
vibration whose amplitude increases and decreases in a periodic fashion.
Electric motors have inherent vibration due to “torque pulses”. Torque pulses are generated as the rotating
magnetic field of the motor energizes the stator poles. Since each motor pole is essentially energized twice
for each cycle of AC current, the vibration resulting from torque pulses will be two times the line frequency
powering the motor. Thus, if an AC line frequency is 60 Hz or 3600 cpm, torque pulse frequency will be
120 Hz or 7200 cpm. This vibration is rarely troublesome except where extremely low vibration levels are
required, or if the torque pulses should happen to excite a resonance condition in the machine or structure.
Torque pulses have also been known to excite loose rotor bars and loose stator windings at frequencies of
2x, 3x, and even 4x torque pulse frequency. Care should be taken if a significant 7,200 cpm vibration is
seen in velocity vibration readings as this can also be caused by stator distortion brought on by a severe
misalignment or a “soft foot” condition.
Demodulated readings will demodulate the frequencies above the high pass filter or inside the band pass
filter. This range will usually also include the rotor bar pass and slot pass frequencies. A modulation of the
rotor bar or slot passing frequencies by 2 x line frequency is not uncommon and does not necessarily mean
that there is a defect in the motor. It is the author’s experience that inequalities in the power factor at each
of the three phases will cause very large changes in this modulation. If frequencies at 120 Hz (7,200 cpm)
are apparent in your demodulated spectrum, the first thing to check is the individual phase power factor.

33
Vibration Analysis on Rotating Equipment

Broken Rotor Bars


AC induction motors experience a wide range of mechanical problems common to most machinery such as
misalignment, looseness, bad bearings etc. However, they also have their own unique set of problems that
are often linked to the electro-magnetically generated fields in the stator and the rotor. Frequency analysis
of the motor load current has been consistently proven to be able to detect the presence of broken rotor bars,
end ring resistances and cast rotor blow holes. Each of these three problems give the same effect to the
motor - variations in current draw, as the defective bar cuts the lines of flux. In the 1960’s Aberdeen
University in Scotland carried out a joint research project with Shell Exploration & Production on several
off-shore oil and gas production facilities in the North Sea. The object of the project was to determine a
reliable and repeatable method for the evaluation of broken rotor bars on A.C. induction motors. The
driving force for this project was the large number of motor failures suffered at these production facilities.
The project concluded that the presence of sidebands at the motor slip multiplied by the number of poles
around the line supply frequency indicated not only the presence of broken rotor bars (or equivalent) but
also how many bars were affected. The quantitative analysis of the number of broken rotor bars relies
heavily on the height of the sideband compared to the height of the line frequency spike (generally
expressed in dB Amps) at a fixed speed and a steady load of at least 50% during the measurement. To
identify the separate spikes clearly the spectrum must have a fine line resolution. These measurements are
usually taken with 1600 or 3200 lines with a bandwidth of 80 Hz (for 60 Hz line supplies) or 65 Hz (for 50
Hz line supplies). A 3200 line spectrum with a bandwidth of 80 Hz will take 40 seconds to collect during
which time the load and speed of the motor under test must not vary significantly.

DC Motors

How DC Power Is Created.


Direct current is created by taking three phase alternating current and converting it through a bank of silicon
controlled rectifiers (SCRs) into direct current. AC is supplied in the United States at 60 cycles per seconds
or 60 Hz (Hertz). Industrial power in the United States is supplied in a three phase 60.Hz format. This
means there are three individual alternating current waveforms being supplied simultaneously at 60 Hz but
120 degrees out of phase. The three waveforms are identical in amplitude and duration; thus, a one second
snapshot of three phase AC will reveal 180 positive and 180 negative amplitude peaks
When AC is rectified to DC these peaks are electronically processed to allow only positive peaks to remain
in the wave form. These peaks, although no longer alternating, create a pulsing which is detectable through
vibration analysis.
When three SCRs (half-wave rectified) are used to convert AC to DC then a pulsing or frequency equal to
the 180 Hz is created in the DC drive System. When six SCRs are used to convert the AC to DC, (full-
wave rectified or High Efficiency System) then a pulsing or frequency of 360 Hz (6 x 60 Hz) is created in
the DC drive system.
Those who are more comfortable using cpm (Cycles Per Minute), rather than Hz, need only multiply the
frequency in Hz by 60 seconds.
Thus 180 Hz x 60 sec. = 10,800 cpm and 360 Hz x 60 sec. = 21,600 cpm.

34
Vibration Analysis on Rotating Equipment

Figure 25 The creation of DC power

Single phase alternating current frequency is 60 Hz

or 60 Hz x 60 see. = 3,600 cpm

• Three phase alternating current frequency is 180 Hz or 3 x 60 Hz x 60 sec. = 10,800


cpm and also has 180 positive and 180 negative amplitude peaks per second.
• Half-wave rectified direct current (3 SCRs) is 3 x 60 Hz @ 180 Hz or 180 Hz x 60
sec. @ 10,800 cpm
• Full-wave rectified direct current ( 6 SCRs ) is 6 x 60 Hz = 360 Hz or 360 Hz x 60
sec. = 21,600 cpm

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

Figure 26 FFT spectrum of half wave rectification

Figure 27 FFT spectrum of full wave rectification

DC Systems and Controls


Direct current drive systems use rectified alternating current to power an electric motor. This DC source
can be varied through system controls to change the running speed of the motor. This can be controlled
manually or be adjusted automatically by allowing the control system to monitor the motor speed through
the use of a tachometer, self adjusting the power source to achieve the desired speed. The self controlled
systems or closed loop systems use low voltage control components to fire or open the pathways which
allow the full DC power to be supplied to the motor. This allows the motor to run at the desired speed or

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

Figure 28 Basic DC system circuit

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.

DC Control Firing Cards


The firing cards, which control the opening and closing of the SCRs, must perform correctly to allow the
system to function normally. Because there is one firing order per SCR there is a potential for one or more
of the SCRs to perform erratically or not at all if the firing card malfunctions.
Vibration analysis has been used to determine malfunction on this card or SCR’s based on frequencies
which have sub multiples of the DC dominant frequency present. DC power frequencies will always be
constant at 21,600 cpm on a full-wave system or 10,800 cpm on a half-wave system. If one sees a
frequency of 3,600 cpm and frequency separation of 3,600 cpm between existing spectrum peaks or a
frequency of 7,200 cpm and frequency differences of 7,200 cpm then faulty firing cards or SCRs may be the
cause.

37
Vibration Analysis on Rotating Equipment

1
S.C.R. problems

Figure 29 FFT spectrum full wave DC firing card frequencies

Figure 30 FFT spectrum after repair

Example of a Firing Card Fault


Vibration analysis was requested on a 300 HP DC motor which was critical to a production system. One
day prior to the request for analysis the system would not maintain the expected speed and the motor was
pulsing. A mechanical problem within the gearbox or drive train was initially suspected because the
electrical system appeared to be functioning normally. The motor was uncoupled from the drive train and a
visual inspection performed on the motor when run under a no load condition. The pulsing was still
apparent even under a no load condition, however, the motor did start and run. A motor bearing then
seemed the next logical failure point. The production system was critical to the plant operation, as
previously mentioned, and a decision to recouple the motor and operate until a vibration analysis could be
performed to determine bearing wear or failure seemed the best alternative. The analysis was performed

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.

Figure 31 FFT spectrum showing half wave firing card frequencies

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

Figure 32 FFT spectrum of same motor (no load)

DC Control Comparitor Card


The comparitor card is another low voltage control component which is responsible for determining the
difference between the system actual speed and the set or predetermined speed performance. When this
component malfunctions it has been observed that there are side-bands present around DC frequencies.
These side-bands are not of a particular set frequency but are always equally spaced from the DC
frequencies. it has also been observed that these side-bands will grow or diminish as the motor RPM is
varied, however, they will remain equally spaced. It has not yet been determined if these side-bands are
related to the RPM fluctuation or hunting which often accompanies comparitor card problems or if the
constant collapsing and regenerating of the magnetic field of a system that is hunting is the cause. The side-
bands do, however, exist regardless of the cause and should be considered a warning of this component
failure. These side-bands may occur at small cpm increments as shown below and may require a high
resolution FFT to differentiate them from the dominant frequencies.
To resolve side bands related to comparitor card malfunctions a FFT spectrum of 3200 lines of resolution at
Bandwidth of 24,000 cpm is suggested.
Another possible reason to see these FFT characteristics could be a faulty or malfunctioning tachometer
which would corrupt the voltage going to the card. Testing the voltage output of the tach. should confirm
this situation.

