This action might not be possible to undo. Are you sure you want to continue?

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

114 What is resonance? ---------------------------------------------------------------------------------------------.130 Impact hammer ----------------------------------------------------------------------------------------------------.93 Stage 1 -----------------------------------------------------------------------------------------------------------------93 Stage 2 -----------------------------------------------------------------------------------------------------------------94 Stage 3 -----------------------------------------------------------------------------------------------------------------94 Stage 4 -----------------------------------------------------------------------------------------------------------------94 Including the Component Failure in the PFA Tree.118 Air & gases---------------------------------------------------------------------------------------------------------.109 Bodé plots ----------------------------------------------------------------------------------------------------------.130 Set up for FFT-type analyzers --------------------------------------------------------------------------------.96 Measurement Windows ---------------------------------------------------------------------------------------------99 Averaging ----------------------------------------------------------------------------------------------------------.109 Triggering ----------------------------------------------------------------------------------------------------------.115 Mechanical ---------------------------------------------------------------------------------------------------------.108 Representation by complex numbers --------------------------------------------------------------------------.122 Critical Speed (Balance Resonance) -------------------------------------------------------------------------.127 The bump test ----------------------------------------------------------------------------------------------------.110 Orbits----------------------------------------------------------------------------------------------------------------.101 Octave Analysis ---------------------------------------------------------------------------------------------------.110 INTRODUCTION TO RESONANCE -----------------------------------------------------. --------------------------------------------------------.106 Advanced functions ---------------------------------------------------------------------------------------------.129 Running machine Bump Test -----------------------------------------------------------------------------------.105 Two Channel Analysis------------------------------------------------------------------------------------------.115 Natural Frequency ----------------------------------------------------------------------------------------------.126 Phase---------------------------------------------------------------------------------------------------------------.115 Liquids & pumping systems-------------------------------------------------------------------------------------.126 Mode Shape ------------------------------------------------------------------------------------------------------.106 Two channel functions -------------------------------------------------------------------------------------------.123 IDENTIFYING RESONANCE IN MECHANICAL SYSTEMS ----------------------. --------------------------------------------------------------.108 Cascade & waterfall plots----------------------------------------------------------------------------------------.95 Conclusion ----------------------------------------------------------------------------------------------------------.130 Reverse Bump or “Plucking” the Suspect Part.100 Real Time Bandwidth and Overlap Processing --------------------------------------------------------------.102 Analysis -----------------------------------------------------------------------------------------------------------.131 4 .103 Severity charts -----------------------------------------------------------------------------------------------------.120 Karman Vortices --------------------------------------------------------------------------------------------------.Vibration Analysis on Rotating Equipment Developing a Potential Failure Analysis for Rolling Element Bearings --------------------------------.

142 Modifying the structure----------------------------------------------------------------------------------------.Vibration Analysis on Rotating Equipment CALCULATING NATURAL FREQUENCY IN MECHANICAL SYSTEMS-----.139 Springs --------------------------------------------------------------------------------------------------------------.147 5 .139 Vibration isolators ----------------------------------------------------------------------------------------------.147 Sources of vibration---------------------------------------------------------------------------------------------.146 WHOLE BODY VIBRATION ---------------------------------------------------------------.145 Changing the Mass------------------------------------------------------------------------------------------------.145 Damping ------------------------------------------------------------------------------------------------------------.137 DETUNING RESONANT STRUCTURES-----------------------------------------------.140 Rubber --------------------------------------------------------------------------------------------------------------.135 Plates---------------------------------------------------------------------------------------------------------------.135 Uniform Beams --------------------------------------------------------------------------------------------------.146 Changing the stiffness --------------------------------------------------------------------------------------------.147 Frequency ranges -----------------------------------------------------------------------------------------------.

Vibration Analysis on Rotating Equipment List of illustrations Figure 1 Velocity spectrum showing imbalance ______________________________________________ 9 Figure 2 .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 .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 .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.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.Velocity spectrum showing fan imbalance _________________________________________ 10 Figure 3 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 ._____________________________________ 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 .3D view_______________________________________________________________ 28 Figure 21 Two channel time waveform .full wave rectified _______________________________________ 43 Figure 37 Same motor .radial ________________________________________ 11 Figure 4 Velocity spectrum of misaligned fan .

___________________________________________ 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.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 .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.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. _______________________________________________ 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 .

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 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.

The AC drive motor rotates at just under 1200 rpm. The fan was driven from the AC motor via a V-belt and rotated at about 720 rpm. Euler in 1744 and Bernoulli in 1751 developed the equation for the vibrations of beams and developed the normal modes for various boundary conditions. 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. 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.C. who studied the origin of musical sounds and the vibration of strings. In 1638 Galileo described the vibrations of pendulums. 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 1882 Hertz developed the first successful theory for impact. the phenomenon of resonance and the factors influencing the vibration of strings. 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.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. 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 Vibration Analysis .

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. 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). 10 . Misalignment Probably 40% of all bearing and shaft failures are caused by misalignment of the components creating an extra axial thrust on the bearings. 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. Notice in the spectrum that there is one big spike which is labeled at 716. Figure 2 . 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.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. 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. Because the one spike is so dominant that is the one that we are concerned about.18 cpm which is exactly twice fan speed.59 cpm and there are two much smaller spikes just to the right.Vibration Analysis on Rotating Equipment mm/s or ips (inches/second). The spectrum was recorded from a vibration transducer which was mounted radial to the shaft (vertical in this case).

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. allowing the component to rattle.Vibration Analysis on Rotating Equipment Figure 3 Velocity spectrum of misaligned fan . 11 .axial However. Looseness Looseness exists when the component is not directly attached to the structure or rotating element and has a relatively large clearance.radial In figure 3 we see what initially looks like an imbalance condition of the fan. a high axial velocity vibration relative to the radial vibration is almost always indicative of component misalignment to the belt. on a belt driven train. although the amplitudes are relatively low. in figure 4 we are now looking at the vibration taken axial to the shaft. looking at our spectrum explanation charts we see that. Figure 4 Velocity spectrum of misaligned fan . Again.

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. The motor was running at 590 rpm and immediately we see the large family of harmonics of run speed. 12 .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 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.

As the looseness deteriorates the envelope readings will decrease but then the velocity readings will start to increase.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. 13 . In other words they are non-synchronous. Notice that the spike at about 8. We will study bearing defects in detail later but notice that the main defects are not multiples of run speed. be looking for harmonics of run speed. 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. Figure 6 shows the early stages of looseness of a bearing inside the fan drive motor. and using demodulated or enveloped acceleration readings.772 cpm is marked BPIR which stands for Ball Pass Outer Race.

inner race defect Figure 8 Inner race spall 14 .Vibration Analysis on Rotating Equipment Figure 7 Enveloped acceleration spectrum of bearing .

Figure 9 uses a pendulum. rotating electrical fields. In addition to frequency the amplitude is the other necessary quantity that must be known in order to characterize vibration. The frequency of vibration is often expressed in terms of cycles per second or HERTZ after the German physicist Heinrich Hertz. it has a velocity associated with each position and. When viewing the time domain data it can be quite difficult to separate these components of vibration. If the pendulum swings back and forth 100 times in one minute. The purpose of this section is to describe the relationship between each of these and how they are used on rotating machinery. In vibration work.001 inch and one micron (µm) is 0. the velocity also varies between a positive peak and a negative peak. gear meshing forces. To understand the concept of vibration analysis. a rotating mechanism or a diving board. then the frequency is 100 cycles per minute. Similarly if a machine is rotating 100 times in one minute. 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. in the frequency domain it is much easier to separate these elements to determine the importance of each.001 mm. Since the pendulum is continuously moving. However for predictive maintenance techniques where rotational speed is often the key to vibration peaks. a mass and spring combination. The time domain waveform is composed of a machine’s response to many individual forces such as imbalance. the displacement is often expressed in terms of mils or micron. 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 displacementAmplitude meters are often calibrated to give the peak to peak value because it is the displacement extremes that are of interest. However. velocity and displacement. it is important to realize that the motion of the measured surface varies with time. One Mil is equal to 0. cycles per minute are used in preference to Hertz.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. acceleration. its speed is 100 revolutions per minute or 100 RPM. Simple Harmonic Motion Simple harmonic motion can be visualized by many common examples such as a pendulum. misalignment. and many other factors. like displacement. Vibration amplitude is measured using three different parameters. 15 .

and at position A the velocity is maximized. The rate of change of displacement is the velocity. 16 . Acceleration Vibration is described by its frequency and amplitude. Displacement . Since the peak positive velocity occurs 1/4 cycle before the peak positive displacement. velocity or acceleration. instead of the usual mils. then to the left.. The 90 degree phase lead is shown in the diagram on figure 10.. called the frequency. velocity is said to lead displacement by 90°. not magnitude.. The negative peak velocity differs only in direction.. Velocity . Frequency is usually expressed in terms of cycles per second. first to the right. then the product 2πfD will be the velocity in inches per second which are the units used for velocity in vibration work. therefore if D is expressed in terms of inches.Vibration Analysis on Rotating Equipment (48) * FREQUENCY * AMPLITUDE . one cycle C B A An oscillating system will produce a certain number of cycles per unit time. Velocity amplitude is expressed only in terms of zero to peak or zero to RMS. the velocity is zero. Figure 9 Simple Harmonic Vibration Figure 10 shows that at position B and C... The amplitude is expressed in units of either displacement. or Hertz.

Acceleration leads velocity by 90°.81 m/sec2 and is derived from the acceleration due to earth’s gravity. A B A C A C B A Highest Velocity Figure 4. Figure 11 shows that at position B and C the acceleration is maximum. velocity is to the right and just after it is to the left. The diagram in figure 11 shows these phase leads and also the acceleration amplitude relationship. At B therefore the rate of change of velocity. Note that acceleration reaches its maximum at Points B and C just as displacement does. The distance between the extremes of motion is the peak-to-peak displacement. 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°. is maximum. The amplitude is one half of the peak-to-peak value for a sine wave. the acceleration is proportional to the square of the frequency. This says that for any given value of displacement. o Figure 10 Integration from acceleration to velocity As velocity is continuously changing. The unit of acceleration is the “g” which is equal to 9. it is decreasing. The velocity amplitude is directly proportional to frequency for a given displacement. but at B acceleration is to the left whereas the displacement is to the right. Therefore a 90 phase shift exists between displacement and velocity. Velocity is highest where displacement is zero and is zero where displacement is maximum. this acceleration is also associated with the motion. A = (2πf)2D. Just prior to point B. Conversely just prior to point A velocity is increasing and just after. Acceleration is the third way to express vibration amplitude. an acceleration is also associated with the velocity. Displacement meters are often calibrated in peak-to-peak units. the acceleration.Vibration Analysis on Rotating Equipment (49) A B A C A Disp Peak Peak to Peak C B A Figure 3. Therefore the rate of change of velocity (the acceleration) must be zero at A. Vel Disp 17 .

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.e. Electrical (AC) voltage is also measured in rms. Velocity is a quantitative measure of the effective velocity and reflects the power or energy being used to vibrate the machine mass. Acceleration amplitude varies as the square of frequency for a given value of displacement. acceleration is maximum to the left and displacement o maximum to the right. squaring the amplitude value of each point. stereo speakers power values are measured in rms values. 18 Accel Vel Disp . The simple relationships which exist between average. a 180 phase shift. Using rms values can be compared to the use of rms in electrical circles i. During most of the cycle the value of the instantaneous voltage is somewhere between zero and peak. is an attempt to apply a single quantitative value -which reflects the effective value of this varying function. Peak Avg Amplitude RMS Peak to Time Peak Avg = 0.Vibration Analysis on Rotating Equipment (50) A B A C A C B A Acceleration Acceleration Figure 5. At B. calculating the mean of the squared values and then finding the square root of the mean. True peak can be seen in the time waveform. is a continuously varying quantity.637 x Peak RMS = 0. ranging from zero to a peak value.707 x Peak Figure 6. then. RMS and Peak amplitude values for sine waves are not valid for combination or random waveforms. Figure 11 Integrating to displacement RMS vs. RMS. This. PEAK The rms or root mean square value is calculated by breaking the waveform down into a number of points. This same logic applies to vibration. Peak value is the maximum amplitude seen during the measurement. 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. like vibration signals.

With respect to machinery vibration. The method most analyzers use to transform signals from the time domain to the frequency domain is called :- 19 .Typically the signal would be displayed on an oscilloscope. its position represents its frequency.Vibration Analysis on Rotating Equipment (51) Peak Avg Amplitude RMS Peak to Time Peak Simple Sine Wave Peak Amplitude Peak to Peak RMS Time Complex Waveform Figure 12 Peak -v.time. Its height represents its amplitude.RMS Time Domain The traditional way of observing signals is to view them in what is called the time domain. We now have axes of amplitude v frequency instead of amplitude -v. 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. The time domain is a record of what happened to a parameter compared to time. Every sine wave separated out by the FFT appears as a separate line.

The FFT is simply a clever set of operations which implements Fourier's theorem.400. The number of samples determines the resolution (number of lines) of frequency. and displays the spectrum of these measured frequency components. which passed through the filter. then the sampled data will accurately represent the input signal. which performs this transformation automatically and then stores the computed spectra into memory. It takes 0. The FFT Analyzer An FFT spectrum analyzer works in an entirely different way. Fourier's theorem states that any waveform in the time domain can be represented by the weighted sum of sines and cosines. Suppose that you take 1024 samples at 2560 Hz. of lines (2560 / 2. but over what frequency range? The highest frequency will be determined by the in-built ratio of Fmax to data sampling rate .56) 2.56 x Fmax usually). The input signal is digitized at a high sampling rate. Nyquist's theorem says that as long as the sampling rate is greater than twice the highest frequency component of the signal. The FFT spectrum analyzer samples the input signal. was measured to determine the signal strength in certain frequency bands.800. The corresponding FFT yields a spectrum with discrete frequency samples. By tuning the filters and repeating the measurements.4 Seconds to take this time record. In fact. and compute its frequency spectrum.56 No samples / 2.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. The original digital time record comes from discrete samples taken at the sampling rate.3200 or even 6400 Lines. The FFT of this record yields 400 frequency points or lines.200. The lowest frequency is just the Fmax divided by the number of lines: Fmax No. a spectrum could be obtained.56. The resulting digital time record is then mathematically transformed into a frequency spectrum using an algorithm known as the Fast Fourier Transform or FFT. like you would see on an oscilloscope trace. the spectrum has less than half as many frequency points as there are time points (remember Nyquist's theorem). The signal power.5 Hz (the same as the lowest measurable frequency) 20 .56) / (1024 / 2. an analog filter was used to isolate frequencies of interest. Certain analyzers pass the input signal through an analog filter.1600. We therefore must sample and digitize the time domain input. We cannot transform to the frequency domain in a continuous manner. FFT Spectrum Analyzers take a time varying input signal. computes the magnitude of its sine and cosine components.2. In simple terms. 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. Most analyzers have an FFT processor. Most analyzers offer resolutions of 100. (2.56 = = = Fmax / No. Now here's the interesting part. Many of these measurements were once done using analog spectrum analyzers. Of Lines = Bin resolution = data sampling rate / 2. The resulting spectrum shows the frequency components of the input signal.