40
Vibration Analysis on Rotating Equipment

0.0872

5000 16,000

Figure 33 FFT spectrum showing comparitor card defect.

Example of a Comparitor Card Defect


Vibration analysis was requested on a 125 HP DC motor which was thought to be vibrating heavily. The
motor had been uncoupled from the belt to see if the motor vibrated when running solo.
The first set of spectra was collected under these conditions and revealed the above spectrum. A RPM
check using a digital tach. also revealed that the motor was fluctuating or hunting approximately 30 cpm at
an RPM of 1440. The tach. voltage was then checked but seemed to be consistent with the operating speed
and fluctuations. Based on the side-bands a recommendation was made to change the comparitor card (also
called the control card). There was not a spare card available at this time but there were spare firing cards.
The firing cards were changed but did not solve the problem. A spare comparitor card was eventually
located and the replacement accomplished. The speed fluctuations stopped and another set of spectra
collected revealing the spectrum below.
DC motors are different from AC motors because of their power supply which requires different
components. The most obvious of these is the tachometer which extends, usually, from the back of the
motor. These units usually have small bearings which can be monitored in the same manner as any bearing.
Caution: Never Place A Magnetic Accelerometer Mount On A Tachometer
These devices use magnets to generate the voltage which the control system monitors to determine the
motor speed. Placing a powerful magnet on or near the tach may alter or destroy the voltage output causing
the motor to literally speed up until it destroys itself.

41
Vibration Analysis on Rotating Equipment

Figure 34 FFT after the comparitor card was replaced

Figure 35 DC motor components

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.

Importance of Exact RPM


Knowing the exact RPM of the variable speed system which is being analyzed is critical to determining
which system component is creating the vibration frequency of interest. Never assume that the RPM of the
system is what the panel display reads, these displays are seldom accurate. The use of either a good strobe
light, or a digital tachometer will give an accurate RPM. Ensure that your data always has the RPM stored
with it to prevent mistakes when looking at your data for analysis.

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.

Example of Mechanical -v- Electrical Frequencies


Vibration analysis was requested on a 5 HP DC motor which was causing problems on a production line.
The motor is one of four motors which must work in sequence for the line to run efficiently. The spectra
was collected revealing the FFT in figure 33, which had what was initially suspected to be DC electrical
frequencies with a 3,600 cpm separation. A quick reference reading of G/SE (high frequency pulse
detection) was taken and registered a higher than normal value, this is not characteristic of an electrical
problem.
The motor RPM was reduced by 25% from 1004 RPM to 756 RPM and new spectra collected. The
frequency separation was now 2,700 cpm, a 25% difference from 3,600 cpm. The frequency separation
followed the RPM proportionately indicating a mechanical problem.
The motor had a SKF 6208 bearing which has a 3.606 BPFO multiplier (1004 RPM x 3.606 = 3620 cpm).
What was assumed to be a 3600 cpm separation was due to a bearing outer race defect.

Figure 36 FFT from a 5 HP motor - full wave rectified

43
Vibration Analysis on Rotating Equipment

Figure 37 Same motor - speed lowered by 25%

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

then the out of balance forces


= [0.5 x (1000/60⋅2π)2 x 0.3]/9.81
= [0.5 x 104.722 x 0.3]/9.81
= [0.5 x 10,966x 0.3]/9.81
= 167.68 kgF (369 lb.F)
In other words we have added over an eighth of a ton to the apparent weight of the impeller. Let us see
what happens if the impeller is running at 2000 rpm.
= [0.5 x (2000/60.2π)2 x 0.3]/9.81
= 670.71 kgF. (1476 lb.F)
By doubling the speed to 2000 rpm we have quadrupled our out of balance forces to almost three-quarters
of a ton force.
A deep groove Conrad bearing for a shaft of about 3 inches diameter could be a MRC 215-S, which is in the
MRC mid range of bearings. This bearing has a static load rating of 23,680 lb. at 1,000 rpm as calculated
using the 3AFBMA method of evaluating load ratings. The speed factor for calculating radial load at 2,000
rpm is 0.7937 so our load is now
3,680 x 0.7937 = 2,920 lb.
By increasing the speed to 2,000 rpm and allowing an imbalance force of about 1 lb at about 1 foot away
from center we have halved the effective load carrying capacity of the bearing.

2
Data taken from TRW service catalog Form 382-14
3
Anti Friction Bearing Manufacturers Association

45
Vibration Analysis on Rotating Equipment

Vibration due to imbalance

Figure 38 Imbalance slide 1

Figure 39 Imbalance slide 2

46
Vibration Analysis on Rotating Equipment

Figure 40 Imbalance slide 3

Figure 41 Imbalance slide 4

47
Vibration Analysis on Rotating Equipment

Figure 42 Imbalance slide 5

Figure 43 Imbalance slide 6

48
Vibration Analysis on Rotating Equipment

Figure 44 Imbalance slide 7

Figure 45 Imbalance slide 8

49
Vibration Analysis on Rotating Equipment

Figure 46 Imbalance slide 9

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

Figure 47 Misalignment slide 1

Figure 48 Misalignment slide 2

51
Vibration Analysis on Rotating Equipment

Figure 49 Misalignment slide 3

Figure 50 Misalignment slide 4

52
Vibration Analysis on Rotating Equipment

Figure 51 Misalignment slide 5

53
Vibration Analysis on Rotating Equipment

Looseness

Figure 52 Looseness slide 1

Figure 53 Looseness slide 2

54
Vibration Analysis on Rotating Equipment

Figure 54 Looseness slide 3

Figure 55 Looseness slide 4

55
Vibration Analysis on Rotating Equipment

Figure 56 Looseness slide 5

Figure 57 Looseness slide 6

56
Vibration Analysis on Rotating Equipment

Figure 58 Looseness slide 7

Figure 59 Looseness slide 8

57
Vibration Analysis on Rotating Equipment

Figure 60 Looseness slide 9

58
Vibration Analysis on Rotating Equipment

Vibration due to aerodynamic forces


Fans and blowers will inherently have some vibration due to aerodynamic forces. This vibration results
from the fan blades striking the air and will occur at a frequency equal to the number of fan blades times fan
R.PM. Normally, the amplitudes of vibration resulting from aerodynamic forces will be low and no cause
for concern. When excessive vibration at the aerodynamic frequency is encountered a common cause is
resonance of some part of the machine or structure and the checks for resonance, described near the end of
the manual should be carried out to determine the resonant part. If it is confirmed that a condition of
excessive aerodynamic vibration is not due to resonance, the fan should be checked carefully for
obstructions that may disturb the smooth flow of air through the fan. For example, high amplitudes of
aerodynamic vibration are sometimes encountered on cooling tower fans. Many of these fans consist of a
drive motor, mounted outside the fan venturi, coupled to the fan gearbox by means of a long torque tube or
drive shaft. The torque tube can act as an obstruction to the smooth flow of air through the fan and an
aerodynamic pulsation is generated each time a fan blade passes over the torque tube. The result is often
excessive vibration at the aerodynamic frequency and may require that the distance between the blade path
and torque tube be increased to minimize these aerodynamic pulsations.
On centrifugal fans, excessive vibration at the aerodynamic frequency can sometimes result if the fan rotor
is positioned eccentrically in -the f an housing. Therefore this should be checked in the event that the
problem cannot be traced to resonance.
Vibration due to aerodynamic forces can also occur at a frequency equal to 1 x fan R.PM and will appear
similar to normal imbalance. This aerodynamic imbalance will result if the fan blades do not have the same
track or pitch. If the fan operates under a constant aerodynamic load, the force of aerodynamic imbalance
can be compensated by following normal balancing procedures. However, it often occurs that changing the
fan load will produce a corresponding change in the vibration at 1 x R.PM.
For example, a centrifugal fan was balanced with the access doors in the fan housing removed to simplify
the addition and movements of trial weights. After the fan had been satisfactorily balanced, the access
doors were replaced. Unfortunately, however, this produced a significant change in aerodynamic conditions
and the result was a significant increase in vibration. In this case, it was necessary to balance the fan
operating under its normal aerodynamic conditions. If a fan must operate smoothly over a broad range of
aerodynamic loads it may be necessary to check and correct for significant variations in blade track or pitch
before this -can be achieved.

59
Vibration Analysis on Rotating Equipment

Figure 61 Aerodynamic forces

Aerodynamic cross coupling


Aerodynamic cross coupling is a problem occasionally encountered on turbines, centrifugal compressors
and fans operating above the first rotor critical and generally results from eccentric rotation of the rotor
caused by rotor bow or deflection. In the case of a centrifugal compressor the layer of air or other gas being
compressed will have a rotating speed less than that of the rotor, similar to the rotating oil whirl in a plane
bearing. If the rotor is bowed slightly, the layer of rotating gas between the rotor and the machine housing
will produce a torque reaction on the rotor causing the rotor to whirl at the rotating speed of the gas layer.
The frequency at which this whirl occurs can vary from one machine to the next-
The vibration may have the same frequency characteristics of oil whirl and hysteresis whirl where the
lowest natural frequency of the rotor bearing system is excited. In most cases, the vibration frequency will
be less than the rotating frequency. Cases have been reported where sub-multiples of 0.5, 0.33 or perhaps
0.25 times the rpm of the rotor have been excited.
Since rotor whirl generated by aerodynamic cross coupling is excited by -the compressed air or gas, it seems
logical that the condition would be affected by machinery load. In general, the machine will be more likely
to experience this condition under heavily loaded conditions and changing the load of the machine to
determine its effect on the vibration can be useful in -diagnosing this problem.

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.

Choking or Stone Walling


The problem of 'choking or 'stone-walling' in a compressor is essentially the opposite of surging but again is
the result of attempting to operate the unit outside design parameters. Choking occurs when discharge
pressures are too low. When discharge pressures are low, velocities are high and when flow velocity in the
diffuser section approaches Mach 1 a turbulent or circulating flow between the blades will occur which has
the effect of blocking the flow of gas. When this occurs a noticeable drop in efficiency and pressure ratio
can be seen, along with an increase in vibration due to the turbulent flow within the compressor. The
vibration characteristics of choking are essentially the same as those encountered during surging. A check
of other operating parameters such as pressure, temperature, flow etc. should be undertaken to distinguish
between the two.

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.

Elasto Hydrodynamic Lubrication


Let us consider the lubrication of a rolling element bearing. As the roller rotates the pressure point is very
small so the pressure loading is very high. So high, in fact, that the contact point of the roller and the race
(the contact ellipse) becomes elastically deformed, trapping a very small amount of the lubricant into a
“wedge”. The lubricant wedge itself, is very small but is large enough to keep the roller physically
separated from the race by a small distance (in the micron range). THERE IS NO METAL TO METAL
CONTACT IN A PROPERLY LUBRICATED BEARING RUNNING AT NORMAL LOAD.
As the lubrication mechanism starts to break down the lubricant loses its ability to separate the roller from
the race. This may be due to a deterioration of the lubricity, an increase in bearing load, overheating of the
bearing or a fatigue failure of an old bearing at the area most prone to elastic deformation.

Figure 62 Elasto-hydrodynamic lubrication

Rolling element bearings have four stages of failure:


• Stage 1 Lubrication problems
• Stage 2 Marking of the raceways and / or rolling elements
• Stage 3 Spalling of the raceways and / or rolling elements
• Stage 4 Collapse of the bearing

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

First Stage of Bearing Failure


Lubrication problems in the case of grease or oil systems is not necessarily the physical loss of lubricant but
the loss of the oil or grease’s properties. Since the life of a grease lubricant is strongly temperature
dependent and since grease loses half of its life for every 20°F rise in temperature, this can be seen to have a
very significant effect on bearing health. In order to avoid the temperature effect due to over packing,
bearing manufacturers usually suggest that bearings are packed with grease to between 15% to 20% of the
bearing's free volume. We know that over packing is the most common cause of raised bearing
temperature, which leads to reduction in grease life and eventual failure. Under-packed bearings, or
bearings which have lost grease due to physical migration, may generate high bearing temperatures when
running at high speeds.
Low speed, starved bearings usually wear into a condition of excessive looseness and fail without
appreciable temperature increases. While this is fairly common on lightly loaded motor bearings, press
main drive motors will usually fail catastrophically.

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

Figure 64 Loss of Lubricant - Roller Bearing Courtesy of the Torrington Company

Second Stage of Bearing Failure


This the early stage of fatigue. This fatigue may be initiated on the surface or beneath the surface. Surface
fatigue is usually caused by scratches on races, balls, or rollers, abrasive contamination, or brinelling.
These marks produce “stress raisers”, a point on the bearing surface that experiences abnormally high stress
due to the physical conditions at that spot. Simply, a given load over a given area produces stress. If a
crack or contaminant is found at that location, the load is distributed over a different (often smaller) area
and therefore greatly increases stress at that point. This phenomenon limits the number of cycles a bearing
can survive. These raised stress areas provide a start point for micro-crack formation that leads eventually
to pitting, spalling, and wear.
Subsurface fatigue is usually caused by voids, foreign matter or coarse carbides introduced into the material
at the time of formation. These material anomalies again provide for a point of crack formation if they fall
within a high stress area. and once a crack is formed beneath the surface, it works its way outward and
eventually develops into a spall.
As the crack reaches the surface it creates a small void into which the lubricant wedge collapses.
Remember that the lubricant wedge is microscopically small so even a tiny crack in the material can cause
the roller to impact heavily onto the race. As each roller passes the void it impacts onto the race. As with
the first stage, the impacts generate vibration at resonant and ultra sonic frequencies, with the difference,
however, that these frequencies are now modulated by the rate at which the rollers hit the defect. In other
words the resonant frequencies are excited every impact, giving a vibration like the pattern below.
Notice the “roller coaster” in Figure 64 at just below 2 kHz. This is a bearing resonant frequency and the
rise and fall in amplitude coincides with the impacts from the rollers on the outer race defect (in this case).
At this stage of the failure there is no appreciable rise in temperature and the velocity vibration at the defect
frequencies is insignificant.

64
Vibration Analysis on Rotating Equipment

Figure 65 Waterfall plot from a damaged motor bearing

Third Stage of Bearing Failure


Fatigue failure or spalling results from mechanical materialogical failure of the bearing. Literally a stress
related failure of the material which results from cyclic stresses due to operation at high loads. As the
rollers repeatedly impact onto the small crack, the very small area produces very high point loading so the
material starts to flake off around the crack. As the crack propagates, more and more material is removed
until the crack becomes a visible cavity or spall.
At this stage of the bearing failure, the velocity vibration becomes apparent at the defect frequencies and
harmonics, possible also with sidebands of run speed and / or cage frequency. The resonant vibration has
also increased in amplitude at the defect frequencies along with a general rise in floor level. Temperature
will be elevated above normal but not significantly, particularly not if there is any air movement around the
surface of the bearing or bearing housing.

65
Vibration Analysis on Rotating Equipment

Figure 66 Early Fatigue - Ball Bearing Courtesy of the Barden Corporation

Fourth Stage of Bearing Failure


With time both surface and subsurface fatigue flaws spread over the active bearing surfaces causing bearing
wear, growth in spalls, and eventual machine failure. The metal contaminants or wear particles removed
from the bearing during spalling are either washed out with the oil, in oil-lubricated bearings, or are trapped
in the bearing, as is common in sealed and grease-packed bearings. In these latter bearings, continuous
recirculation of the particles causes progressively higher wear to the point where either the bearing becomes
excessively loose and fails to support the load suitably or the induced damage leads to failure.