This is because each frequency component has a phase relative to the start of the time record.Vibration Analysis on Rotating Equipment Everything below 2. This bin actually covers the range from 0 Hz to 2. independent of phase.00125 Hz resolution would require 2.5 Hz (for this example) is considered to be DC. What happens if we make the time record 800 mS or twice as long? Well.5 . the second 2. then the real (cosine) or imaginary (sine) part or the phase may be displayed.25 Hz resolution containing 800 points. It is simply the complex FFT. If we use a synchronous trigger then each frequency component has a well-defined phase. One is memory. Advantages of FFT Analyzers The advantage of this technique is its speed. The first covers 0 . the magnitude of the spectrum is displayed. Measurement Basics An FFT spectrum is a complex quantity. and the 400th covers 997.5 . The range between 1000 Hz and 2. we ought to get 2048 time points (sampling at 2560 Hz) yielding a spectrum from DC to 1000 Hz with 1. Spectrum The spectrum is the basic measurement of an FFT analyzer. The magnitude is the square root of the FFT times its complex conjugate. the entire frequency span is not useable. If we keep increasing the time record. the longer the processing time. (Square root of the sum of the real (sine) part squared and the imaginary (cosine) part squared). then we would need to store more and more points. We just described how we arrived at a DC to 1000 Hz frequency span using a 400 mS time record.000 values.048. (Alternately.5 Hz. then the phase is random and we generally look at the magnitude of the spectrum. but again.1000 Hz. A typical filter has a flat response from DC to 1000 Hz and then rolls off steeply from 1000 Hz to 2. 21 . There is also a frequency bin labeled 0 Hz (or DC). the time record is triggered in phase with some component of the signal. But the analyzer places some limitations on this.) Another limitation is processing time. The magnitude is a real quantity and represents the total signal amplitude in each frequency bin. i. So our final displayed spectrum contains 400 frequency bins. Frequency Spans Before we continue.e. the entire spectrum takes only 400 mS to measure. The length of the time record determines the frequency span and resolution of our spectrum. you may wish to think of the input signal being composed of sines and cosines.) If there is no triggering. 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. (0.2. Normally. let's clarify a couple of points about our frequency span. The more points you take. If there is phase information in the spectrum. The output spectrum thus represents the frequency range from DC to 1000 Hz with points every 2. Because the signal passes through an anti-aliasing filter at the input.56 kHz. To measure the signal with higher resolution the time record is increased.5 Hz.5 Hz (the lowest measurable frequency) and contains the signal components whose period is longer than the time record (not only DC). providing an enormous speed advantage. In the case of a 1000 Hz span and 400 resolvable frequency bins. For vibration measurements phase is usually considered to be relative to the trigger pulse. all frequencies are examined simultaneously. The phase is simply the arc tangent of the ratio of the imaginary and real parts of each frequency component.5 Hz.56 kHz is therefore not useable and the actual displayed frequency span stops at 1000 Hz.

In applying this to rotating machinery displacement measurements are relative readings of the displacement of the shaft to a reference. 20 orders is a frequency of 1200 Hz. 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. Velocity transducers will give an output proportional to the linear speed (velocity) of the transducer in inches/second or millimeters/second. this is severe regardless of the frequency”. Now suppose a velocity of 0. In other words the acceleration vibration gives a good indication of impactive forces inside the machine such as bad bearings. of the rotor as a “simple harmonic motion”. The most prominent advantage of a velocity measurement is that the value of rms velocity is related to the potential for mechanical damage. For higher frequencies however. The transducer sees the force once per rev. 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 example. The output of the transducer will depend on what type of transducer we are using. as near as possible to the bearing and couple the transducer to a spectrum analyzer. The many published vibration severity charts are based on this principle. For predictive maintenance purposes we use accelerometers almost exclusively so we will concentrate on them for now. In summary. According to Newton’s Second Law F=m⋅x a ⋅ 22 . For a 3600 rpm machine. 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. regardless of the frequency. is this severe? At 6 Hz this is not severe at all. velocity or acceleration As previously discussed. We place our transducer on the drive end of the motor in the vertical direction.Vibration Analysis on Rotating Equipment Parameter Selection Selecting displacement. As an example suppose a displacement of 0.13 mils pk-pk. Velocity measurements are especially good for a number of reasons. displacement measurements are good from 0 Hz to 500 Hz. How does it work? Consider a rotating machine (a motor) which has. usually the bearing. Newton’s second law (F=ma) tells us that the acceleration of a body is directly proportional to the force applied to the body.6 in/sec (15mm/s) is observed. 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.5 in/sec is 0. 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. Is this severe? The answer is “Yes.l mils is observed. displacement amplitude is higher at lower frequencies. 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. velocity up to 1 kHz and acceleration from 2 Hz to 20 kHz depending on the design of the accelerometer. to centrifugal) force. very small but certainly a measurable value. • • • Displacement transducers will give an output proportional to the linear displacement of the transducer in thousandths of a inch or micron. significant vibration has a displacement value which -is too small to conveniently measure and velocity or acceleration is more appropriate. At that frequency 0.

200 rev/min the spike will have a frequency at 1. Back to our motor. then we may safely say that the acceleration is also proportional to the square of the speed. the acceleration amplitudes must increase for a given force even if there is nothing wrong with our machine. However. so if we need a value that is independent of frequency for severity analysis purposes we can use the velocity reading. 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. We have already said that we have a concern that the acceleration increases with frequency. we know that acceleration is simply the rate of change of velocity. 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. 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. or the higher up the frequency range we go. So if we integrate our acceleration reading with respect to time we will get a velocity reading. if the motor is running at 1. Integrating acceleration will change our value from: inches/second2 to inches/second effectively finding the square root of the acceleration (for time). Either way.200 cycles/minute (cpm) or 20 Hertz (Hz). If we plot the acceleration against time (time domain) we would see a sine wave which is the result of simple harmonic motion. Newton also says that for a rotating body F=m⋅ϖ2⋅r ⋅ where ϖ r = the rotational speed in radians/second = the radius at which the force is acting.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. 23 . a=ϖ2⋅r ϖ The important point here is that the faster the machine goes. This is the signal that is passed along to the analyzer.

However. Permanently mounting an accelerometer should be done with care. Generally speaking the shear mode accelerometer offers better axial sensitivity with much better mechanical integrity. The way the accelerometer is mounted will affect the resonant frequency and. 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. the useable frequency range.Vibration Analysis on Rotating Equipment Accelerometers Figure 13 Compression mode accelerometer Looking at the figure above we see a schematic of an accelerometer. 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. hence. However. 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. Figure 14 Shear mode accelerometer Many low cost industrial accelerometers are now shear mode. Modern accelerometers are available as compression mode or shear mode. the response from the accelerometer is better if it is permanently mounted. on a motor it is usually not practical to drill into the motor frame for obvious reasons. The best 24 .

The run speed signal is being MODULATED by a signal at 4 Hz which is probably a foundation resonance. 25 . 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. This type of modulation is commonly found in maintenance applications but consider the example below.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. The example above was taken from a large steam turbine running at 3600 rpm. 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.

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 modulation is at this frequency. process. In vibration terms.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).07) Hz = 61. high frequency pulses sitting on top of the high amplitude.F.T. The bearing excitation resonance is shown as small. 26 . 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. low frequency vibration. As can be seen from Figure 2. demodulation is a way of extracting the rate of occurrence of high frequency resonances. The demodulation process extracts the modulating frequency to produce a time waveform which can be handled by the F. 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*3.4 Hz.07 x run speed). In other words the bearing outer race frequency is modulating the bearing resonance frequency. The 2 kHz vibration is the resonance of the bearing which is being excited by the bearing outer race frequency (3.

To modify the signal so as to be suitable for F. 27 .T.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.T.F. Figure 18 Enveloping process With the time domain signal in this format the F.F. 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. conversion would give a single spike in the frequency domain at the resonant frequency which we have earlier said is not what we want.

5 to 4 kHz range (typically).F.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. 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.Vibration Analysis on Rotating Equipment Figure 19 Fast Fourier Transform The signal is NOW passed through the F. Figure 20 FFT . The bearing housing will have its own 28 . and we get a spike in the frequency domain at the bearing defect frequency (figures above and below). while this is often the case it is not always so. 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.T.

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 . A.C.i.C.P.C. The lower plots show the time domain (left) and frequency domain (right) over a 50 mS period of the lower 640 mS plot. 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. This plot shows the signal from the inboard bearing of a 35 H. the carrier wave signal resonance could be coming from any part of the mechanical structure. A. the machine structure will have its own resonance.e.Vibration Analysis on Rotating Equipment resonance. a high frequency oscillation. Figure 21 Two channel time waveform . Motor Example. Ultra-sound data are heterodyned to the audible range so demodulating at 5 to 8 kHz gives acceptable results while A.bearing defect The 2 upper plots are the time domain signal in two planes over a period of 640 mS. 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.C. 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. 29 . motor operating a belt-driven fan. frequency of 60 Hz or 50 Hz. In short. Note also that the lower left portion of the plot is a zoom of the windowed part of the long time record. electric current should be demodulated from the A.

This clearly shows the modulation of the 2 kHz frequency while the 3 kHz frequency is static. The frequency range of 30 . the ball impacts on the defect causing the 2 kHz vibration to suddenly rise and then ring down. Every time one of the bearing balls passes a defect on the outer race.001G. The modulation has been calculated to be equal to the bearing outer race defect frequency of the motor inboard bearing.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. Figure 23 Enveloped acceleration spectrum Note that the demodulated spectrum is clean and extremely easy to analyze. The figure below shows the demodulated spectrum on the left with waterfall plot on the right above a trend of the defect frequency. The 2 kHz is the resonant frequency and the bearing defect frequency (outer race) is the modulating frequency. 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.

Note that the demodulated spectrum is much cleaner and easier to analyze.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.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). Figure 24 Comparison . 31 .

In other words these frequencies will have sidebands of 1x. If phase readings are taken. If the nature of the pulsating vibration can be determined. Electrical problems with induction motors will often cause the motor load current ammeter to swing or pulsate in a cyclic manner. 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. Wound rotor windings Defects in the rotor will cause a modulation of rotor bar pass or stator slot passing frequencies at run speed. A bowed rotor This defect usually occurs on very large. the rotor will appear to swing back and forth. defective bearings and looseness. • • • 32 . the vibration is likely to be due to electrical problems. 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. this can help significantly to identify the specific problem as discussed in the following paragraphs. 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. unbalanced phases. of poles sidebands around line frequency to previous levels. the largest component frequency of vibration resulting from these electrical problems will be 1 x RPM and. unequal air gap and other similar problems. broken rotor bars. Mechanical problems may include imbalance. the problem is more likely to be mechanical in nature. In vibration readings this will look very like imbalance with a small axial component. On the other hand. 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. Generally. If this is the case conventional electrical testing procedures can be carried out to pinpoint the true cause of vibration. 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. if the vibration amplitude decreases only gradually after power is disconnected.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. The unequal magnetic forces may be due to open or shorted windings. this will appear similar to imbalance. or if the motor has a strobe light flashed at run speed. Perhaps an even better indication of the contribution of electrical problems is by observing the time waveform of the vibration as power is disconnected. Vibration caused by electrical problems are normally the result of unequal magnetic forces acting on the rotor or stator. 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. 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. misalignment. it will be seen that the phase is erratic and instead of the strobe “freezing” the rotor. If the vibration disappears the instant the power is shut off.

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. 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. This range will usually also include the rotor bar pass and slot pass frequencies. This vibration is rarely troublesome except where extremely low vibration levels are required.g. 33 . 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. Torque pulses have also been known to excite loose rotor bars and loose stator windings at frequencies of 2x. and even 4x torque pulse frequency. 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. Care should be taken if a significant 7. 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. 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. in this case the pulsation is the result of a beat between two separate frequencies of vibration which are very close together. 60 x 1/3 for a 1200 rpm motor) around 2 x line frequency. The result will be noticeable steady pulsation or “beat” of the vibration amplitude. If demodulated spectra are used this will show up as very high spikes at 120 Hz (100 Hz in Europe) in the frequency domain.5 MΩ for a main drive 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. to produce a vibration whose amplitude pulsates in a cyclic fashion it is necessary that two frequencies of vibration be present. Since each motor pole is essentially energized twice for each cycle of AC current. 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 vibration resulting from torque pulses will be two times the line frequency powering the motor. torque pulse frequency will be 120 Hz or 7200 cpm. However.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. 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. 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. if an AC line frequency is 60 Hz or 3600 cpm. or if the torque pulses should happen to excite a resonance condition in the machine or structure.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.200 cpm) are apparent in your demodulated spectrum. the first thing to check is the individual phase power factor. 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. One of these vibration frequencies may be the result of some imbalance or misalignment occurring at the running speed of the armature. Thus. Torque pulses are generated as the rotating magnetic field of the motor energizes the stator poles. Demodulated readings will demodulate the frequencies above the high pass filter or inside the band pass filter. • • In the case of a stator related problem. Electric motors have inherent vibration due to “torque pulses”. If frequencies at 120 Hz (7. 3x.

Frequency analysis of the motor load current has been consistently proven to be able to detect the presence of broken rotor bars. 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. as the defective bar cuts the lines of flux. = 21. 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).variations in current draw. 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. This means there are three individual alternating current waveforms being supplied simultaneously at 60 Hz but 120 degrees out of phase. When six SCRs are used to convert the AC to DC.C. Thus 180 Hz x 60 sec. These peaks. Those who are more comfortable using cpm (Cycles Per Minute). = 10. looseness. 34 . However. AC is supplied in the United States at 60 cycles per seconds or 60 Hz (Hertz). 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. induction motors. 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.800 cpm and 360 Hz x 60 sec. bad bearings etc. The object of the project was to determine a reliable and repeatable method for the evaluation of broken rotor bars on A. To identify the separate spikes clearly the spectrum must have a fine line resolution. The driving force for this project was the large number of motor failures suffered at these production facilities. 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. thus. (fullwave rectified or High Efficiency System) then a pulsing or frequency of 360 Hz (6 x 60 Hz) is created in the DC drive system. Industrial power in the United States is supplied in a three phase 60. 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. end ring resistances and cast rotor blow holes. Direct current is created by taking three phase alternating current and converting it through a bank of silicon controlled rectifiers (SCRs) into direct current. create a pulsing which is detectable through vibration analysis. 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. rather than Hz.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. DC Motors How DC Power Is Created. although no longer alternating.600 cpm. Each of these three problems give the same effect to the motor . need only multiply the frequency in Hz by 60 seconds. The three waveforms are identical in amplitude and duration.Hz format.

/sec.1 in. This usually means that there is an electrical control problem.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. = 3.800 cpm and a full-wave rectified system will have a dominant electrically related frequency of 21. = 21.600 cpm • • • Three phase alternating current frequency is 180 Hz or 3 x 60 Hz x 60 sec. @ 10.800 cpm Full-wave rectified direct current ( 6 SCRs ) is 6 x 60 Hz = 360 sec. = 10. 35 . peak are detected.600 cpm When these frequencies exist within their respective systems they should be considered normal unless amplitudes greater than 0. Half-wave rectified direct current (3 SCRs) is 3 x 60 Hz @ 180 Hz or 180 Hz x 60 sec.600 cpm Hz or 360 Hz x 60 A half-wave rectified DC drive system will then have a dominant electrically related frequency of 10.800 cpm and also has 180 positive and 180 negative amplitude peaks per second.

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. This allows the motor to run at the desired speed or 36 .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. 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 DC source can be varied through system controls to change the running speed of the motor.