Figure 67 Developed Fatigue on Roller Bearing Courtesy of the Torrington


Company

66
Vibration Analysis on Rotating Equipment

Bearing Defect Frequency Calculation


It is important to understand that as a bearing with a damaged surface rotates, the regularity, or frequency
with which the roller or ball impacts on the defect indicates potential failure and allows us to determine the
type of damage that exists. A number of characteristic frequencies are generated by a damaged bearing ,
and are known as:-
• Cage or fault train frequency = FTF
• Ball pass with respect to the outer race = BPFO
• Ball pass with respect to the inner race = BPFI
• Ball rolling about its own axes = BSF
• Shaft frequency of rotation = RPM

The following equations are used to calculate these frequencies


FTF = RPM/2 [1 - (BD/PD) cos θ] (rotating inner race)
approximated by RPM * 0.45 (rotating inner race)
or RPM * 0.55 (rotating outer race)
BPFO = n * RPM/2 [1 - (BD/PD) cos θ]
approximated by RPM * O.4* n
BPFI = n * RPM/2 [1 + (BD/PD) cos θ]
approximated by RPM * O.6 * n
BSF = (RPM * PD)/2BD * [1 - (BD/PD)2 COS2 θ]
(no valid approximation)

where BD = ball (roller) diameter


PD = pitch diameter
θ = contact angle
n = number of balls (rollers).
When bearing geometry is not known but the number of balls or rollers can be counted or estimated, it is
suggested that the approximate equations be used to establish the bearing frequencies of interest. All the
equations listed above show a direct dependence of the calculated frequency on the frequency of rotation..
The following figure illustrates the bearing geometry used in the above equations

67
Vibration Analysis on Rotating Equipment

Figure 68 Ball Bearing Terminology

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

Figure 70 Bearing damage severity assessment chart

Analysis of bearing defects

Figure 71 Demodulated acceleration spectrum from a dry bearing

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).

Figure 72 Demodulated acceleration spectrum of a marked bearing

71
Vibration Analysis on Rotating Equipment

0.872

Figure 73 Demodulated acceleration spectrum from a slightly more heavily marked


bearing
Figures 71 and 72 show two examples of rolling element bearings which have suffered some marking of the
races (both of these examples show marking of the stationary outer race). Note that the fundamental
frequency has several harmonics but it is the fundamental frequency which will coincide with the generated
bearing defect frequency and help us in our analysis. At this stage of the bearing deterioration it is
sometimes still possible to save the bearing with additional lubrication. At the very least you will extend
the life of the bearing.

1.000

1.000

0 Time Waveform mS 160

Figure 74 Time waveform from a marked bearing.

72
Vibration Analysis on Rotating Equipment

2.500

2.500

0 Time Waveform mS 160

Figure 75 Time waveform from a heavily marked bearing

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

Figure 76 Velocity spectrum from a spalled bearing

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.

Figure 77 Sources of imbalance


Imbalance is the unequal distribution of the weight of a part about its rotating centreline. Illustrated
are some reasons for an unequal weight distribution. The blow hole in the casting, the thicker
web, the eccentric hole location and eccentric machining of the pulley groove all add up to more
weight on one side of the rotating centreline than the other. The faults shown are exaggerated but
could easily exist in almost every rotating part manufactured.

75
Vibration Analysis on Rotating Equipment

Figure 78 Assembly tolerance stack up


More important as a source of imbalance is the stack up of tolerances possible when assembling
rotating parts. The example illustrated is a very common one. The hole in the pulley is necessarily
larger than the shaft diameter and when a key or set screw is attached the take-up in clearance
shifts the weight of the pulley to one side of the rotating centreline of the shaft. Thus the assembly
is out of balance.

Figure 79 Heavy spot


All of the combinations of unequal weight distribution can be considered to be concentrated at one
point called the heavy spot. This heavy spot represents the accumulated results of all of the
imbalance of the pulley. An equal amount of weight at the same radius but opposite the heavy
spot will balance the rotor.

76
Vibration Analysis on Rotating Equipment

Figure 80 Units of measure of imbalance


The units for measuring imbalance are generally ounce-inches, gram-inches or, more commonly
now, gram-centimetre. One ounce-inch of imbalance would be a heavy spot of one ounce in
weight one inch from the rotating centreline. Three ounces located three inches from the centre
would be nine ounce- inches.

Figure 81 Mass centre displacement


The imbalance of a part may also be given in terms of the distance from the rotating centreline to
the mass centre of the part. The mass centre of a part is more commonly known as the centre of
gravity or C.G. The centre of gravity or mass centre is the point about which all of the weight of a
body is equally distributed.
The example illustrated shows a condition of mass centre displacement of two thousandths of an
inch. Such a condition could very easily be duplicated by installing a 100- ounce pulley on a shaft
with a two-thousandths clearance between the hole and the shaft. The resulting imbalance is two-
tenths of an ounce-inch.

77
Vibration Analysis on Rotating Equipment

Figure 82 Force due to imbalance


The effect of imbalance is that it generates a force which causes the part and the supporting
bearing and structure to vibrate. The size of the force generated depends on the speed of rotation
and the amount of imbalance. In the illustration the part has an imbalance represented by a heavy
spot located some distance from the centre of rotation. If the heavy spot and distance are in
ounces and inches, the force generated can be computed by the formula

F = 1.77 (RPM/1000)2 ounce-inch.

“F'' is the force in pounds',


RPM, the rotating speed of the part and
ounce-inch the amount of imbalance.
1.77 is a constant required to make the formula dimensionally correct.

When the imbalance is given in terms of gram-inches the formula becomes

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.

Vibration Related to Imbalance


Balancing is the process by which we learn the amount 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.
We have learned that the more imbalance we have the greater the force, therefore, the greater
the vibration. So we will use the amount of vibration to indicate how much imbalance we have. We
will use the position of the reference mark as seen by the strobe light to tell us the location of the
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

Single Plane Vector Method of Balancing


This question can best be answered by making what we call a vector diagram. A vector is simply
a line whose length represents the size of the imbalance and whose direction represents the
angle of the imbalance. When a trial weight is added to the part we actually add to the original
imbalance in the part and change the position of the total imbalance to some new position
between the two. We see this as a new vibration displacement and reference mark position or
phase. Our original imbalance was represented by 1.8 mils and a phase of 12:40 o'clock. After
adding a trial weight the imbalance due to both the original plus the trial weight is represented by
2.30 mils and a phase of 2.30 o'clock.
Using polar coordinate graph paper we plot our original imbalance vector. We draw a line from the
centre at the same angle we saw our reference mark, 12: 40 o'clock. Each hour is equal to 30°.
We select a convenient scale for the length of the line such as 1 mil equals 2 divisions. We'll call
this vector “O” for the original imbalance. Next we draw the original plus the trial weight imbalance
vector “O + T'' to the same scale. It is drawn at the new angle for the original plus the trial weight
imbalance phase 2: 30 o'clock. This is all the information needed to solve for the effect of the trial
weight alone.

Figure 83 The vector diagram


To solve for the trial weight vector we must complete the vector parallelogram. To complete the
vector parallelogram we first connect the end of the original imbalance vector to the end of the
original plus trial weight imbalance vector, line A. Parallel to line A we draw a line outward from the
centre of the graph, line B. Then parallel to the original imbalance vector and from the end of the
original plus the trial weight vector, O + T, we draw another, line C, until it crosses line B. Line B
then represents the vector for the trial weight. The length of line B represents the amount of the
trial weight. The direction in which the trial weight acts with respect to the original imbalance is
represented by the direction of line B. In order to balance the part we want the trial weight to be
equal and opposite to the original imbalance as shown by the dotted line. We can see that the trial
weight must be moved by an angle equal to the angle between the trial weight vector and the
place where we want it to be. The direction we must move the weight, however, is not clockwise
as shown here but is counterclockwise. Remember our experiment where we learned that the
reference mark moves in the opposite direction to that of the heavy spot. Therefore, we should
adopt this rule. When the reference mark after adding a trial weight moves clockwise from its

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.