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. low voltage. 37 .Vibration Analysis on Rotating Equipment desired rate of speed change to meet the needs of the driven system. The comparitor card. the tachometer (tach.600 cpm between existing spectrum peaks or a frequency of 7.200 cpm and frequency differences of 7. 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.600 cpm and frequency separation of 3. 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. 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. DC power frequencies will always be constant at 21. If one sees a frequency of 3. must perform correctly to allow the system to function normally.200 cpm then faulty firing cards or SCRs may be the cause. based on the voltage differential between the constant voltage pot and the tach.800 cpm on a half-wave system. low voltage is compared to a constant or predetermined voltage variance by the comparitor card. As the motor turns. DC Control Firing Cards The firing cards. 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. which control the opening and closing of the SCRs. signals the firing cards.) also turns which creates a low voltage proportionate to its speed. The tach. There is a firing card or order for each SCR which controls the power flow through the SCR.600 cpm on a full-wave system or 10.

The production system was critical to the plant operation. 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 motor did start and run. A motor bearing then seemed the next logical failure point. 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.Vibration Analysis on Rotating Equipment 1 S.R. however. The pulsing was still apparent even under a no load condition. as previously mentioned. The analysis was performed 1 Thanks to Bill Rinehart for his permission to use this data 38 .C. 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.

six SCR system). This DC control system uses three firing cards to control six SCRs which is typical of many DC control systems. The FFT spectrum in figure 27 was taken after the repair.00847 in/sec peak while the amplitude of the spectrum above at 7. The 21.600 cpm if the system has: • • • One firing card for each SCR. The firing cards were examined and a loose connection on one card was found and repaired.600 cpm. then suspect firing card or SCR problems in the DC control. yet each spectrum represents the same firing card malfunction. Figure 31 FFT spectrum showing half wave firing card frequencies Half-wave rectified AC power sources will tend to have 1/3 multiples of 10.600 for a full wave. A simple voltage test of the three incoming AC phases should confirm this situation if present. The 7.800 cpm or frequency separations of 3.600 cpm. Although the frequencies are the same in each spectrum the amplitudes are considerably load dependent. Dominant frequencies of 7. Another possible reason for seeing these 1/3 multiples would be if one phase of the AC power source was not present. The amplitude at 7. 14. When one firing card is not functioning then 1/3 of the power is lost.200 cpm.200 cpm is 0. not bearings or mechanical defects When an FFT appears with dominant 1/3 frequencies of the DC full pulse frequency (21.400cpm frequencies are now gone.200 cpm on the spectrum below is only 0. 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). The dominant frequencies are 7.200 cpm and 14.Vibration Analysis on Rotating Equipment revealing the FFT spectrum in figure 26.400 cpm or frequencies equivalent to multiples of 1/3 of 21.400 cpm and 21. Full-wave rectified AC power sources can also have frequency separation of 3.600 cpm frequency is the normal frequency of a full wave system and should be present. 39 .600 cpm are representative of this situation. These frequencies are related to DC ELECTRICAL problems.200 cpm and 14.3037 in/sec peak. This would affect one-third of the system power and virtually render one bank of SCRs inactive.

The sidebands do. 40 . Testing the voltage output of the tach. it has also been observed that these side-bands will grow or diminish as the motor RPM is varied. To resolve side bands related to comparitor card malfunctions a FFT spectrum of 3200 lines of resolution at Bandwidth of 24. 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. however. Another possible reason to see these FFT characteristics could be a faulty or malfunctioning tachometer which would corrupt the voltage going to the card. they will remain equally spaced.000 cpm is suggested. 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. 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. however. should confirm this situation. exist regardless of the cause and should be considered a warning of this component failure.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.

0872 5000 16. The most obvious of these is the tachometer which extends. A RPM check using a digital tach. Example of a Comparitor Card Defect Vibration analysis was requested on a 125 HP DC motor which was thought to be vibrating heavily.Vibration Analysis on Rotating Equipment 0. A spare comparitor card was eventually located and the replacement accomplished. voltage was then checked but seemed to be consistent with the operating speed and fluctuations. 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. also revealed that the motor was fluctuating or hunting approximately 30 cpm at an RPM of 1440. There was not a spare card available at this time but there were spare firing cards. 41 . The tach. Based on the side-bands a recommendation was made to change the comparitor card (also called the control card).000 Figure 33 FFT spectrum showing comparitor card defect. The first set of spectra was collected under these conditions and revealed the above spectrum. 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. DC motors are different from AC motors because of their power supply which requires different components. The speed fluctuations stopped and another set of spectra collected revealing the spectrum below. usually. The firing cards were changed but did not solve the problem. These units usually have small bearings which can be monitored in the same manner as any bearing. from the back of the motor. The motor had been uncoupled from the belt to see if the motor vibrated when running solo.

readings at one times the motor RPM will rise in amplitude. Ensure that your data always has the RPM stored with it to prevent mistakes when looking at your data for analysis. peak or even higher in extreme cases. once again. these displays are seldom accurate. 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. Another frequency associated with the commutator is the number of slots on the commutator times the motor RPM.3 ins. Although the significance of this frequency has not been specifically related to a problem. sometimes reaching 0. The use of either a good strobe light.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. Never assume that the RPM of the system is what the panel display reads. usually made of a carbon alloy. or a digital tachometer will give an accurate RPM. the brush wear would. be suspect. Brushes. 42 . It has been observed that as these brushes wear./sec. When the brushes arc it has been observed that these one times RPM readings will increase dramatically. ride against the commutator and supply the DC power to the commutator.

600 cpm separation. which had what was initially suspected to be DC electrical frequencies with a 3. be sure that the resolution of the spectrum is enough to see the difference between closely positioned frequencies.Vibration Analysis on Rotating Equipment When determining a frequency relationship to a machine component. Figure 36 FFT from a 5 HP motor .Electrical Frequencies Vibration analysis was requested on a 5 HP DC motor which was causing problems on a production line.700 cpm.606 = 3620 cpm). A quick reference reading of G/SE (high frequency pulse detection) was taken and registered a higher than normal value. The frequency separation followed the RPM proportionately indicating a mechanical problem.full wave rectified 43 . If necessary take multiple spectra at different resolutions to capture and differentiate all possible frequencies. this is not characteristic of an electrical problem. Remember. Example of Mechanical -v. The spectra was collected revealing the FFT in figure 33. This is a powerful analysis tool when determining a mechanical or electrical frequency. The motor had a SKF 6208 bearing which has a 3. The motor RPM was reduced by 25% from 1004 RPM to 756 RPM and new spectra collected.600 cpm. 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. What was assumed to be a 3600 cpm separation was due to a bearing outer race defect. a 25% difference from 3. The motor is one of four motors which must work in sequence for the line to run efficiently. mechanically related problems will follow the speed increases or decreases proportionately and electrical frequencies will remain constant.606 BPFO multiplier (1004 RPM x 3. The frequency separation was now 2.

Vibration Analysis on Rotating Equipment Figure 37 Same motor .speed lowered by 25% 44 .

The speed factor for calculating radial load at 2.F) By doubling the speed to 2000 rpm we have quadrupled our out of balance forces to almost three-quarters of a ton force. This bearing has a static load rating of 23.5 x (2000/60. Let us see what happens if the impeller is running at 2000 rpm.3]/9.722 x 0.680 lb.966x 0.920 lb.3]/9.000 rpm as calculated using the 3AFBMA method of evaluating load ratings. at 1.I. 2 3 Data taken from TRW service catalog Form 382-14 Anti Friction Bearing Manufacturers Association 45 . A deep groove Conrad bearing for a shaft of about 3 inches diameter could be a MRC 215-S.68 kgF (369 lb.5 x 104.7937 = 2.2π)2 x 0.3 meter from the center (we will use S. Let us imagine that this impeller has an out of balance corresponding to 0. By increasing the speed to 2.Vibration Analysis on Rotating Equipment Rotating Equipment Imbalance Let us consider a fan impeller of 50 kg weight which is running at 1000 rpm.7937 so our load is now 3.5 x (1000/60⋅2π)2 x 0.680 x 0.81 = 670.5 x 10.81 using S.81 = [0.5 kg at 0.I.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.F) In other words we have added over an eighth of a ton to the apparent weight of the impeller. which is in the MRC mid range of bearings.3]/9.000 rpm is 0.71 kgF. units here to make the math easier).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. units F is force in kg.81 = [0.3]/9. Newton says that: F = (m ω2 r)/9.81 = 167. (1476 lb. = [0.

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 .

50 . 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. 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 .Vibration Analysis on Rotating Equipment Figure 46 Imbalance slide 9 To confirm a suspected imbalance check the time waveform. Out of balance may occur in more than one plane. Most spectrum analyzers have a two plane balancing function built into the software.

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 .

However.PM. 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. Therefore this should be checked in the event that the problem cannot be traced to resonance. it was necessary to balance the fan operating under its normal aerodynamic conditions. After the fan had been satisfactorily balanced. high amplitudes of aerodynamic vibration are sometimes encountered on cooling tower fans. Vibration due to aerodynamic forces can also occur at a frequency equal to 1 x fan R. described near the end of the manual should be carried out to determine the resonant part. the force of aerodynamic imbalance can be compensated by following normal balancing procedures. the amplitudes of vibration resulting from aerodynamic forces will be low and no cause for concern. however. Normally. this produced a significant change in aerodynamic conditions and the result was a significant increase in vibration. Many of these fans consist of a drive motor. For example. 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. 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. If it is confirmed that a condition of excessive aerodynamic vibration is not due to resonance. Unfortunately. coupled to the fan gearbox by means of a long torque tube or drive shaft. the access doors were replaced. 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. 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. the fan should be checked carefully for obstructions that may disturb the smooth flow of air through the fan.PM. it often occurs that changing the fan load will produce a corresponding change in the vibration at 1 x R. mounted outside the fan venturi. excessive vibration at the aerodynamic frequency can sometimes result if the fan rotor is positioned eccentrically in -the f an housing. On centrifugal fans.PM and will appear similar to normal imbalance. a centrifugal fan was balanced with the access doors in the fan housing removed to simplify the addition and movements of trial weights. For example. If the fan operates under a constant aerodynamic load. In this case. This aerodynamic imbalance will result if the fan blades do not have the same track or pitch. 59 .Vibration Analysis on Rotating Equipment Vibration due to aerodynamic forces Fans and blowers will inherently have some vibration due to aerodynamic forces.

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. blade configurations and other factors. similar to the rotating oil whirl in a plane bearing. In general. When this occurs. Since rotor whirl generated by aerodynamic cross coupling is excited by -the compressed air or gas. number of compressor stages.Vibration Analysis on Rotating Equipment Figure 61 Aerodynamic forces Aerodynamic cross coupling Aerodynamic cross coupling is a problem occasionally encountered on turbines. In most cases. Cases have been reported where sub-multiples of 0. a reversal of gas 60 . The problem of surge occurs when. These requirements are met by selecting rotor speed. mass flow and rpm. for a particular operating speed.33 or perhaps 0. 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. 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. Typically a compressor is designed to deliver air or other gases over a specified mass flow range and at a specified pressure ratio. Attempting to operate the unit outside the design range can result in excessive vibration and damage to the machine. If the rotor is bowed slightly. The frequency at which this whirl occurs can vary from one machine to the nextThe 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 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.5. 0.25 times the rpm of the rotor have been excited. it seems logical that the condition would be affected by machinery load. The manufacturer of the compressor can supply performance characteristics curves showing the range of stable operation in terms of pressure ratio. the delivery pressure to inlet pressure ratio is too high or if mass flow is too low relative to design conditions. 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 vibration frequency will be less than the rotating frequency.

Of course. if this condition is allowed to continue. This is caused by the turbulent flow within the compressor exciting the various natural frequencies of the rotor wheels. casing. extensive damage to the compressor c result. rotor blades. flow etc. When this occurs a noticeable drop in efficiency and pressure ratio can be seen.Vibration Analysis on Rotating Equipment flow in the compressor will result. In the initial stages of surge. In cases of mild surge. at full surge the gas flow reverses its direction and flows from the discharge to the inlet. The vibration characteristics of choking are essentially the same as those encountered during surging. shaft and other components. When discharge pressures are low. temperature. 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. erratic vibration usually covering a rather broad frequency range. The vibration characteristics resulting from compressor surging can vary depending on the extent of the problem. along with an increase in vibration due to the turbulent flow within the compressor. Rotating stall shows itself as a vibration at approximately 35% of run speed but is dependent on the physical configuration of the compressor. 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. 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. A check of other operating parameters such as pressure. 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. the flow reversal may only occur in the boundary layers of the rotor blades (a rotating stall) however. 61 . should be undertaken to distinguish between the two. Choking occurs when discharge pressures are too low. diffuser blades.

As the lubrication mechanism starts to break down the lubricant loses its ability to separate the roller from the race. THERE IS NO METAL TO METAL CONTACT IN A PROPERLY LUBRICATED BEARING RUNNING AT NORMAL LOAD. an increase in bearing load. So high. The lubricant wedge itself. This may be due to a deterioration of the lubricity. Elasto Hydrodynamic Lubrication Let us consider the lubrication of a rolling element bearing. Let us consider the lubrication of rolling element bearings. 62 .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. that the contact point of the roller and the race (the contact ellipse) becomes elastically deformed. Figure 62 Elasto-hydrodynamic lubrication Rolling element bearings have four stages of failure: • • • • Stage 1 Stage 2 Stage 3 Stage 4 Lubrication problems Marking of the raceways and / or rolling elements Spalling of the raceways and / or rolling elements Collapse of the bearing We will discuss each stage of the failure and discover how each stage may be identified with predictive maintenance techniques. trapping a very small amount of the lubricant into a “wedge”. overheating of the bearing or a fatigue failure of an old bearing at the area most prone to elastic deformation. As the roller rotates the pressure point is very small so the pressure loading is very high. in fact. is very small but is large enough to keep the roller physically separated from the race by a small distance (in the micron range).

4 Ultra sonic . While this is fairly common on lightly loaded motor bearings. 63 . In order to avoid the temperature effect due to over packing. starved bearings usually wear into a condition of excessive looseness and fail without appreciable temperature increases. bearing manufacturers usually suggest that bearings are packed with grease to between 15% to 20% of the bearing's free volume. press main drive motors will usually fail catastrophically. For a rolling element bearing these frequencies will be in the range of 1 to 4 kHz. 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. Figure 63 Loss of Lubricant . Under-packed bearings. We know that over packing is the most common cause of raised bearing temperature. Both of these frequency ranges do not experience any modulation at this stage as the metal to metal contacts are irregular. Low speed. or bearings which have lost grease due to physical migration. Set at above 20 kHz for industrial applications. may generate high bearing temperatures when running at high speeds.just like hitting a bell with a hammer. this can be seen to have a very significant effect on bearing health. Because of this.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. The metal to metal contacts also generate 4ultra sonic frequencies at between 30 and 50 kHz. the metal to metal impacts excite the resonant frequencies of the bearing .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.above the human audible range. which leads to reduction in grease life and eventual failure.

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

As the rollers repeatedly impact onto the small crack.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. At this stage of the bearing failure. the very small area produces very high point loading so the material starts to flake off around the crack. 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. Literally a stress related failure of the material which results from cyclic stresses due to operation at high loads. the velocity vibration becomes apparent at the defect frequencies and harmonics. As the crack propagates. more and more material is removed until the crack becomes a visible cavity or spall. 65 . particularly not if there is any air movement around the surface of the bearing or bearing housing. possible also with sidebands of run speed and / or cage frequency.

or are trapped in the bearing.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. 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. and eventual machine failure. growth in spalls. In these latter bearings.Vibration Analysis on Rotating Equipment Figure 66 Early Fatigue . in oil-lubricated bearings. as is common in sealed and grease-packed bearings. The metal contaminants or wear particles removed from the bearing during spalling are either washed out with the oil. Figure 67 Developed Fatigue on Roller Bearing Courtesy of the Torrington Company 66 .

the regularity.6 * n BSF = (RPM * PD)/2BD * [1 .4* n BPFI = n * RPM/2 [1 + (BD/PD) cos θ] approximated by RPM * O.(BD/PD)2 COS2 θ] (no valid approximation) where PD θ n BD = ball (roller) diameter = pitch diameter = contact angle = number of balls (rollers). it is suggested that the approximate equations be used to establish the bearing frequencies of interest.45 (rotating inner race) RPM * 0. The following figure illustrates the bearing geometry used in the above equations 67 . When bearing geometry is not known but the number of balls or rollers can be counted or estimated.Vibration Analysis on Rotating Equipment Bearing Defect Frequency Calculation It is important to understand that as a bearing with a damaged surface rotates..(BD/PD) cos θ] approximated by RPM * O.55 (rotating outer race) BPFO = n * RPM/2 [1 . All the equations listed above show a direct dependence of the calculated frequency on the frequency of rotation. 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. and are known as:• • • • • Cage or fault train frequency Ball pass with respect to the outer race = BPFO Ball pass with respect to the inner race = BPFI Ball rolling about its own axes Shaft frequency of rotation = BSF = RPM = FTF The following equations are used to calculate these frequencies FTF = RPM/2 [1 .(BD/PD) cos θ] approximated by or (rotating inner race) RPM * 0. A number of characteristic frequencies are generated by a damaged bearing .

Vibration Analysis on Rotating Equipment Figure 68 Ball Bearing Terminology 68 .

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. The severity chart in figure 69 is applicable for motors and fans running between 500 to 3600 rpm.5 kHz has risen steadily.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. For machines that are outside the speed ranges quoted the amplitudes will be lower for slower machines and higher for faster machines. We must also wait until the defect frequencies with sidebands and/or harmonics show up in the velocity spectrum. The envelope (or demodulation) amplitudes are quoted in dBG re 0.001G. This frequency is typical of the bearing resonance. Notice that the vibration at about 1. Velocity readings are heavily affected by the mass of the machine so care should be taken when assessing very small or very large motors. 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).not the derived peak seen in the frequency spectrum. 5 69 . 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.Vibration Analysis on Rotating Equipment Figure 69 Waterfall of early damage to a motor bearing collected every 1. True peak is defined as the peak seen in the waveform . Don’t forget that roller bearings will stand more impactive forces than ball bearings.

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 .

If a plug is fitted opposite the grease fitting make sure that you remove it. A simple rule of thumb for bearing condition is that if the vibration goes down and stays down.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).Vibration Analysis on Rotating Equipment is a raised carpet level. the bearing only had a lubrication problem and you have just fixed it. If the bearing is fitted with seals then it is sometimes possible to force some grease pass the seal with a grease gun. If the vibration level goes down but rises again then the bearing is damaged . In the case of a main drive motor the bearing is almost certainly grease lubricated. 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 grease lubricated bearing may be of shielded (or sealed) construction or non-shielded construction. Figure 72 Demodulated acceleration spectrum of a marked bearing 71 . If the bearing is non-shielded then the bearing should be lubricated following the bearing and motor manufacturers instructions. the chances are that the bearing is starting to suffer lubrication problems. 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. 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.

Vibration Analysis on Rotating Equipment 0. At this stage of the bearing deterioration it is sometimes still possible to save the bearing with additional lubrication.000 1.872 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). 72 . 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.000 0 Time Waveform mS 160 Figure 74 Time waveform from a marked bearing. 1. At the very least you will extend the life of the bearing.

we can correlate this time difference with the frequencies seen in the frequency spectrum. Generally.Vibration Analysis on Rotating Equipment 2.0463 Figure 76 Velocity spectrum from a spalled bearing 73 . 0.500 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.500 2. Looking at the time interval between the peaks in the time domain. 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.

74 . then that bearing is spalled. Figure 75 shows a velocity spectrum from a spalled bearing with multiple harmonics of the outer race defect frequency. irrespective of amplitude. As a general rule of thumb.Vibration Analysis on Rotating Equipment As the bearing deteriorates the bearing defect frequencies start to show up in the velocity spectrum. if you see the same defect frequencies in the demodulated acceleration and the velocity spectra.

or other type of key phasor) for phase measurements. In-place Balancing eliminates the need to disassemble the machine. However. The vibration must be due to imbalance. certain conditions must be satisfied. 75 . The following pages deal directly with this application. 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. the thicker web. In-place Balancing is a straight forward process which involves following a few simple rules. we must be able to make the weight corrections on the rotor. The faults shown are exaggerated but could easily exist in almost every rotating part manufactured.Vibration Analysis on Rotating Equipment Balancing Imbalance has been named the most common cause of vibration in our machines. Illustrated are some reasons for an unequal weight distribution. Before we can balance a part with the Vibration Analyzer. Figure 77 Sources of imbalance Imbalance is the unequal distribution of the weight of a part about its rotating centreline. 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 process of balancing a part without taking it out of the machine is called In-place Balancing. The blow hole in the casting. balance under artificial conditions. operating as it normally does. before we discuss balancing we should first understand imbalance. where it comes from and what must be done to correct it. and assures smooth operation of the machine when you are done. 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. transport the part to a balancing machine. Since this is true it would be to our advantage to be able to correct this condition as easily as possible.

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. The example illustrated is a very common one.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. 76 . An equal amount of weight at the same radius but opposite the heavy spot will balance the rotor. 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.

G. The centre of gravity or mass centre is the point about which all of the weight of a body is equally distributed. 77 . more commonly now. Three ounces located three inches from the centre would be nine ounce.inches.ounce pulley on a shaft with a two-thousandths clearance between the hole and the shaft. gram-centimetre. 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. gram-inches or. The mass centre of a part is more commonly known as the centre of gravity or C. Such a condition could very easily be duplicated by installing a 100.Vibration Analysis on Rotating Equipment Figure 80 Units of measure of imbalance The units for measuring imbalance are generally ounce-inches. The example illustrated shows a condition of mass centre displacement of two thousandths of an inch. The resulting imbalance is twotenths of an ounce-inch. One ounce-inch of imbalance would be a heavy spot of one ounce in weight one inch from the rotating centreline.

therefore. We have learned that the more imbalance we have the greater the force. In the illustration the part has an imbalance represented by a heavy spot located some distance from the centre of rotation. We will use the position of the reference mark as seen by the strobe light to tell us the location of the imbalance. 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. 78 . 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. So we will use the amount of vibration to indicate how much imbalance we have. For instance. “F'' is the force in pounds'. at 3600 RPM nearly 23 pounds of force is generated for each ounce-inch of imbalance. 1. When the imbalance is given in terms of gram-inches the formula becomes F = 1/16(RPM/1000)2 gram-inches. the rotating speed of the part and ounce-inch the amount of imbalance. RPM. the force generated can be computed by the formula F = 1. the greater the vibration.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.77 (RPM/1000)2 ounce-inch. If the heavy spot and distance are in ounces and inches.77 is a constant required to make the formula dimensionally correct. The size of the force generated depends on the speed of rotation and the amount of imbalance.

If we move the weight 30 degrees counterclockwise from its first position we will still see the same amount of vibration. 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. 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. The third thing that can happen is the usual one where the trial weight is added neither at the heavy spot or opposite it. How can we use this information to balance? How to Balance . What we must do is find a weight and position for that weight which would oppose the original imbalance. The change caused by the trial weight can be used to learn the size and location of the original imbalance. if we are lucky we might add the trial weight right on the heavy spot. Starting with the part out of balance. 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. we see 1. Let's start by adding a trial weight. If the trial weight were smaller than the imbalance. The reference mark shifts in the opposite direction to a shift of the heavy spot. When this happens the reference mark shifts to a new position and the vibration displacement changes. 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.crease and the reference mark will appear in the same position it did on the original run. Note that the reference mark appeared at the same place each time. 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. 2. The angle the reference mark shifts is equal to the angle of shift of the heavy spot.creased or decreased to be equal but opposite the original imbalance? 79 . The amount of vibration is proportional to the amount of the imbalance. By what angle should the trial weight be shifted. 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. in which direction. 10 mils. 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. The new total imbalance in the part will be represented by a new amount and phase of vibration.it is correct to use the amount of vibration to tell us how much imbalance is in the part. 10 mils. 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. If we do. The experiment tells us two things: 1. The amount of vibration is 5 mils and the reference mark appears at 12: 00 o'clock. This experiment shows . To balance all we would have to do would be to add more weight until we reached a satisfactory vibration level. We change the original imbalance by adding a trial weight to the part. but the reference mark will now appear at 11: 00 o'clock or 30 degrees counterclockwise from where it was before. The part will vibrate a certain number of mils and the reference mark will appear at some definite position when the strobe flashes.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 second thing that could happen is we could add the trial weight in exactly the right place opposite the heavy spot.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. First. 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 we would reduce the trial weight amount until we reached a satisfactory vibration level.8 mils of vibration and a reference mark position at 12:40 o'clock. the vibration will in. Three things can happen. and should its amount be in.

line A. 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. Then parallel to the original imbalance vector and from the end of the original plus the trial weight vector. We draw a line from the centre at the same angle we saw our reference mark. 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. Our original imbalance was represented by 1.30 o'clock. When the reference mark after adding a trial weight moves clockwise from its 80 . Parallel to line A we draw a line outward from the centre of the graph. Remember our experiment where we learned that the reference mark moves in the opposite direction to that of the heavy spot. Using polar coordinate graph paper we plot our original imbalance vector. Each hour is equal to 30°. A vector is simply a line whose length represents the size of the imbalance and whose direction represents the angle of the imbalance. until it crosses line B.30 mils and a phase of 2. The direction we must move the weight. Figure 83 The vector diagram To solve for the trial weight vector we must complete the vector parallelogram.8 mils and a phase of 12:40 o'clock. however. Next we draw the original plus the trial weight imbalance vector “O + T'' to the same scale. Line B then represents the vector for the trial weight. we draw another. We'll call this vector “O” for the original imbalance. This is all the information needed to solve for the effect of the trial weight alone. It is drawn at the new angle for the original plus the trial weight imbalance phase 2: 30 o'clock.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. O + T. The direction in which the trial weight acts with respect to the original imbalance is represented by the direction of line B. we should adopt this rule. line B. 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. The length of line B represents the amount of the trial weight. is not clockwise as shown here but is counterclockwise. We select a convenient scale for the length of the line such as 1 mil equals 2 divisions. We see this as a new vibration displacement and reference mark position or phase. line C. Therefore. 12: 40 o'clock. After adding a trial weight the imbalance due to both the original plus the trial weight is represented by 2. 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.

measure the angle between “O'' and “T''. 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. Figure 84 Simplified vector diagram Our vector method of balancing then becomes quite simple. The correct balance weight is equal to the trial weight x O/T. the angle between the original imbalance vector and line A. connect the ends of the two. By following these instructions carefully the part should now be balanced by moving the trial weight 75 degrees counterclockwise and reduce its size. We know this because the trial vector. and A. We draw the “O'' vector then the O + T vector. However. and this becomes the “T'' vector or Trial Weight Vector. in shifting the weight. 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. and adjusting the weight to the proper amount can result in some remaining vibration still due to imbalance. O. is the same as the angle between the trial weight vector. O + T. Therefore. Furthermore.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. and its position to get the desired effect to balance the part. In the example shown the trial weight is larger than it needs to be. The direction of this angle is opposite from the shift of the reference mark on the trial run. To solve this problem measure the lengths of line “O'' and line ”T''. to solve the vector problem we actually need only the three lines. the dotted line. or line A in the more complex vector diagram shown above. 81 . very small errors in measuring the phase angle. line B. 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. line B. is longer than the “O'' vector.

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 1. Original Reading First Trial Reading 5.0 mils 3.0 mils @ @

Phase 3: 00 o'clock 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 2. Original Reading First Trial Reading 3.0 mils 4.0 mils @ @ Phase 7: 00 o'clock 7: 00 o'clock

3. Original Reading First Trial Reading

9.0 mils 4.5 mils

@ @

10: 30 o'clock 4: 30 o'clock

4. Original Reading

18.0 mils

@

9: 00 o'clock

84

However. Balancing in One Run Do not let the title of this section fool you. 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. The imbalance corrected was therefore 40 gram-inches.0 mils @ 9: 00 o'clock 5. After the part is balanced to a vibration of 0. If the system is changed in any way. Knowing how much weight to add is important. how much and where weight must be added will also change. 24 gram-inches is required. The next time it is necessary to balance this particular unit or one just like it we will know 8 graminches of correction is required for each mil of vibration. Original Reading First Trial Reading 9. This is the original reading which.5 mils @ @ 2: 00 o'clock 9: 00 o'clock 6. To show how this is done we will use the motor and pulley system illustrated before.0 mils 9. Therefore. Then try some real problems on a simple system like the one pictured here in the text. For example if we measure 3 mils. and with the vibration pickup in the same place. Original Reading First Trial Reading 9.0 inches was required. of course.0 mils 1. This is true of course only for the machine mounted the same way.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. 85 . After the corrections indicated by the readings above have been made a new reading was taken indicating additional correction is required. 12: 00 o'clock Determine the corrections required.Vibration Analysis on Rotating Equipment First Trial Reading 3. lf the weight is added at two inches 12 grams will be needed.0 mils @ @ 2: 00 o'clock . 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. We will assume that it is the pulley that is out of balance.0 mils of vibration. operating at the same speed. This information can be learned for any system but applies only to the individual system. The pulley's position with the reference mark at 2: 00 o'clock is the position of the pulley when the strobe light flashed.1 mil we can learn how much weight was added to correct for 5. Second Trial Reading 3.0 mils of vibration or 8 gram-inches per mil. 40 gram-inches of balance correction was required for g 5. Before a trial weight was added to the pulley the phase or position of the reference mark was 2:00 o'clock. 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.1 mils with the reference mark stopped at 2: 00 o'clock. To date no method has been devised to permit balancing in only one run the first time a unit is balanced. for the example is 5. but where is even more important. We will assume that 20 grams of weight added at a radius of 2. The heavy spot or point of imbalance is 180° away or at 11: 00 o'clock when the strobe light flashed.

The reference mark is simply a way to see the position of the pulley 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 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 location of the heavy spot with respect to the position of the pickup when the strobe light flashes should be apparent. So that you can be sure you understand we suggest the following experiment. 86 . Balance a pulley mounted on a motor similar to the one we have used here. Do the same thing with the heavy spot in other locations. These principles are also useful in two plane balancing. 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. After these relationships have been learned about a part then many parts may be balanced on a production basis. Now put the pickup in a new location and repeat the experiment.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. 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.

put a thin layer of grease on the accelerometer and a thin layer of epoxy adhesive on the top of the tap block. Drilled & tapped 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. Width across flats should be just wider than the base of the accelerometer. 1/4” Figure 89 Typical tap block for mounting an accelerometer When it comes time to mount the accelerometer. for this reason the material quoted is bright steel or austenitic stainless. Whichever adhesive is used. The motor mounting surface should be prepared with a hand grinder or similar to remove all paint and get down to bare metal. 87 . This block should be machined with a “gramophone” finish on the reverse side to give good adhesion to the mounting surface. holding the tap block in place while the adhesive is setting with duct tape or similar. stock so that if it is necessary to remove the block then it can be wrenched off with a suitable spanner. Be very careful not to over-tighten the accelerometer onto the tap block . 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. it must set hard in order to transmit the vibration.follow the manufacturers torque settings.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. The block is designed for mounting to the surface with an epoxy adhesive such as Loctite DEPEND or similar. 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. Notice that the tap block is made from hex. 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. This will help the transmission of vibration through the tap block and allow for easy disassembly. The tap block should be mounted without the accelerometer and allowed to set firmly.