Figure 84 Simplified vector diagram


Our vector method of balancing then becomes quite simple. We draw the “O'' vector then the O +
T vector, connect the ends of the two, and this becomes the “T'' vector or Trial Weight Vector, or
line A in the more complex vector diagram shown above. To solve this problem measure the
lengths of line “O'' and line ”T'', measure the angle between “O'' and “T''. The correct balance
weight is equal to the trial weight x O/T. The balance weight should be added at an angle equal to
the angle between “0'' and “T'' away from where the trial weight was added. The direction of this
angle is opposite from the shift of the reference mark on the trial run. By following these
instructions carefully the part should now be balanced by moving the trial weight 75 degrees
counterclockwise and reduce its size. However, very small errors in measuring the phase angle, in
shifting the weight, and adjusting the weight to the proper amount can result in some remaining
vibration still due to imbalance.

81
Vibration Analysis on Rotating Equipment

Figure 85 Additional corrections


To make further corrections we simply observe the new amount and phase of vibration, .5 mils at
9:00 o'clock. We must plot this reading as a new “O + T'' (shown above). We then connect the
ends of “O'' and the new “O'' + “T'' vector to find the new “T'' We can see that we shifted the trial
weight too far and must move it back by the angle between “O'' and the new “T'' This may be
repeated as many times as necessary but always using the original “O'' line. This then is the
vector method for single plane balancing. It is simple to use and provides accurate information to
balance a part in the fewest number of starts and stops of the machine as possible.
Four-step Method of Balancing Single Plane
Another method for balancing which is not as precise as the vector method follows the same
basic procedure except we do not construct a vector diagram. The four- step method follows a
few simple rules to find the proper location for the correction weight after which the amount is
adjusted to balance the part. Since it generally requires many starts and stops of the machine it is
not too popular except when the number of balancing runs is not important.
First we observe the amount of vibration and the phase or position of the reference mark when
the strobe light flashes. In the illustration these readings are 2.4 mils at 5: 00 o'clock. This is the
original run. AII future data will be referred to these readings just as we did in the vector method.
Next we add a trial weight to the part at any convenient location and again observe the amount
and phase; 1.8 mils at 7:00 o'clock.

82
Vibration Analysis on Rotating Equipment

Figure 86 Direction to shift the weight


Our first goal is to shift the trial weight to a position where the reference mark returns to its original
position or 180° away. We do this by shifting the trial weight in a direction opposite the shift of the
reference mark. Do not be afraid to move the trial weight by a large angle because if you move it
too far, the phase will direct you to move it back. The reference mark shifted from 5: 00 o'clock to
7: 00 o'clock after we added the trial. This is a clockwise shift of the reference mark. Therefore,
we move the trial weight in the counterclockwise direction.
A new reading might be 2.4 mils at 3:û0 o'clock (not illustrated) . This would indicate we moved
the weight too far because the mark is counterclockwise from the first position of 5: 00 o'clock.
Shifting the weight again we see the reference mark has been returned to its original position, and
the vibration has been reduced to 1 mil. This means the weight is in the proper position and we
need only to increase the size of the trial weight. If the reference mark had appeared 180° away
from its original position the weight would have been in the proper place but too large. If the
reference mark had returned to its original position but with an increased vibration then the trial
weight would have been on the heavy spot.
It should be apparent that the vector method and the four-step method both follow the same basic
rules. Know and understand both of these methods well.
Sample Problems
Following are some examples of balancing problems.
You are given the original readings and the readings after a trial weight has been added. See if
you can solve each problem correctly. Your answer should indicate first the direction to move the
trial weight, second the angle, and third, the adjustment required to get the correct size of the trial
weight

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

Figure 87 Sample problem vector diagram


Discussion:
The vector diagram was constructed as shown above.
The “O'' vector was scaled off to five units long representing five mils vibration.
The “O'' + “T'' vector was scaled off to the same scale as the “O'' vector. The ends of the “O” and
“O” + ''T'' vector were connected. This is the ''T'' vector.
The angle between the “O'' vector and the “T'' vector is the angle the trial weight must be moved
from its position for the trial run.
The length of the “T'' vector was measured using the same scale used for the “O'' and “O'' + ''T''
vector The effect of the trial weight is smaller than the original imbalance because the “T'' vector is
shorter than the “O'' vector. The amount the trial weight must be increased was calculated using
formula
Corr. Wt. = T.W. x O/T.

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

3. Original Reading 9.0 mils @ 10: 30 o'clock


First Trial Reading 4.5 mils @ 4: 30 o'clock

4. Original Reading 18.0 mils @ 9: 00 o'clock

84
Vibration Analysis on Rotating Equipment

First Trial Reading 3.0 mils @ 9: 00 o'clock

5. Original Reading 9.0 mils @ 2: 00 o'clock


First Trial Reading 1.5 mils @ 9: 00 o'clock

6. Original Reading 9.0 mils @ 2: 00 o'clock .


First Trial Reading 9.0 mils @ 12: 00 o'clock

Determine the corrections required.

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

Figure 88 Determining the flash angle


The angle between the point where the pickup is applied and the position of the heavy spot when
the strobe light flashes is called the Flash angle for the system.
Therefore, for this system the heavy spot will always appear at 11: 00 o'clock or 30 degrees
counterclockwise from the pickup when the strobe light flashes. The heavy spot can be at any
angle around the pulley but it will always appear at 11: 00 o'clock when the strobe light flashes.
The reference mark has nothing to do with the relationship since it can be placed anywhere on the
pulley. The reference mark is simply a way to see the position of the pulley when the strobe light
flashes.
After these relationships have been learned about a part then many parts may be balanced on a
production basis. These principles are also useful in two plane balancing. So that you can be sure
you understand we suggest the following experiment.
Balance a pulley mounted on a motor similar to the one we have used here. Remove the
reference mark from the pulley to avoid any confusion regarding it’s position and the heavy spot.
Next put a heavy spot on the pulley and observe its position when the strobe light flashes. Do the
same thing with the heavy spot in other locations. The location of the heavy spot with respect to
the position of the pickup when the strobe light flashes should be apparent. Now put the pickup in
a new location and repeat the experiment.

86
Vibration Analysis on Rotating Equipment

Single Channel Analysis


Taking measurements
The figure below shows a typical tap block for mounting an accelerometer to a motor. This block should be
machined with a “gramophone” finish on the reverse side to give good adhesion to the mounting surface.
The material used for the tap block should be capable of being magnetized so the tap blocks can also be
used for off line measurements with magnet and accelerometer prior to an accelerometer being permanently
fitted; for this reason the material quoted is bright steel or austenitic stainless.
The block is designed for mounting to the surface with an epoxy adhesive such as Loctite DEPEND or
similar. Whichever adhesive is used, it must set hard in order to transmit the vibration. The motor
mounting surface should be prepared with a hand grinder or similar to remove all paint and get down to
bare metal. The tap block should be mounted without the accelerometer and allowed to set firmly, holding
the tap block in place while the adhesive is setting with duct tape or similar.
Before mounting the accelerometer it is worth considering the purchase of accelerometers with local
connectors rather than those with integral cables so that if the motor has to be dismantled it is easier to
disconnect the accelerometer.

Width across flats should be Drilled & tapped


just wider than the base of
the accelerometer.
1/4” 28

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”

Figure 89 Typical tap block for mounting an accelerometer


When it comes time to mount the accelerometer, put a thin layer of grease on the accelerometer and a thin
layer of epoxy adhesive on the top of the tap block. This will help the transmission of vibration through the
tap block and allow for easy disassembly. Be very careful not to over-tighten the accelerometer onto the tap
block - follow the manufacturers torque settings. Notice that the tap block is made from hex. stock so that if
it is necessary to remove the block then it can be wrenched off with a suitable spanner. Do not mount the
accelerometer directly onto the surface of the motor with the adhesive unless you want to lose your
accelerometers every time your motor goes away for repair.