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. Do not connect both ends of the shield to ground as you will get beautiful ground loops that will really dominate your signal.Vibration Analysis on Rotating Equipment Particularly on motors.the end furthest away from the motor .5 volt drop for a 10 mA signal and 5 volt at full speed (20 mA). keep the cable as far away as possible from the motor power cable to avoid “cross-talk”.L. To use this output the 4-20 mA must be dropped across a resistor of about 6250 Ω to give a 2.on to a good electrical ground. Failure to do this will result in high amplitude multiples of electrical line frequency (60 Hz). care should be taken to ground the shield wire at one end only .L. 6 V = I. is not adversely affected by the resistor. 88 .L. Before installing this resistor check with the engineers responsible for the drive system to make sure the drive system or the P..L. which gives a 4-20 mA output. R is ohms. triggered by a key or from the installed tachometer channel on the P.R where V is volts. pulse.C.T. once per rev from an installed proximity switch.C. A speed output should be installed for variable speed motors. I is amps. Case isolated accelerometers are usually less susceptible to picking up electrical frequencies. This could be in the form of a T.T. When running your accelerometer cable back to the junction box. pulse.

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 .

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 . 100 Hz + 4 0 + 3 0 + 2 0 + 1 0 0 -1 0 -2 0 1 .Vibration Analysis on Rotating Equipment Thick double sided adhesive disk Double sided adhesive disk Thin 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 P ro b e D u a l R a il M a g n e t F la t M a g n e t M o u n t in g P a d A d h e s iv e M o u n t S tu d M o u n t Sensitivity Deviation (dB) ~ Ref.

Examples could be: lubrication problems. 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. Base cause As the name suggests this is the root cause of any potential failure. 91 . Failure type This is the failure that we could expect from the base cause. for example misalignment could cause a bearing failure or a shaft breakage. misalignment. 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. manufacturing defects and so on. 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 base cause often branches out to more than one failure type.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. 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.

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. but what type of vibration is best? In this case the best vibration parameter is velocity. The heat exchanger temperature change may be detected by a mercury-in-glass thermometer or alternatively by the use of infra-red technology. An example would be that a misalignment would show itself as an increase in vibration at the machine. At this stage you need to have detailed understanding of the failure modes and predictive maintenance applications.Vibration Analysis on Rotating Equipment External manifestation This is where we start getting technical. We have already said that misalignment shows itself as a rise in vibration so the technology to use would be a vibration based technology. 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. 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. 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. Technology So now we know how the problem will show it’s ugly head but what is the best technology to detect it. Parameter So the technology of choice for detection misalignment is vibration. 92 . 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. 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. 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. 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.

5 .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. Hamming or Kaiser = 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. 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.5 kHz (for electric motor bearings up to about 250 HP) = 400 or 800 = Hanning. above. Stage 1 Notice that the first stage of the defect is linked to lubrication problems. It shows the four stages of a rolling element bearing failure. 93 . 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 Envelope filter No lines Window Averages = BPFI x 8 (or thereabouts) = 2.

Note that if you are using seismic velocity transducers you will not be able to look this high in the frequency range. Hamming or Kaiser = 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. Usually 1. We may also see sidebands of run speed or cage frequency around the defect frequencies or their harmonics. the appropriate Fmax would be: Fmax = 1024 0.5 kHz (do not confuse AC motor frequencies with a brg defect) = 400 or 800 = Hanning. To calculate the Fmax in the point setup screen we use the following formula: Fmax = No.024 points. So the choice here would almost certainly be ESP. At this stage we start to see the defect in velocity usually at the third or fifth harmonic of the bearing defect frequency.656 Hz So we would set the Fmax to 2. At this stage the bearing overheats dramatically and very quickly.2 milliseconds (mS). If we expect resonance to occur at about 2 kHz then the Fmax should be about 3 kHz. The velocity reading should be set up as follows: Fmax or No lines Window Averages = BPFI x 8 (if also using ESP or similar reading) = 2. 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. Temperature measurement using embedded thermo-couples have been used 94 . If we require a resolution of 1.0502 x 3 x 2. of samples time for 3 revs x 2.024 samples is adequate resolution.Vibration Analysis on Rotating Equipment Stage 2 The next stage of failure involves light marking of the bearing. The ESP reading should be the same as that set up in stage 1. The setup for the ESP reading would be the same as in stage 1. We should be taking an ESP and a velocity reading.56 = 2.0502 seconds or 50. Stage 3 The third stage of the bearing failure is when the bearing starts to spall.5 kHz or the nearest available frequency. Stage 4 The final stage of failure is when the bearing physically collapses. The time duration should be calculated to be enough to show about 3 revs of the shaft. For a bearing on a shaft running at 1195 rpm the time for on rev is: 60/1195 = 0.56 Example. 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. 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.

5) Bad SCR is colder Unbalanced supply Hot spot on stator On Line Looseness Run speed harmonics & subharmonics Vibration @ SCR Frqys Changes in SCR temps. which is operating under these particular conditions. 2.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.C. See sct. Vibration Velocity On Line Bad S. Not only have we considered the possibility of bearing failure but we have included insulation breakdown. Ron Frend .5 Peaks @ 12 G (roller) 7 G (ball brg) Interval On Line Set Up Fmax = BPFI x 8 Radial G acceleration = 3 revs Fmax = BPFI x 8 (higher if not using resonance parameter) Radial = 3 revs G acceleration Fmax = 10 x rpm Axial Fmax = 10 x rpm Vertical Fmax = 120 kcpm Drive End Compare SCR to SCR Compare phase to phase 1 baseline On Line On Line Spalled bearings Bearing Failure Harmonics of bearing defect frequencies Vibration Velocity Time Waveform On Line Misalignment Bearing Failure/ Shaft Failure Bearing Failure/ Shaft Failure/ Structural Failure Reduced Power Motor Burnout High axial vibration Vibration Velocity Multiples of run speed axial > radial Multiples of run speed up to 15x. 2.C. & harmonics and/or sidebands.2. We now include all of the information from our failure study of the rolling element bearings into our potential failure analysis tree (figure 2). rotor bar defects.PreDiCon Figure 94 PFA for a main motor 95 . may eventually suffer. s/band on DC(Sect. rise/ Thermography Uneven temp. 2. loose components and even drive problems on D. Including the Component Failure in the PFA Tree. 1/3 DC pulse. We have carried out potential failure analyses on all of the possible defects that we think a motor. 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 Base Cause Dry/Marked bearings Failure Type Bearing Failure External Manifestation High frequency resonance @ bearing defect frequency intervals Technology Vibration Parameter Demodulation (Ultra Sound) Time Waveform Analysis Spikes @ Brg defect frequencies (20 dBG) See sect.R.2) Sidebands of 1x (Sect 2. Possible sub-harmonics 240 Hz.5 MΩ Sidebands of slip x No. distribution Earth current leakage Insulation test Should be > 1. Notice that the bearing failure is only one of the potential failures that we now know of.5 Peaks in waveform Spikes @ Brg defect freqys. of poles around 60 Hz (Sect.2) 6 month 6 month Motor de-energized Press idling Rotor defects Rotor failure Line frequency + sidebands Vibration @ rotor bar pass frequency Motor load Motor current current analysis frequency Vibration Velocity Motor burnout On Line Fmax = RBPF x 3 Radial c.s Vibration Thermography Velocity Thermal Image Amps Thermal image (spot temp) Megger On Line 6 month 6 month 6 month Winding defects Motor burnout Different current / Current Draw input phase Stator temp. motors.

CCTV (reference marks req'd) DRIVE ASSEMBLY GREASE SEAL FAILURE MATERIAL CHOICE DESIGN ROTARY SIEZURE BEARING COLLAPSE BEARING CLEARANCE INCREASE OVER TEMPERATURE LOADING MATERIALS / LOADING LUBRICATION / LOADING DESIGN DESIGN DESIGN DRIVE SIEZURE FORIEGN BODY INGRESS CORROSION BRINNELLING BEARING DEFECTS SLIDE BUSH SIEZURE LUBRICATION / CLEARANCE MATERIAL COMPATIBILITY / CORROSION DESIGN DESIGN GEAR WEAR / DEFECTS DIRT/MATERIALS CHOICE NOISE / VIBRATION INSPECTION .CCTV FORIEGN BODY INGRESS CORROSION FRETTING/CLEARANCE INCREASE INSPECTION . The engineer or technician defining the measurements must be aware of the limitations of the technology which is used for the measurement. don’t limit yourself to only taking vibration measurements . DISSOLVER FEED CHUTE FAILURE MODES CHUTE ASSEMBLIES LINER WEAR CLAMP/SEAL FAILURE CAM SIEZURE INSPECTION . Finally. 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.there are a host of other technologies out there which complement vibration in a predictive maintenance application.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.CCTV (reference marks req'd) HEAD 1 96 .CCTV (wieght ?) INSPECTION .

Vibration Analysis on Rotating Equipment BEARING DEFECT DETECTION BEARING COLLAPSE OVER TEMPERATURE FORIEGN BODY INGRESS CORROSION PROBLEMS BEARING DEFECTS ULTRA-LOW SPEED PARTIAL ROTATION INNER BEARING MOUNTS BRINNELLING FREQUENCY RESOLUTION NOT POSSIBLE TIME-BASE READINGS REQ'D IN : ACCELERATION STRESS WAVE PROBLEM RADIATION HEAD 2 97 .

5 Peaks in waveform Spikes @ Brg defect freqys. 3.5 On Line Fmax = 10 x rpm Vertical 1 Baseline c. & harmonics and/or sidebands. Ron Frend .Vibration Analysis on Rotating Equipment Press Flywheel PFA Tree Base Cause Dry/Marked bearings Failure Type Bearing Failure External Manifestation High frequency resonance @ bearing defect frequency intervals Technology Vibration Parameter Demodulation (Ultra Sound) Time Waveform Analysis Spikes @ Brg defect frequencies (20 dBG) See sect. 3.PreDiCon 98 .5 Peaks @ 12 G (roller) 7 G (ball brg) Interval On Line Set Up Fmax = BPFI x 8 Radial G acceleration = 3 revs Fmax = BPFI x 8 (higher if not using resonance parameter) Radial = 3 revs G acceleration On Line On Line Spalled bearings Bearing Failure Harmonics of bearing defect frequencies Vibration Velocity Time Waveform On Line Looseness Bearing Failure/ Shaft Failure/ Structural Failure Run speed harmonics & subharmonics Vibration Velocity Multiples ofBPFO or BPFI in velocity Sect. 3. See sct.

or "Uniform" windowing. 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. and noise floor. but it will be periodic (zero at each end). it is as if the spectral lines are wider. If the filter is wide. then only frequencies near the bin will contribute to the bin. amplitude accuracy. on the other hand. If a sine wave is passing through zero at the beginning and end of the time series. BlackmanHarris. since the filter is narrow. 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. and is generally unpredictable for real signals. or side lobes. Uniform The uniform window is actually no window at all. This is done by multiplying the data samples by a "smoothing window" function.Vibration Analysis on Rotating Equipment Measurement Windows Many people get confused with this topic. It may not be identical from record to record. it falls off from center rapidly. the resulting FFT spectrum will consist of a single line with the correct amplitude and at the correct frequency. They start and stop at zero and are smooth functions in between. However. This is true but it's actually a bit worse than that. and the resulting time record is by definition periodic. If. "Flat". If the filter is narrow. While the smoothing window does a good job of forcing the ends to zero. energy in the signal "leaks" from its proper location into the adjacent lines. In the frequency domain. resulting in a discontinuity in the sampled signal. This smearing will also change wildly between records because the amount of mismatch between the starting value and ending value changes with each record. Hanning. The different types of windows trade off selectivity. which can have several different shapes. These sidebands. A narrow filter is called a selective window . If there is no windowing function used. When the time record is windowed. truncation of the waveform will occur. it is necessary that the signal level is forced zero at the beginning and end of the time series. time bin by time bin. Several types of window functions are available including Uniform (none).it selects a small range of frequencies around each bin. The net result of windowing is to reduce the amount of smearing in the spectrum from signals not exactly periodic with the time record. Leakage could be avoided if the time series zero crossings were synchronized with the sampling times. the signal can actually smear across the entire spectrum. then frequencies far from the bin will contribute to the bin amplitude but those close by will not be attenuated significantly. This discontinuity causes problems with the FFT process. and the result is a smearing of the spectrum from a single line into adjacent lines. a window acts like a filter. it also adds distortion to the time series which results in sidebands in the spectrum. windowing algorithms give extra weight to the values near the middle of the sequence. (It is exactly periodic 99 . Windows are functions defined across the time record which are periodic in the time record. and Kaiser. 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. Flattop. In order to reduce the effect of leakage. its points are multiplied by the window function. This means that even frequencies close to the bin may be attenuated somewhat. the signal level is not at zero at one or both ends of the time series record. but this is impossible to achieve in practice. The shape of the "leaky" spectrum depends on the amount of signal truncation. The time record is used with no weighting. The measured amplitude of the weighted signal is also incorrect because a portion of the signal level is removed by the weighting process. A signal will appear as narrow as a single bin if its frequency is exactly equal to a frequency bin. This is called "leakage". What is windowing? Let's go back to the time record. If the time record does not start and stop with the same data value. this is called "Rectangular". effectively reduce the frequency resolution of the analyzer. To make up for this reduction in power.

If its frequency is between bins. The filter is steep and narrow and reaches a lower attenuation than the other windows. With a sufficient number of averages. this window is only useful when looking at transients which do not fill the entire time record. Blackman-Harris The Blackman-Harris window is a very good window to use with the spectrum analyzer. RMS Averaging RMS averaging computes the weighted mean of the sum of the squared magnitudes (FFT times its complex conjugate). Thus.7 dB) than the Hanning. the Hanning window can limit the performance of the analyzer when looking at signals close together in frequency and very different in amplitude.5 dB (for signals between bins) and provides reasonable selectivity. Unlike the Hanning. a very good approximation of the actual random noise floor can be displayed. combines excellent selectivity and reasonable accuracy (about 0. it is the best window to use for measurements requiring a large dynamic range. It has better amplitude accuracy (about 0. This allows signals close together in frequency to be distinguished. Hanning The Hanning window is the most commonly used window. displaying the real or imaginary part or phase of an RMS average has no meaning. it will affect every bin of the spectrum. signals appear wide but do not leak across the whole spectrum.8 dB for signals between exact bins). Since RMS averaging involves magnitudes only. (The real part is averaged separately from the imaginary part. averaging many spectra together improves the accuracy and repeatability of measurements. Averaging In general. Vector (Synchronous Time) Averaging Vector averaging averages the complex FFT spectrum. the Flattop window has a wide pass band and very steep rolloff on either side. RMS averaging reduces fluctuations in the data but does not reduce the actual noise floor. Because of these properties.Vibration Analysis on Rotating Equipment within the time record).02 dB. As a result. In general. However. which is available on IRD analyzers. Kaiser The Kaiser window. Its filter roll off is not particularly steep. Its between-bin amplitude variation is about 0. 100 . These two cases also have a great deal of amplitude variation between them (up to 4 dB). Flattop The Flattop window improves on the amplitude accuracy of the Hanning window. The RMS average has no phase information. even when their amplitudes are very different. The Kaiser window has the lowest side-lobes and the least broadening for non-bin frequencies. It has an amplitude variation of about 1. The weighting is either linear or exponential.) This can reduce the noise floor for random signals since they are not phase coherent from time record to time record. the selectivity is a little worse. very good selectivity and the fastest filter rolloff.