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

Max temp 250°C (482°F)

Thin layer of
bees wax

Max temp 40°C (100°F)

Methyl cyanoacrylate
cement (super glue)
Methyl cyanoacrylate
cement (super glue)

soft glue

Steel stud

Max temp 80°C (178F)

Figure 90 Accelerometer mounting techniques a-d

89
Vibration Analysis on Rotating Equipment

Thin double sided


adhesive disk
Thick double sided
adhesive disk

Double sided adhesive


disk

Max temp 95°C (200°F)

Max temp 150°C (300°F)

Hand held probe

Figure 91 Accelerometer mounting techniques e-g

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

(dB) ~ Ref. 100 Hz

+ 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 )

Figure 92 Overview of accelerometer mounting techniques

90
Vibration Analysis on Rotating Equipment

Potential Failure Analysis


A methodology for objective set up
by
Ron Frend

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.

The PFA Tree


The tree is structured in the following way:
Setup
Interval
Analysis
Parameter
Technology
External Manifestation
Failure Type
Base Cause

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

Developing a Potential Failure Analysis for Rolling Element


Bearings

Figure 93 PFA development for rolling element bearings

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:

Fmax = No. of samples


time for 3 revs x 2.56
Example. For a bearing on a shaft running at 1195 rpm the time for on rev is:
60/1195
= 0.0502 seconds
or 50.2 milliseconds (mS).
If we require a resolution of 1,024 points, the appropriate Fmax would be:
Fmax = 1024
0.0502 x 3 x 2.56
= 2,656 Hz
So we would set the Fmax to 2.5 kHz or the nearest available frequency.

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.

Including the Component Failure in the PFA Tree.


We now include all of the information from our failure study of the rolling element bearings into our
potential failure analysis tree (figure 2). Notice that the bearing failure is only one of the potential failures
that we now know of.
We have carried out potential failure analyses on all of the possible defects that we think a motor, which is
operating under these particular conditions, may eventually suffer. Not only have we considered the
possibility of bearing failure but we have included insulation breakdown, rotor bar defects, loose
components and even drive problems on D.C. motors.
The procedure for building the PFA tree is the same for any equipment or component:
• Identify the possible failure mechanisms
• Determine how the failure mechanism will show itself
• Quantify the best contemporary method for detecting the failure mechanism
• Define set up parameters to best use the detection method for this failure mechanism

Press Main Drive Motor PFA Tree


External
Base Cause Failure Type Manifestation Technology Parameter Analysis Interval Set Up

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

Time Peaks @ 12 G (roller) On Line = 3 revs


Waveform 7 G (ball brg) G acceleration

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

Figure 94 PFA for a main motor

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.

DISSOLVER FEED CHUTE FAILURE MODES

CHUTE ASSEMBLIES

LINER WEAR INSPECTION - CCTV (wieght ?)


CLAMP/SEAL FAILURE INSPECTION - CCTV
CAM SIEZURE FORIEGN BODY INGRESS INSPECTION - CCTV
CORROSION (reference marks req'd)
FRETTING/CLEARANCE INCREASE

DRIVE ASSEMBLY

GREASE SEAL FAILURE MATERIAL CHOICE DESIGN

ROTARY SIEZURE BEARING COLLAPSE LOADING DESIGN


BEARING CLEARANCE INCREASE MATERIALS / LOADING DESIGN
OVER TEMPERATURE LUBRICATION / LOADING DESIGN

DRIVE SIEZURE FORIEGN BODY INGRESS


CORROSION BEARING DEFECTS
BRINNELLING

SLIDE BUSH SIEZURE LUBRICATION / CLEARANCE DESIGN


MATERIAL COMPATIBILITY / CORROSION DESIGN

GEAR WEAR / DEFECTS DIRT/MATERIALS CHOICE NOISE / VIBRATION

INSPECTION - CCTV (reference marks req'd)


HEAD 1

96
Vibration Analysis on Rotating Equipment

BEARING DEFECT DETECTION

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

Press Flywheel PFA Tree


External
Base Cause Failure Type Manifestation Technology Parameter Analysis Interval Set Up

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

Time Peaks @ 12 G (roller) On Line = 3 revs


Waveform 7 G (ball brg) G acceleration

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

c. Ron Frend - PreDiCon

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.

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.

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.

Real Time Bandwidth and Overlap Processing


What is real time bandwidth? Simply stated, it is the frequency span whose corresponding time record
exceeds the time it takes to compute the spectrum. At this span and below, it is possible to compute the
spectra for every time record with no loss of data. The spectra are computed in "real time". At larger
spans, some data samples will be lost while the FFT computations are in progress.
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.

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.

Octave Band Center Frequencies


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 or 1/3
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)

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

If there is a significant defect raise a work request.

104
Vibration Analysis on Rotating Equipment

Severity charts

Figure 95 General severity chart for vibration

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

Two Channel Analysis

Two channel functions


Two-channel analyzers offer additional measurements such as transfer function, cross-spectrum, coherence
and orbit. These measurements are discussed below.

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

Representation by complex numbers


Sometimes the representation of the spectrum is carried out by the use of rotating vectors instead of sine
waves. For complex signals this is often much more convenient. Consider the vector below.

a a+b
ϖt
0 φ
b

Figure 96 Vector addition of 2 vibrations

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

Cascade & waterfall plots

Figure 97 Cascade of fan over 20mS

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.

Once per rev pulse trigger


As the name implies this is a simple pulse, once per rev of the shaft, which triggers data collection at a
specific time. This function is useful in cascade plots, orbits, or for Bodé plots.

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.

Figure 98 Bode plots

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

Figure 99 Orbit showing misalignment

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.

Changes in the filtered amplitude can be determined using orbits


analysis. By superimposing the phase angle input signal onto the orbit a shift in this parameter can
be easily determined. By noting the operating speed at which the resonance frequencies occur, a
change in this frequency may indicate the "possibility" of a crack existing.

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.

LOOSE ROTATING PART


A loose rotating part can generate unusual vibration signals. They may
fluctuate in amplitude and the phase angle may shift, also. This fault is
diagnosed easiest using filtered, synchronous orbit analysis. Imagine a

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.

Spectra displays of rub conditions are characterized by distinct


frequencies that occur at multiples of a fundamental frequency. The
fundamental frequency will depend upon the relationship of the
shaft speed to the first natural resonance frequency. At shaft speeds
up to twice the natural resonance frequency, the fundamental rub frequency will coincide with the
shaft speed with multiples at 2X, 3X, etc. Between twice and three times the first natural resonance
frequency, the fundamental rub frequency will be ½ shaft speed with multiples at 1X, 3/2X, 2X,
5/2X, etc. Between three and four time the natural resonance frequency, the fundamental rub
frequency will be shaft speed with multiples at 2/3X, 1X, 4/3X, 5/3X, 2X, 7/3X, etc.

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.

NATURAL FREQUENCY + EXCITING FORCE = RESONANCE

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

Equation 1 Simple pendulum time period Figure 100 Simple pendulum

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

The main three components in the system shown are:

Figure 101 Mass/Spring system

• 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

The forces acting on the mass are:


-kx –cx2 + Po sin ωt
Newton's law gives:
d2x
m ------ = mx3 = -kx –cx2 + Po sin ωt
dt2
Equation 5 Differential equation of motion of a single-degree-of-freedom system

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.

Liquids & pumping systems

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.

Speed of sound in various liquids (m/s)


Table 1 Speed of sound in liquids

Liquid Velocity m/s


water 1497
methyl alcohol 1123
Benzene 1326
castor oil 1500
Glycerin 1923

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

For sound in air, V(air) = √ (bulk modulus/mass density)


For longitudinal waves in a solid, V(solid) = √(Young's modulus/ mass density).
For sound waves in water, V(water) = √(bulk modulus/ mass density)
The speed of sound in water is approximately 1500 m/s. It is possible to measure changes in ocean
temperature by observing the resultant change in speed of sound over long distances. The speed of sound in
an ocean is approximately:
c = 1449.2 + 4.6T - 0.055T^2 + 0.00029T^3 + (1.34-0.01T)(S-35) + 0.016z
Equation 7 Speed of sound in the ocean

T temp in degrees Celsius,


S salinity in parts per thousand
z is depth in meters

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.