Averaging takes place according to the formula. And what is the processor doing while it waits? Nothing. the display shows the real or imaginary part or phase of the complex peak value. Remember. further changes in the spectra are detected only if they last sufficiently long. With overlap processing. The resulting display shows the peak magnitudes which occurred in the previous group of spectra. then the new data is stored. Otherwise. 101 . the peak magnitude values are stored in the original complex form. The signal of interest must be both periodic and phase synchronous with the trigger. 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. This is because the complex information is preserved. With narrow spans. Peak Hold detects the peaks in the spectral magnitudes and only applies to Spectrum. With vector averaging. these points are “re-used" and appear as middle points in other time records. With overlap. New Average = (New Spectrum * 1/N) +(Old Average) * (N-l)/N where N is the number of averages. it is the frequency span whose corresponding time record exceeds the time it takes to compute the spectrum. This is done on a frequency bin by bin basis. Peak Hold Peak Hold is not really averaging. it is possible to compute the spectra for every time record with no loss of data. and Octave Analysis measurements. Instead it uses data from the previous time record as well as data from the current time record to compute the next FFT. the points at the ends of the time record do not contribute much to the FFT. Once in steady state. Make sure that the number of averages is not so large as to eliminate the changes in the data that might be important. This is why overlap effectively speeds up averaging and smoothes out window variations. Linear Averaging Linear averaging combines N (number of averages) spectra with equal weighting in either RMS. instead. The spectra are computed in "real time". At larger spans. Exponential Averaging Exponential averaging weights new data more than old data.Vibration Analysis on Rotating Equipment Vector averaging requires a trigger. Thus. and if the new data is larger. Exponential averages "grow" for approximately the first 5N spectra until the steady state values are reached. some data samples will be lost while the FFT computations are in progress. If the real or imaginary part or phase is being displayed for spectrum measurements. Real Time Bandwidth and Overlap Processing What is real time bandwidth? Simply stated. the analyzer does not wait for the next complete time record before computing the next FFT. most window functions are zero at the start and end of the time record. However. At this span and below. the real and imaginary parts as well as phase displays are correctly averaged and may be displayed. This type of averaging is useful for eliminating transients. this could be quite slow. PSD. the real and imaginary parts of the signal will not add in phase and instead will cancel randomly. This speeds up the processing rate. Vector or Peak Hold fashion. The update rate would be no faster than one spectra per time record. the new spectral magnitudes are compared to the previous data.

For vibration measurements.8% is the typical maximum.Vibration Analysis on Rotating Equipment Overlap Percentage The amount of overlap is specified as a percentage of the time record. Octave analysis measures spectral power closer to the way people perceive sound. Octave Analysis The magnitude of the normal spectrum measures the amplitudes within equally divided frequency bins. yet sufficient samples are gathered for a valid analysis. = 1 kHz x 2n The 1/3 octave bands have center frequencies given by: Center Freq. For vibration analysis of rotating machinery a good overlap is 50% as this ensures that no data is zeroed out by the smoothing windows. Octave analysis computes the spectral amplitude in logarithmic frequency bands whose widths are proportional to their center frequencies. Typically the shape of each band is a third-order Butterworth filter whose bandwidth is either a full or 1/3 octave.11 (1986). 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. in octaves. 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. Octave Band Center Frequencies The center frequency of each band should be calculated according to ANSI standard S1. that is. The full octave bands have band centers at: Center Freq. = 1 kHz x 2(n-30/3) 102 . The bands are arranged in octaves with either 1 or 3 bands per octave (1/1 or 1/3 octave analysis). 0% is no overlap and 99.

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. Of Rolling Run Speed x No. Predictive Maintenance Procedure Date:July 11.if not you must change the speed reference before continuing. One common bearing frequency is just over 3 x run speed for BPOR on a 8 element bearing. Of Rolling Run Speed x 0. If the trend is deteriorating then further checks are necessary. 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.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. In the envelope spectrum see if any of the generated frequencies coincide with (or are close to) any significant spikes. Once a spike at a bearing frequency has been identified you should check the baseline for this type of machine for the trend.800 cpm (for a 3600 rpm motor) . • • • • • 3) As a rule of thumb: Ball Pass Frequency Outer Race Elements x 0. 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.4 Ball Pass Frequency Inner Race Elements x 0.4 2) Note that this vibration is not necessarily direction specific. Be careful that you do not confuse a run speed harmonic or an electrical frequency with a bearing defect frequency. If you have frequency information for the machine ensure that the reference speed is accurate . Identify the run speed accurately.6 Cage Frequency ≅ ≅ ≅ Run Speed x No.Vibration Analysis on Rotating Equipment Analysis The following procedure gives an overview of the procedure to take in the analysis of vibration frequency spectra. Activities: 1) Select the first plot of the machine which will be at the driver outboard and will be an enveloped acceleration reading. 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 10-15 dB(G) spike above carpet > 20 dB(G) spike above carpet under lubrication minor marking 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. 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. In the velocity spectrum the following patterns indicate the associated defects:

Dominant Frequency Secondary Frequency Harmonics of Dominant Frequency Sidebands Dominant Direction Defect Suggested Maximum Amplitude @ Dominant Frequency 6 mm/s 3 mm/s 4 mm/s 0.5 mm/s 5 mm/s 7 mm/s 1 mm/s 5 mm/s

6)

7)

1x 1x 1x BPFO / BPFI 7,200 cpm Any Gear Mesh 2x Belt

Nil 1/2 or 1/3 x 3x 1x Rotor Bar Any 1x 1x

Nil Multiple 2 or 3 Multiple 2 or 3 Nil 3 or 4 2 or 3

Nil Nil Nil 1x / Cage 2xLF of RBF Nil 1x N/A

Radial Radial Axial Radial Any Any Radial Radial

Imbalance Looseness Misalignment Bearings Electrical Resonance Gearing Belts

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 “delayedby-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

its phase extends over the full 360 degrees. The phase is relative to the pulse of the trigger. Phase In general. 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. below full scale. This avoids the problem of taking the log of negative voltages. even if a signal is small. The Log Mag display graphs the magnitude of the spectrum on a logarithmic scale.It turns out that the log display is both easy to understand and shows features which have very different amplitudes clearly. This avoids the messy phase display associated with the noise floor. If we wanted it to be 1 inch high on the graph. another way of displaying amplitude is the Log Magnitude. The real and imaginary parts are always displayed on a linear scale. the top of the graph would be 833 feet above the bottom . Why is the Log Mag display useful? Remember that the 16 bit analyzer has a dynamic range of about 90 dB. Imagine what something 0.s. phase measurements are only used when the analyzer is triggered. (Remember. usually from -180 to +180 degrees. The phase is displayed in degrees or radians on a linear scale.Vibration Analysis on Rotating Equipment Spectrum Although the linear magnitude scale is used most often for displaying spectra.) 107 .).01% of full scale would look like on a linear scale.

A complex number can represented graphically by a point in a plane where the real numbers 1. For numerical addition. Consider the vector below. Similarly if we say that “a complex number represents harmonic motion” we imply that the real part of such a number. j3.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. j2. We do not necessarily need to make much use of this number but we do need to know where it comes from. written in the form “a(cos ϖt + j sin ϖt)” represents that motion. 108 . Harmonic motions are represented by rotating vectors. The first use that we will make of the imaginary number is the Nyquist chart which is useful for identifying resonances. a ϖt 0 b φ a+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. 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. Almost all of the algorithms in the analyzer which involve phase make reference to the imaginary number. For complex signals this is often much more convenient. etc.2.3 etc. There exists a simpler method of handling the vectors numerically by employing imaginary numbers. 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. however the vectors must be resolved geometrically which loses almost all of the advantage of ease. are plotted horizontally and the imaginary numbers are plotted vertically. With the notation j = √-1 these imaginary numbers are j. But this horizontal projection is also the real part of the vector.

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.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. 109 . This trigger is often a 5 volt. Whichever is the particular application. Triggering As mentioned previously a trigger may be set up on a machine to control the acquisition of data. The OR25 series analyzer is limited to displaying data at the acquired Fmax or 1/10th or 1/100th. once per rev of the shaft. 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. orbits. This could be a very high speed machine over a short time duration or a slow speed machine over a very long time period. A third type of trigger is a single pulse or voltage change and is known as an event trigger. once per rev pulse or it could be from an encoder giving many pulses per rev. or for Bodé plots. 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. Ensure that the data collection rate is valid for the necessary analysis. which triggers data collection at a specific time. A many pulse per rev trigger requires an external clock input to the collector which is programmed to the number of pulses per rev. Once per rev pulse trigger As the name implies this is a simple pulse. Encoder trigger The encoder will give many pulses per rev and must be input to the external clock input. This function is useful in cascade plots. the collection of data must be considered very carefully before data acquisition takes place.

110 . Figure 98 Bode plots Orbits Displacement sensors measure the amount of shaft displacement. This can be very useful if the subject machine is suffering from a transient fault. The displacement sensors are mounted orthogonally which is so that each sensor is 45° from the vertical. When the trigger is activated the data storage will commence. The Bodé plot is useful when assessing the resonant condition of a machine or even bearing deterioration as in the example below. the total motion of the shaft's orbit within the bearing. then some data collected before the trigger event will be collected as well as some data after the trigger. Bodé plots When the cross channel properties are calculated by the analyzer. or rather. 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. In the first example we see an elliptical orbit which is typical of preload caused by misalignment. they can be displayed as amplitude or phase against speed (Bodé) plots or real-imaginary (Nyquist) plots. A post-trigger will collect data only after the event. The analyzer is actually collecting data all of the time but is only storing as much as the memory will allow. If the trigger is set up as a pre-trigger. The trigger level is set and the analyzer just sits there waiting for the trigger amplitude to reach the predefined level.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.

This can be accomplished by viewing the spectra from any single Eddy Probe sensor. synchronous orbits with the phase angle superimposed on the orbit allows rapid identification of this condition. LOOSE ROTATING PART A loose rotating part can generate unusual vibration signals. or shaft. such as.peak amplitude exceeds acceptable limits. They may fluctuate in amplitude and the phase angle may shift. damping. of the orbit will grow larger until the peak-to.basically anything that can influence the "system" mass.. By superimposing the phase angle input signal onto the orbit a shift in this parameter can be easily determined. A similar diagnosis can be made by viewing the filtered signals from two orthogonally mounted Eddy Probes sensors as orbits.. and sometimes the shape. and/or stiffness. Changes in the filtered amplitude can be determined using orbits analysis. a change in this frequency may indicate the "possibility" of a crack existing. foundation problems. Spectral analysis can be used to identify this fault. synchronous orbit analysis. The "possibility" must be emphasized and carefully analyzed because many other causes can produce these changes. As the balance condition deteriorates the size. and/or a change in the resonance frequency as the machine starts or stops.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. CRACKED SHAFT A crack in a rotor. can generate several different effects on how the machine behaves: a change in the vibration level. loose rotating parts. also. a damaged or loose bearing support. Imagine a 111 . By noting the operating speed at which the resonance frequencies occur. but observing filtered. This fault is diagnosed easiest using filtered. a change in the operating phase angle.

As the machine accelerates the whirl frequency will increase as machine speed increases. but the changes may be too rapid for the sampling rate of the instrument. appears to have several phase marks. also. 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. such as an impeller.Vibration Analysis on Rotating Equipment mass. This is observable using a spectrum analyzer. EXCESSIVE PRELOAD All journal bearing machines have some amount of preload so that a stable oil wedge can be established. This phenomenon creates an individual sub-synchronous frequency which can occur within a frequency range from 35% to 48% of rotor speed. This fault is caused by a condition which prevents the rotor from creating a stable oil wedge on which ride. it can rotate freely on the shaft independently. which has come loose. An improperly designed bearing is the usual source for oil whirl conditions. The display. but a change in the fluid viscosity or machine alignment state are other possibilities. but closer observation of the vibration signals and the machine conditions causing these signals will produce different. OIL WHIRL Oil whirl and oil whip are sometimes listed as a single machine fault. Generally. The orbit display will no longer rotate because the oil whirl frequency has coincided with the first natural resonance. or critical speed. with the phase input superimposed on the display. The orbit will appear to vibrate slightly as this condition is created. The orbit will be more or less round in shape with an amplitude that nearly approximates the bearing clearance. an oil whirl condition precedes an oil whip condition. The inception of a loose part condition will produce a "nervous" filtered. distinct signal displays for each condition. The two phase marks will not be displayed symmetrically on the orbit because the whirl frequency is not at exactly ½ machine speed. 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 part may be slipping and then sticking on the shaft just prior to becoming a full fledged loose rotating part. This display will be round in shape and the amplitude will greater that the amplitude noted during oil whirl. the orbit will appear to have two phase marks on it. synchronous orbit produces a distinctive display. An oscilloscope set up to observe a filtered orbit will sample continuously so that the changes can be seen. Observing oil whirl as a filtered. The preload may be internally or 112 . depending upon the machine/bearing design or construction. be observed using an oscilloscope. and when the phase angle is superimposed upon the display. OIL WHIP Oil whip occurs during the later stages of an oil whirl condition and it has a distinctive orbital display. This characteristic is due to filtering at shaft speed and the fault being generated at a sub-synchronous frequency. Spectral and orbit analysis can be used to identify either condition. and has "locked" onto this frequency. synchronous orbit. 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. The phase shifting can.

After the orbit has been flattened into the "banana" shape a 2X frequency is present on spectra displays. 3X. Spectra displays of rub conditions are characterized by distinct frequencies that occur at multiples of a fundamental frequency. These sources of preload create an elliptical orbit that is flattened in the direction of the preload vector. 1X. 2X. A light rub will produce a "tear drop" shaped orbit. the fundamental rub frequency will coincide with the shaft speed with multiples at 2X. 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. they usually increase the clearances until the rub has been cleared or. etc. At shaft speeds up to twice the natural resonance frequency. 5/3X. RUB A common problem in newly rebuilt or modified rotors is a slight rubbing condition as the rotor is initially operated. if not corrected. they will wear away the internal clearances until the machine cannot be operated. etc. 2X. The severity of the rub will affect the shape of the orbit. 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. 7/3X. These internal loops will have their own phase marks displayed and the loops will be located symmetrically on the display.Vibration Analysis on Rotating Equipment externally produced. Rotor rubs are not a phenomenon which continues over an extended period. Heavy preloads further distort the orbit into a figure eight shape. External preloads may be from coupling misalignment or piping and support system thermal changes. Internal sources of preloads are from gear meshing or hydraulic loading during pumping actions. 4/3X. the fundamental rub frequency will be shaft speed with multiples at 2/3X. Between twice and three times the first natural resonance frequency. As the preload increases the orbit is further flattened. 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. etc. 3/2X. As the rub gets heavier the orbit will flattened and may appear as an excessive preload. The fundamental frequency will depend upon the relationship of the shaft speed to the first natural resonance frequency. the fundamental rub frequency will be ½ shaft speed with multiples at 1X. 5/2X. 113 . As preload increases the shaft centreline will shift in the direction of the preload vector. Between three and four time the natural resonance frequency. with the point of the tear drop coinciding with the impact spot.

the speeds. He knows the complete structure of the machine. the number of teeth on the gears. 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 . Most failures are due to material fatigue from normal operation leading to mechanical failure. a vibration is encountered which does not fall into any of the above categories and the vibration analyst is left scratching his head. Eventually. The vibration consultant.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. on the other hand. the belt and pulley dimensions and still none of the pre-calculated defect frequencies line up with the big spike in the frequency spectrum. 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. This is not the case. Many consultants in the field of vibration analysis will tell you that 90% of all failures are directly caused by vibration 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. however. He knows the bearing numbers. 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. In many cases the big spike is related to a resonance of a natural frequency or perhaps running a machine near a critical speed.