Air & gases

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.

Figure 102 Harmonic series for the tone C.

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

The Newton-Laplace equation for the speed of sound in a gas is


v = √( (γγ P) / ρ)
Equation 9 Newton-Laplace eq. for the speed of sound in a gas

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

where v represents speed,


γ is the specific heat ratio,
T is the temperature in Kelvin,
k is Boltzman's constant,
m is the mass of one molecule,
R is the ideal gas constant, and
M is the molecular weight of the gas.
The speed of sound in a gas depends on the temperature, molecular weight, and molecular structure, but not
on the pressure of the gas. The speed of sound in air at a temperature of 0°C (32°F) and 50% relative
humidity is 331.6 m/s. The speed is proportional to the square root of absolute temperature and it is
therefore about 12 m/s greater at 20°C (64°F). The speed is nearly independent of atmospheric pressure but
the resultant sound velocity may be substantially altered by wind velocity.
Just as in the liquid pipe, we can calculate the natural frequency of the air inside a pipe by using the
formula:
c = f ⋅λ
Eq.8
If a tuning fork is set in vibration and held over the open end of a tube closed at one end, sound waves
which travel down the tube, will be reflected back from the closed end. If the returning wave is exactly in
phase, a condition of constructive interference exists which we call resonance and the loudness of the sound
will be greatly increased. The displacement of the air will be greatest at the open end of the tube, whereas
the pressure difference will be greatest at the closed end, and vice versa.
The lowest frequency where a condition of resonance exists in the tube occurs when the wavelength is four
times the length of the tube. This is called the fundamental mode of vibration. Higher modes of resonance
exist when the wavelength is 4/3(third harmonic), 4/5(fifth harmonic),... of the length of the tube. The first
four resonance modes for tubes open at both ends, and closed at one end are shown below:

121
Vibration Analysis on Rotating Equipment

Figure 103 Sonic vibration in a tube

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

Figure 104 Karman vortices

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

Figure 105 Tacoma Narrows Bridge

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.

Critical Speed (Balance Resonance)


Critical speed is, in general, any shaft rotational speed which is associated with high vibration amplitudes.
Often, the speed which corresponds to a rotor lateral mode resonance frequency excited by rotor unbalance,
in which case it is more correctly called the balance resonance speed.

123
Vibration Analysis on Rotating Equipment

Figure 106 Graphical identification of critical speed

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

Figure 107 Campbell Diagram

125
Vibration Analysis on Rotating Equipment

Identifying resonance in mechanical systems


Mode Shape

Figure 108 Mode shape node points

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.

Figure 109 Mode shape readings

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.

Figure 110 Phase relationships


This phenomenon can be used to help us determine resonance. Typically the running machine cannot be
shut down or have its speed varied for resonance tests, yet the phase change method to determine whether a
resonant condition exists can still be used. Remember that almost all troublesome resonances occur in non-
rotating parts, such as pipes, bases, pedestals, beams, decks and so on. These parts can usually have their
resonant frequencies changed while the machine is still running. Parts may be temporarily braced to add
rigidity and, therefore, moved to a higher resonance frequency that is at least 20 to 25 percent away from
the source frequency.

128
Vibration Analysis on Rotating Equipment

Figure 111 Phase / frequency relationships @ resonance

The bump test

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.

Running machine Bump Test


Normally, bump tests are performed with the machine shut down. However, in certain cases a bump test can
be performed while the machine is running. A running machine spectrum is obtained. Another is taken
while bumping the part. The two spectra can then be compared to determine if there is a resonance. If peak-
hold averaging with running display capability is available, then its use is preferred. With the instrument set
up accordingly, take data with the machine running. Then with the machine still running, bump the
structure repeatedly. If the bumps generate natural frequency peaks that correspond to a peak produced by
the running machine (before bumping), then the amplitude on the analyzer screen should increase. The
peak’s width will also increase if the peak is affected by resonance.

Reverse Bump or “Plucking” the Suspect Part.


Very large columns and beams, such as main steel support structures of buildings, do not as easily respond
to the regular bump test. A structure such as a platform supported by columns will have individual resonant
frequencies for each column and the platform. However, the total structure as a combination of columns
and platform, etc., also has several flexibilities depending on the direction being considered. For example,
a horizontal force can be applied in one direction at the top of the structure. Assume that the force is
applied to the horizontal beam supporting the platform. The force will deflect the beam but will also cause
the whole structure (including columns) to deflect as well. A regular impact bump test can be used for the
horizontal beam, but will not necessarily work well for the total structure that includes all the vertical
columns, beams and platform. To determine the resonant frequency of the total structure, a “reverse” bump
or pluck test works very well. Usually a clamp is attached to the top of the structure. A strong rope is tied
to the clamp. Tension is applied to the rope using any practical means such as pulling with a turnbuckle,
pulley, a “come-along” or even a lift truck or tractor. If the structure is small, such as one supporting a fan
or the total frame of a vertical pump, tension on the rope deflects the structure by several mils. For a very
large structure, such as a tower of several stories, the structure can be deflected as much as a mm or two
without harming the structure. Position the pickup and set the vibration instrument for the expected
frequency range. Now cut the rope. The structure will spring back, giving its resonant frequency in a
manner similar to regular bump tests. The rope will need to be retied, tension applied, cut, etc. Repeat the
procedure several times to make sure repeat readings are obtained. It takes more time and patience than
regular impact bump tests, but often reveals that the magnification due to resonance is not that of an
individual part but instead that of the total structure. Care should be exercised during this type of test as the
rope will whip when cut and could cause injury to anyone in its path. The area where the rope could whip
must be cordoned off and all bystanders moved out of harms way.

Set up for FFT-type analyzers


FFT-type analyzers have various settings which determine the type of spectra obtained. The procedure for
setting up the instrument varies slightly with each instrument but the general procedure is as follows:
1) Mount the transducer on the structure to be tested in the direction that the excessive vibration was
recorded. Ensure that the transducer weight is less than 10 percent of the system weight.
2) Set the Fmax (maximum frequency) to a suitable value higher than the suspected frequency.

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

Figure 112 Impact hammer response


The response can be analyzed with spectrum, Bode and Nyquist plots to identify resonances.

Figure 113 Impact hammer specification sheet

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

Table 2 Fourier spectra of pulses

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

Calculating Natural Frequency in Mechanical Systems


At certain times it is very useful to be able to calculate the natural frequencies of certain components of the
system – especially the supporting structure. The structure can usually be broken down into a number of
components which can be calculated quite easily. This section gives guidelines for the calculation of
natural frequencies for simple structural elements.

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

ωn = 22 * (30,000,000 * 220/(22 * 1444))0.5 (calculator notation)


= 29.00117 rad/s
= 4.616 Hz

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

The plate constant (in this case) is defined by

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

Detuning Resonant Structures


To take any structure out of resonance we must do one of two things:
• Isolate the structure or component from the exciting force, or
• Change the natural frequency of the structure or component
In the case where it is not feasible to change the natural frequency we must isolate.

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:

Table 5 Damping ranges of vibration isolators

Material Effective Damping Range Lowest Amplitude


Expected
Steel Springs 1.2 to 3.8 Hz 3.5 mils p-p
72 to 228 cpm
Rubber 4.0 to 10.5 Hz 0.66 mils p-p
240 to 630 cpm
Cork 8.0 to 12.6 Hz 0.058 mils p-p
480 to 756 cpm
Felt 19.6 to 30 Hz 0.045 mils p-p
1176 to 1800 cpm

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

Figure 114 Stress/strain diagram for steel

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.