Sometimes the pushes would add to the amplitude of the swing but sometimes the pushes would reduce the amplitude of the swing. You are the adult who pulls the child back and releases the swing so that the child swings back and forth. If the exciting force is not large enough to overcome the damping. which means the pendulum swings back and forth once every 2 seconds. we can see a host of interconnectivity which can give rise to a large number of natural frequencies. The damping acts like the friction in the child/swing system. In fact. In this case the child/swing system is in RESONANCE with the exciting force. however. we mean any part or component (or combination of either) of the machine and/or the structure. repeatedly applied force causes the amplitude of an oscillating system to become very large is called resonance. Any oscillating object has a natural frequency. Looking at the average machine and associated structure. 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. most of the systems have inherent damping. The natural period of oscillation of a pendulum is calculated using the following formula: 115 .Vibration Analysis on Rotating Equipment What is resonance? Imagine a child on a swing. even though there may many natural frequencies. The rate at which the child swings is the NATURAL FREQUENCY of the child/swing system. the natural frequency of a pendulum 1 m (39 in) long is 0. When we say a system. which is the frequency an oscillating object tends to settle into if it is not disturbed. 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. In this case the system is not in resonance with the exciting force. Every time that the child swings back to your position.5 Hz. gear mesh. Mechanical For example. Fortunately for us. Such forces include imbalance. then the amplitude at that frequency will diminish. bearing frequencies etc. 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. If. 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. misalignment. If the pendulum is struck lightly once every 2 seconds. anything which can be measured as a vibration may be considered to be an exciting force. you give the swing a small push – you are the EXTERNAL EXCITING FORCE. Each small push (as long as it is large enough to overcome friction or damping) will add to the amplitude of the swing. Let us consider the pendulum mentioned above. Imagine what would happen if you pushed at a different frequency. 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. The phenomenon in which a relatively small. the amplitude of the swing increases gradually until the amplitude of oscillation is very large.

sin( φ ) dφ = Notice that the time period of the swing is affected slightly by the angle of the pendulum but the effect is only slight.Vibration Analysis on Rotating Equipment π L 4. . such as the one shown below. g 2 1 0 Equation 1 Simple pendulum time period Figure 100 Simple pendulum 1 2 2 x . 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. we have to consider all aspects of the system that affect the natural frequency. 116 . In this case the time period is: f Hz 1 Ts = 1 Ts Equation 2 Calculating frequency from time period In a conventional mechanical system.

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. 2 (a t) 2 Equation 4 Newton's law of motion (x) (x2) (x3) Where s = displacement u = velocity a = acceleration t = time In the single degree of freedom system such as the mass/spring system shown. So: Spring force = "-kx" The damping force is also negative. The forcing mechanism is defined harmonically as Po sin ωt. because the damping is always in the opposite direction to the velocity. and by applying Newton's laws of motion we can determine the instantaneous velocity and displacement. and this could be as simple as someone pushing and pulling the mass with their hand. s u . The sign of the spring force is negative as it is always acting in the opposite direction to the extension. 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.t 1. if the spring force has magnitude kx. Note: Po sin ωt is a combination of direct force "Po" and angular position "sin ωt" where ω = rotational velocity & t = time 117 . being "-cx2". and the extension "x" is zero then there is no force. the spring/mass system will move vertically up and down at a rate which is determined by the spring rate.

Vibration Analysis on Rotating Equipment The forces acting on the mass are: -kx –cx2 + Po sin ωt Newton's law gives: d2x m -----. If we impact one end of the pipe. Sound travels at different speeds in different liquids but in fresh water it travels at roughly five times the speed of sound in air. The damping controls the amplitude response. The damping qualities of a structure control the resonant response. 2. 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. For example a spring is a good example of a low damped system. The higher the damping effect the lower the amplitude of vibration will be as a result of a resonant condition. There are two different types of vibration in liquids: 1. consider a pipe filled with liquid. A shock absorber is a good example of a highly damped system where a large input results in a small response. a relatively small amplitude of input results in a large amplitude response. damping and stiffness directly affect the amplitudes of oscillation in. You can easily see the three components of mass. Sonic vibration and Pulse vibration Sonic vibration To think of a hydraulic resonance. Speed of sound in various liquids (m/s) Table 1 Speed of sound in liquids Liquid water methyl alcohol Benzene castor oil Glycerin Velocity m/s 1497 1123 1326 1500 1923 118 . acceleration velocity and displacement respectively.= 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". Liquids & pumping systems Pumping systems in general and hydraulic systems in particular are often associated with hydraulic resonances. 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.

01T)(S-35) + 0. V(solid) = √(Young's modulus/ mass density). and the properties of the substance through which the sound travels is given by: V = √(elastic property/inertial property). For sound waves in water.055T^2 + 0.6T . S salinity in parts per thousand z is depth in meters Now. Equation 6 Velocity of sound in materials For sound in air.Vibration Analysis on Rotating Equipment Mercury air (reference) liquid helium 1440 320 240 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. V. 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. 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 decreases as the mass of the molecules increases because the mass of the molecules is related to the force per acceleration of the molecule.0. V(water) = √(bulk modulus/ mass density) The speed of sound in water is approximately 1500 m/s.016z Equation 7 Speed of sound in the ocean T temp in degrees Celsius. if we know how fast the impulses travel through the liquid. The exact relation between the velocity of sound. 119 . The speed of sound in an ocean is approximately: c = 1449. However.00029T^3 + (1. the speed of sound increases as the strength of the interaction between the molecules increases. V(air) = √ (bulk modulus/mass density) For longitudinal waves in a solid.2 + 4.34-0.

and so on). the player isolates and makes predominant one of the higher harmonics. the vibrating segments become smaller. In “over-blowing” a wind instrument. others. quarters. or timbre. which is usually perceived as the basic pitch of the musical sound. let us look at the historical use of air resonance – music. Air & gases Before we get too deep into the vibratory aspects of air-borne sound for vibration analysis. particularly if they are both at the same frequencies. The subsidiary vibrations produce faint overtones (second and higher harmonics or partials). 120 . such as the bugle. 4:1. it pushes liquid in front of it because most liquids are virtually incompressible. black notes show pitches that do not correspond exactly with the Western tuning system. The usual outcome is a very noisy operating environment. thirds. As in liquids (described above). a stretched string or an enclosed air column. The impulsive frequency can also excite any mechanical resonances in the system from pipework or supporting structures. Harmonics contribute to the ear's perception of the quality.Vibration Analysis on Rotating Equipment Impulsive vibration The other type of vibration in liquids is from the pulses generated by the system itself. They result when the vibrating body. As the series progresses. Figure 102 Harmonic series for the tone C. and so on). the valves of the trumpet and the slide of a trombone add extra tubing. The harmonics of string players are flutelike tones produced when they cause the string to vibrate solely in halves (or thirds. on a clarinet. of a sound: On a flute. wave-motion vibration. producing wave frequencies that are in simple ratios with the fundamental frequency (2:1. for example. In musical sound the full-length vibration produces the fundamental tone (or first harmonic or first partial). The shock wave from this pulse then travels right through the system giving system-wide excitation at that impulsive frequency. 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. for example: • • • • Vane pass Piston pulse Hydraulic valve operation Impeller eccentricity As each hydraulic pulse enters the piping system. In musical instruments harmonics are a series of subsidiary vibrations that accompany a primary. the frequencies higher. vibrates simultaneously as a whole and in equal parts (halves. we must first know the speed of sound in air. and the musical pitches closer together. It has been the authors experience that heavy vibrations on pumping systems are usually caused by an interreaction of these impulsive and sonic vibrations. produce only the tones of the harmonic series. but mechanical failure often results. or fundamental. The harmonic series for the tone C is given in the accompanying notation. thus extending the range of the instrument upward. 3:1. and so on). Unvalved brass instruments. to calculate the resonant frequency in a pipe. certain harmonics of the series are most prominent. for example. the instrument can thus produce more tones. giving the player a new fundamental wavelength with a new harmonic series.

we can calculate the natural frequency of the air inside a pipe by using the formula: c = f ⋅λ Eq. The lowest frequency where a condition of resonance exists in the tube occurs when the wavelength is four times the length of the tube. and molecular structure. If the returning wave is exactly in phase. 1.8 If a tuning fork is set in vibration and held over the open end of a tube closed at one end. molecular weight. γ is the specific heat ratio. 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 331.30 for triatomic gases. of the length of the tube.2 to 1. 1. whereas the pressure difference will be greatest at the closed end. T is the temperature in Kelvin. sound waves which travel down the tube. will be reflected back from the closed end. and closed at one end are shown below: 121 . and M is the molecular weight of the gas.Vibration Analysis on Rotating Equipment The Newton-Laplace equation for the speed of sound in a gas is v = √( (γ P) / ρ) γ Equation 9 Newton-Laplace eq. and in the range of Applying the Ideal Gas Law.. 4/5(fifth harmonic).6 m/s. γ = cp over cv Equation 10 Ratio of specific heats (gamma) ≈ 1. 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. Higher modes of resonance exist when the wavelength is 4/3(third harmonic). a condition of constructive interference exists which we call resonance and the loudness of the sound will be greatly increased.40 for diatomic gases. m is the mass of one molecule. The speed is nearly independent of atmospheric pressure but the resultant sound velocity may be substantially altered by wind velocity.. k is Boltzman's constant. The displacement of the air will be greatest at the open end of the tube. This is called the fundamental mode of vibration. R is the ideal gas constant. The speed of sound in a gas depends on the temperature. γ = ratio of the specific heat of the gas at a constant pressure over the specific heat at a constant volume.1 for polyatomic gases. 1..67 for monatomic gases. for the speed of sound in a gas where and P =pressure ρ = density. Just as in the liquid pipe. and vice versa. The first four resonance modes for tubes open at both ends. 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).

5v/4f and so on for a tube closed at one end.3L1) or 1/2(3L3 . where e is the end correction. both the velocity of sound may be determined by using the relation c = f⋅λ. 122 . It is actually a small distance beyond it. When resonance exists. three positions of resonance can usually be found. the resonance lengths are v/4f. but the antinode is not exactly at the open end. If the positions of resonance from the open end of the tube are L1.L2) or (L3 . 3v/2f for a tube open at both ends. 3v/4f. and the end correction may be calculated from 1/2(L2 . The acoustic length of the tube is equal to its physical length plus the end correction. 2v/2f.5L2) or 1/4(L3 . Karman Vortices Everyone knows the story of the Tacoma Narrows suspension bridge.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). the displacement is a minimum (node) at the closed end. L2.L1) or 2(L3 .5L1).L1). The mechanism for the resonance-destruction of the bridge was the same mechanism that causes a flag to flutter – Karman Vortices. and L3. Since the frequency of the tuning fork is known. From these relations the wavelength may be calculated from 2(L2 . and tuning forks of frequency about 500 Hz. the wavelength of the sound wave is equal to 4(L1 + e) or 4/3(L2 + e) or 4/5(L3 + e). The bridge began to sway violently when the wind blew steadily at a certain velocity. In a typical laboratory experiment with a water reservoir-resonance tube apparatus about 1m long. If the frequency is fixed. This extra distance beyond the end of the tube is called the end correction . as in the case of the tuning fork held over the open end of a tube. and v/2f. and the value of the end ⋅ correction may be determined from the relations above.

any shaft rotational speed which is associated with high vibration amplitudes. 123 . As the wind hit the side structure of the bridge it developed Karman vortices. the speed which corresponds to a rotor lateral mode resonance frequency excited by rotor unbalance. f * D = 0. in which case it is more correctly called the balance resonance speed. 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. There is a distinct relationship between the frequency (f) of the force.22 V Equation 12 Karman vortices Figure 105 Tacoma Narrows Bridge In the case of the Tacoma Narrows bridge.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. Critical Speed (Balance Resonance) Critical speed is. 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. a steady wind blew along the valley onto the side of the bridge. 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. Often. in general. perpendicular to the air flow. the velocity of the wind (V) and the width or diameter (D) of the object.

A diagram used in rotating machinery design. 124 .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. The resonance speed is at the point which is 90° phase away from the angle of the heavy spot for that resonance mode.48X). When the speed increases or decreases. oil whirl (. i. damping. These changes may not happen at the same frequency due to non-linearity. 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. etc. blade or vane passing frequencies. rotational speed (1X). This may be slightly different than the rotational speed peak amplitude point. gear mesh frequencies..40X to . 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.e. and/or asymmetry in the system stiffness. The X axis represents the various possible excitation frequencies. Campbell Diagram.. The term is sometimes used incorrectly to describe the cascade plot and waterfall plot. The Y axis represents the rotor lateral and torsional natural frequencies.

Vibration Analysis on Rotating Equipment Figure 107 Campbell Diagram 125 .

radial lines). Eventually it will break. Remember to 126 . The break has the characteristics of a pure fatigue failure which is often mistakenly called “crystallization. If non-resonant. characteristic of a resonant condition. gear cracking or failure (usually in equal-spaced. Identifying resonance in mechanical systems Mode Shape Figure 108 Mode shape node points Whenever cracking of the machine or structure is reported.” 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. But sometimes the pipe. Cracking almost always occurs at one of the mode shape nodes. The phenomenon is similar to bending a wire back and forth until it breaks. The same phenomenon is very common for pressure gauges cantilevered on a pipe. but when it does. The material does not break due to excessive stresses but due to the many millions of reversed stresses concentrated about the nodes. by feeling with fingers along the small diameter. with a gauge acting as a weight. such as at the bearings. is resonant to a specific frequency of machine vibration. However. even though the calculated stresses indicate the shaft should last a lifetime without breaking. This simple procedure can be performed with the most rudimentary of instruments and is very effective in identifying the mode shapes. lubrication. causing fatigue. usually at its point of connection. there is no problem. gear driven shafts broken with the characteristic 45° torsion break. it can be determined if there are tell tale large “loops” (antinodes) and nodes. 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.Vibration Analysis on Rotating Equipment . Premature foundation or floor cracking. Sometimes rotating machinery does not show excessive vibration at the usual points of measurement. have all occurred as the result of metal or concrete fatigue at one of the nodes due to resonance. pipe weld breakage (often repetitively). cooling water or instrumentation piping. resonance should be suspected. 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. lost lubricant can wreck an expensive machine or a flammable product can be released. It may take many months or even years for the tube or pipe to crack at one of the nodes.