Figure 115 Stress/strain diagram for rubber

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.

Figure 116 Resonance Curve

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 k = the spring constant (refer to Hartog p.429)


k (linear spring) = Gd4
8nD2

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.

Figure 117 Resonance curve for spring


material with different internal damping

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.

Figure 118 Schematic representation of the internal


damping properties of rubber. The elliptical area
indicates the loss off energy

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

Figure 119 Vibrations sequence with single impact


for steel and rubber springs

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 120 Relationship between rubber


hardness and shear modulus

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.

Figure 121 The dependence of the compression


modulus upon the shape factor

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.

Modifying the structure


As we mentioned at the beginning of this section, if it is not feasible to isolate the source from the structure,
we must de-tune the structure so that the natural frequency is removed from the forcing frequency or
somehow reducing the amplitude. This can be achieved by changing the mass, the stiffness or the damping.
Consider the generalized formulae:
ωn = k √(Stiffness/Mass) and
Ra = k (Forcing_Function/Damping)
Where ωn = natural frequency
Ra = amplitude of vibration
k = a constant
Let us consider what effect each of these three factors will have on the resonant condition of the
machine/structure system.

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.

Changing the Mass


Remember the formula for natural frequency?

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.

Changing the stiffness

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

Whole body vibration


Sources of vibration
Exposure to localized vibration occurs when a part of the body comes in contact with a vibrating object,
such as power hand tools. Localized vibration from hand-held power tools can increase grip force and
symptoms of hand-arm vibration syndrome (HAV) and carpal tunnel syndrome. Exposure can occur while
holding:
• vibrating tools (e.g., pneumatic and electric rotary tools such as grinders [380 m/sec]),
• impact tools such as chippers (425 m/sec) and riveters, and
• gasoline powered tools such as chain saws (70 m/sec).
Guidelines established by the American Conference of Government Industrial Hygienist (ACGIH, 1993)
suggest that hand vibration in excess of 12 mm/sec should have exposures of less than 1 hour. (The tool
manufacturer should be contacted to obtain tool operation characteristics.) The ACGIH guidelines are based
on symptoms of white finger or Raynaud's phenomenon. ACGIH and other agencies that recommend
vibration guidelines update their guidelines when new information becomes available. •Holding vibrating
work pieces such as when grinding or polishing. •Holding vibrating controls (e.g., lawn mowers).
Exposure to whole-body vibration can also occur while standing or sitting in vibrating environments or
objects, such as trucks or heavy machinery. Prolonged exposure to whole-body vibration has been
associated with back and neck musculoskeletal disorders.
Sources of whole body vibration include virtually all modes of transportation (the jarring and jolting
associated with off-road vehicles, for example). Even a building's vibration may reach levels of concern
(such as in process control rooms or at machine operators' control platforms).
Some examples of whole body vibration sources include the following:
• on-road and off-road vehicles (buses, trucks, tractors, dozers and construction, forestry or mining
equipment);
• rail vehicles;
• aircraft (helicopters, fixed-wing aircraft);
• sea vessels (hovercraft, ships); and
• buildings (generators, pumps, heavy machinery, ventilation systems).

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.

Table 6 Whole body vibration (frequency ranges)

Frequency Range Physiological Result

< 1Hz Motion sickness

4 – 8 Hz Gastrointestinal & spinal problems

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.

High Pass Filter

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.

Low Pass Filter


A measurement filter that removes data above the cutoff frequency.

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

Real Time Bandwidth


What is real time bandwidth? Simply stated, it is the frequency span whose corresponding time record
exceeds the time it takes to compute the spectrum. At this span and below, it is possible to compute the
spectra for every time record with no loss of data. The spectra are computed in "real time". At larger
spans, some data samples will be lost while the FFT computations are in progress.
Rectangular Window (Uniform or No Window)

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

Interval, 4, 91, 92 RUB, 113


Rubber, 139, 140, 142, 143, 144, 145
—K—
Kaiser, 93, 94, 99, 100, 154, 159
—S—
S.C.R, 3, 38
Setup, 4, 91, 92
—L— Severity, 4, 105
Leakage, 154 Shear mode, 6, 24
Line, 154 Sideband, 158
Liquids, 118 Simple Harmonic Motion, 2, 15
Logarithmic, 154 Single Plane, 3, 79, 80, 82
Looseness, 2, 3, 6, 7, 11, 54, 55, 56, 57, 58, 104 spall, 6, 13, 14, 64, 65, 94
lubricant wedge, 62, 63, 64 spectrum, 6, 7, 9, 10, 11, 12, 13, 14, 15, 18, 20, 21,
22, 29, 30, 31, 33, 34, 36, 37, 38, 39, 40, 41, 43,
50, 69, 70, 71, 72, 73, 74, 92, 94, 99, 100, 101,
—M— 102, 103, 104, 106, 107, 108, 112, 114, 124, 130,
magnetic center, 33 131, 132, 133, 151, 153, 154, 155, 156, 157, 158,
Mass, 117, 145, 146 159
Mercury, 119 Spectrum, 2, 20, 21, 101, 106, 107, 129, 151, 152,
Misalignment, 2, 3, 6, 10, 51, 52, 53, 104, 114 153, 156, 158
Mode shape, 127 Speed of sound, 118, 119, 121
mounting techniques, 7, 89, 90 Spring, 117, 142
Stone Walling, 3, 61
Surging, 3, 60
—N—
Natural frequency, 135, 142 —T—
Newton, 8, 22, 23, 45, 117, 118, 121, 150
Nyquist, 132 Tachometer, 158
Technology, 3, 91, 92
time, 6, 15, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 29,
—O— 30, 32, 33, 34, 41, 50, 59, 64, 66, 73, 79, 84, 85,
Octave, 4, 101, 102, 155, 156 87, 92, 94, 99, 100, 101, 102, 103, 106, 109, 110,
Orbit, 7, 106, 111, 155 113, 115, 116, 117, 118, 127, 130, 131, 133, 150,
Orbits, 4, 110 151, 152, 153, 154, 155, 156, 157, 158, 159
Overall, 155 Time, 2, 4, 7, 19, 72, 73, 101, 129, 157, 158, 159
Overlap, 4, 101, 102, 155 tolerance stack up, 7, 76
Transfer Function, 106, 158
Triggering, 4, 109
—P—
Parameter, 2, 4, 22, 91, 92 —U—
PEAK, 2, 18
Peak Hold, 101, 154, 156 Uniform, 5, 99, 100, 135, 157, 158, 159
PFA, 3, 4, 7, 91, 93, 95
Phase, 4, 7, 50, 83, 84, 107, 127, 128, 129, 156 —V—
Piston pulse, 120
PRELOAD, 112 Vane pass, 120
vector, 7, 80, 81, 82, 83, 84, 101, 108, 113, 158, 159
velocity, 2, 6, 10, 11, 13, 15, 16, 17, 18, 22, 23, 31,
—R— 33, 50, 61, 64, 65, 69, 74, 92, 94, 104, 117, 118,
119, 121, 122, 123, 127, 150, 152, 154
Resolution, 154, 157
Vibration isolators, 139
resonance, 4, 7, 9, 25, 26, 28, 31, 33, 59, 69, 94, 111,
112, 113, 114, 115, 116, 118, 120, 121, 122, 123,
124, 126, 127, 128, 129, 130, 139, 140, 141, 142, —W—
143, 146, 147, 149
Resonance, 2, 4, 8, 28, 104, 123, 141 waterfall, 4, 6, 30, 31, 69, 109, 124
RMS, 2, 6, 16, 18, 19, 100, 101, 154, 155, 156, 157 waveform, 6, 7, 15, 18, 19, 20, 26, 27, 29, 32, 33, 50,
rotor bars, 32, 33, 34 69, 72, 73, 94, 99, 133, 153
Rotor Bars, 2, 34 whip, 112, 130
Route, 157 whirl, 60, 112, 124

160
Vibration Analysis on Rotating Equipment

Whole body vibration, 148, 149 Windows, 4, 99, 159


windings, 32, 33, 71

161