On piping sections the direction indicating the greatest “curl” is the resonant direction. 3. Displacement units are preferred but velocity or acceleration can also be used. On a horizontal zero line drawn on a sheet of graph paper mark the measurement locations spaced 1” or 25mm apart. then the force and resulting deflection will remain in phase. Then the process starts all over again with another phase angle reversal. 4. usually with a letter to avoid confusion. 5. Divide each reading by 2 (for displacement). the deflection will time-lag the force. 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. and so on. the force and its resulting deflection will move in the same direction at the same time. when the vibration frequency enters the resonance frequency or the critical speed range. 6.” and if the frequency of that vibrating force is below the critical speed range or resonance range. Measure the vibration amplitude at the frequency of interest at each location. When the vibration frequency reaches the actual critical speed or resonance frequency. Figure 109 Mode shape readings Phase If a force is slowly applied to a spring system. If a vibrating force is applied to the same object or “spring system. However. until the second resonance frequency range is approached. Divide the component to be tested into a number of roughly equal parts and identify each location. Join the dots “curve fit” if required. The 180° relationship will remain the same for all frequencies above the resonant frequency range. 1 cm = 25 microns. This phenomenon is shown in the 127 .Vibration Analysis on Rotating Equipment plot both directions (horizontal and vertical). As the frequency increases further. the force will precede the deflection by 90° . or “in phase” with each other. 2. Overall readings can be used but filtered data is preferred. The procedure for performing the mode shape plot is as follows: 1. the force increases its lead until the force finally precedes the resultant deflection by 180° .

Figure 110 Phase relationships This phenomenon can be used to help us determine resonance. such as pipes.Vibration Analysis on Rotating Equipment resonance diagram of a rotor’s vibration amplitude and phase relative to resonance of some part. part of a steel base or beam. Remember that almost all troublesome resonances occur in nonrotating parts. beams. such as the rotor support system. or more rarely. Parts may be temporarily braced to add rigidity and. bases. 128 . These parts can usually have their resonant frequencies changed while the machine is still running. resonance of the rotor itself. decks and so on. therefore. yet the phase change method to determine whether a resonant condition exists can still be used. moved to a higher resonance frequency that is at least 20 to 25 percent away from the source frequency. Typically the running machine cannot be shut down or have its speed varied for resonance tests. pedestals.

however. the natural frequency generated in this way will be indicated on the analyzer. Although care should be taken that the frequency at which the "bumps” are struck is not read as a resonant frequency. Since an object will undergo free vibration at its natural frequency when bumped or struck. are almost always limited to determining the first critical frequency of the part bumped. 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. Also. Bump tests. simply bump the machine or structure with a force sufficient to cause it to vibrate.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. if the part tested is connected to another part that is relatively flexible. 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. The calculations are fairly straightforward for simple structures. 129 . With the machine shut down and a vibration pickup held or attached to the machine. One method of determining the natural frequency of a structure is to calculate it. 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. 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.

such as main steel support structures of buildings. covers. tension on the rope deflects the structure by several mils. It takes more time and patience than regular impact bump tests. However. This test is based on the principle that when a “spring system” or part is bumped or deflected in some way. do not as easily respond to the regular bump test. If the bumps generate natural frequency peaks that correspond to a peak produced by the running machine (before bumping). it will vibrate for several cycles at its natural frequency. A structure such as a platform supported by columns will have individual resonant frequencies for each column and the platform. columns. Set up for FFT-type analyzers FFT-type analyzers have various settings which determine the type of spectra obtained. If peakhold averaging with running display capability is available. floors. A running machine spectrum is 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. but its frequency remains the same. pedestals and welded steel bases. A strong rope is tied to the clamp. The low frequency of the repeated bumps is not measured.. For example. The structure will spring back. then the amplitude on the analyzer screen should increase. but will not necessarily work well for the total structure that includes all the vertical columns. Usually a clamp is attached to the top of the structure.Vibration Analysis on Rotating Equipment Bump tests are also ideal for rotors. The peak’s width will also increase if the peak is affected by resonance. The two spectra can then be compared to determine if there is a resonance. Position the pickup and set the vibration instrument for the expected frequency range. beams and platform. pulley. 2) Set the Fmax (maximum frequency) to a suitable value higher than the suspected frequency. For a very large structure. such as one supporting a fan or the total frame of a vertical pump. Now cut the rope. The area where the rope could whip must be cordoned off and all bystanders moved out of harms way. including cast iron or cast steel bearing supports and brackets. a “come-along” or even a lift truck or tractor. the structure can be deflected as much as a mm or two without harming the structure. However. A regular impact bump test can be used for the horizontal beam. To determine the resonant frequency of the total structure. take data with the machine running. Running machine Bump Test Normally. bump the structure repeatedly. They are also practical for relatively rigid parts. Its amplitude decreases with each cycle. Another is taken while bumping the part. rolls. Very large columns and beams. Tension is applied to the rope using any practical means such as pulling with a turnbuckle. 130 . beams. bump tests are performed with the machine shut down. The force will deflect the beam but will also cause the whole structure (including columns) to deflect as well. Ensure that the transducer weight is less than 10 percent of the system weight. The rope will need to be retied. only the frequency that is generated between bumps. If the structure is small. a “reverse” bump or pluck test works very well. a horizontal force can be applied in one direction at the top of the structure. etc. such as a tower of several stories. cut. in certain cases a bump test can be performed while the machine is running. Then with the machine still running. also has several flexibilities depending on the direction being considered. tension applied. etc. With the instrument set up accordingly. fan blades. 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. giving its resonant frequency in a manner similar to regular bump tests. Assume that the force is applied to the horizontal beam supporting the platform. the total structure as a combination of columns and platform. Repeat the procedure several times to make sure repeat readings are obtained. Reverse Bump or “Plucking” the Suspect Part. then its use is preferred.

For expected resonant frequencies in the range of normal speed machines.Vibration Analysis on Rotating Equipment 3) Set the instrument up for manual amplitude scaling (disable the autorange feature) and select an appropriate amplitude scale. 6) Take a spectrum and observe the background vibration data (if any). When the hammer impacts one part. However. such as 4 to prevent excessive data collection time. For higher expected frequencies. the results are usually acceptable for most situations. Special windows and trigger settings are available on most data collectors today. 5) If the analyzer has the capability to display continuously updated (running) spectra using “peak-holdtype” averaging. using an FFT. A typical spectrum obtained from a bump test will look similar to regular spectra obtained from vibrating running machines. 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. 7) The structure should now be impacted in the direction of transducer orientation using a soft hammer or piece of wood. If you have an analyzer which can perform negative averaging. although not necessarily in all instances. then the number of averages should be minimized. 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. The more the lines of resolution. The largest peak usually represents the resonant frequency of the part that receives the bump test hammer’s impact. These peaks. then that part may cause the largest amplitude peak. the impact also causes some deflection and release on parts that are connected to the part receiving the hammer blow. 4) Set the resolution to an appropriate number such as 400. Typically. then use this capability to remove the background vibration after the bump test. If “peak-hold” is not available. If multiple blows are required then random impacts are preferred. Impact hammer A better way of determining natural frequencies is to use the impact hammer. the longer it takes for the instrument to acquire the data. 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. bumping with a rubber mallet is preferred over harder materials such as steel or plastic. If this is used. If the other part is more flexible or has much less damping. there is one peak that has a much higher amplitude than the rest. They can be used but are not necessary for this simplified test. A single bump test. 131 . except that the various peaks will not necessarily be related to running speed. Despite the loss of accuracy due to the lower resolution. the peaks will relate to the resonant frequencies of various machine parts that were displaced by the bumping action. Most modern data collectors have the capability to generate a “live” time display. Some condition monitoring software programs will subtract spectra as well. If autorange remains active the instrument will tend to range to the initial bump rather than the “ring down” afterwards. representing vibration coming from other machines will have to be mentally subtracted from the bump test results. The amplitude scale and strength of impact should be set to respond to the “ring down” rather than the impact. harder hammer heads such as plastic are usually recommended. An accelerometer is mounted on the structure and the resultant output is generated by the natural frequencies of the structure. may not always determine with certainty which peak represents a specific part. then its use is recommended for the best results. 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. Soft wood (such as a “2x4”) can also be used. Instead. Start with a sensitive scale and adjust the scale as required to suit the strength of the impact and the size of the structure. then a setting of 100 lines may be more appropriate. it causes a deflection and release. such as under 5000 rpm.

132 . model 086C41.Vibration Analysis on Rotating Equipment Figure 112 Impact hammer response The response can be analyzed with spectrum. For this course we will use medium sledge from IMI. 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.

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. above: Note that if the individual oscillations are smeared into one pulse in the time domain. Consider table 2. then the FFT cannot have relevant frequency domain data. 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”. when 133 . In some cases it is possible to get a quasi-triangular pulse which.

in fact. will give a frequency response which looks good but is. 134 .Vibration Analysis on Rotating Equipment passed through the Fourier transform. totally meaningless.

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. Table 3 Natural frequency calculation of uniform beams 135 . Uniform Beams One of the most common sources of resonant vibration is the structural supporting beam. This section gives guidelines for the calculation of natural frequencies for simple structural elements.

Vibration Analysis on Rotating Equipment The notation a1 through a5 is for each vibration mode. The weight per unit length is a part of the beam identifier. calculating a1 for a clamped-clamped beam will return the natural frequency of the first vibration mode. • • For low carbon steels E (Young’s Modulus) can be estimated at 30. calculating a2 will return the frequency for the second mode etc.natural frequency of beams The difficulty in making this type of calculation is getting the values for E.000 psi. E. 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.000. standard text books such as the “Machinery’s Handbook” have all of the required information for different types of beams. 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 . ωn a 1 ..l Equation 13 General formula . For example. Luckily. I 4 µ . I and µ. This is µ in our calculation above. however.

ωn l D µ1 2 Where And µ1 = mass per unit length D = the plate constant The plate constant (in this case) is defined by 7 Available in Borders bookstore.0 (from table 2) = 12 feet = 144 inches ωn = 22 * (30. Let us calculate the natural frequency for a low carbon steel. 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. / in = 22. The way that the example above is drawn. Whenever you are given properties in other units. designated W18x106. length of side t.000 = 220 in4 = 106 lb./ft = 8.000.000. “clamped-clamped” beam. all edges clamped.l 4 E µ a1 l = 30.616 Hz I (section Y-Y) (calculator notation) Notice that we have been consistent with our units – we have only used pounds and inches in the equation.833 lb. 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. 12 feet long.5 = 29. you must convert to the units for which your formula is designed.95.000 * 220/(22 * 1444))0. the fundamental mode is 36. I µ .Vibration Analysis on Rotating Equipment Notice that there are two values if I. (DOVER 0-486-64785-4) 137 . ω n a 1 .00117 rad/s = 4. the section X-X will give us the moment of inertia horizontally and the section YY will give the moment of inertia vertically. $12. E. For a square plate.

of side 6 feet we would have ωn = 0.221 Hz If your plate is clamped at each end but free on the other two sides.3) For a low carbon. 138 . t 12 1 3 D µ 2 Where t = thickness (inches) µ = Poisson’s ratio (≈ 0.611 Hz If the same 3 foot plate were ½” thick: ωn =1. then treat the plate as a clamped/clamped beam.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. steel plate. clamped at the edges. the plate constant D will be 4.293 x 104 So that for a square plate.Vibration Analysis on Rotating Equipment E. ¼” thick.

6 to 30 Hz 1176 to 1800 cpm 0. which has more or less the same kind of properties. the locked-in air is eventually pressed out. but their disadvantage is that they provide very little damping.058 mils p-p 0. excessive movement occurs when running the machine through the resonance range.5 mils p-p 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. The benefit of these is that they permit relatively high deflections.045 mils p-p 0.66 mils p-p Lowest Amplitude Expected 3. The types of spring material which are used most often are rubber and steel. However.0 to 10.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. The same applies to felt. Because of this. which is why it is no longer used in more demanding applications. 8 Taken from information prepared by Larry Riley (3/6/96) 139 .0 to 12.8 Hz 72 to 228 cpm Rubber 4.6 Hz 480 to 756 cpm Felt 19. Vibration isolators This section explores some of the isolation materials that we can use. 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.2 to 3.5 Hz 240 to 630 cpm Cork 8. The following 8table suggests some appropriate damping ranges for popular vibration isolation materials: Table 5 Damping ranges of vibration isolators Material Steel Springs Effective Damping Range 1. Another alternative is air springs. thus impairing the isolation result and causing it to have no effect at all after a period of use. Steel springs are normally used in the form of coil springs or leaf springs. Vibration isolation is based on installing machinery on springs or resilient material of uniform stiffness. Cork does not permit a large spring travel. and more often than not special devices have to be installed in order to limit the deflections. 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.

The stress-strain diagram illustrates clearly the difference in the strain properties of rubber and steel. The maximum tensile force is reached at the ultimate strength. There is no yield point. but the stress increases continuously until there is abrupt failure. To ensure effective isolation. and the proportionality constant is equal to the modulus of elasticity E. Figure 115 Stress/strain diagram for rubber The tensile sequence for rubber is quite different. Consequently. rubber has high internal damping. At the tensile yield limit. has good elasticity and chemical resistance. This states that stress is proportional to strain. or yield point. In such cases.Vibration Analysis on Rotating Equipment Figure 114 Stress/strain diagram for steel Compared with steel. and these are properties which make rubber particularly suitable as a spring material. Steel is elastic up to the elastic limit and restores to its original shape (this value is less than 0. rubber does not "stretch" proportionally to the load. The principle relating to vibration-isolating with springs is that they are placed between the machine and the base or plinth. and figure 2 illustrates the corresponding test on rubber. and Hooke's law does not apply in this case apart from approximately in a small strain range of the magnitude of 5-10 %.1 %). the springs must be calculated very accurately. the modulus of elasticity is not constant. the force can be reduced to only 2 or 3% of the force of a rigidly mounted machine. the material is extended without the tensile force increasing. Rubber achieves high strain levels even at very low tensile stress. The increase in load is neither linear. Figure 1 illustrates in diagram form a tensile test on steel. Springs Springs have long been a favorite with fan manufacturers for separating fan vibration from the floor. and in this range Hooke's law applies. In favorable cases. otherwise the result could be impaired performance. The stress strain curve for steel is linear up to the proportionality limit. it also absorbs noise. you could say that the vibrations are practically 140 . which is just above the proportionality limit. in other words.

Here. 3) The force transmission percentage is reduced substantially by correct calculation and suitable mountings being installed between the machine and base. All machines have more than one resonance point as. that is. the magnification factor can increase considerably and may amount to several hundred per cent. the force transmission percentage will be small. If Z=1. The resonance points can be determined. then. The magnification factor can be regarded as being 100%. Figure 116 Resonance Curve As can be seen in figure 10. the machine speed (rpm) = the natural vibrations of the system. 2) In the case of an unsuitable spring system. by means of springs. which has minor internal damping and in which the amplitude. when Z=1. depending on the hardness of the rubber. that is. As a rule you usually know the operating speed of a machine (interference frequency). B drops rapidly. Z should be between 3 and 5. The term “4D2 Z2” can generally be neglected completely except in the resonance range. then it is possible to change the force transmitted. The low elasticity and shear moduli of rubber are used to achieve a low natural frequency. it is said that there is resonance. The magnification factor B. The relative magnitude of the transmission of force depends entirely on the tuning ratio Z. For practical applications. which is therefore forced to be a part of the movement of the machine.Vibration Analysis on Rotating Equipment eliminated. Vibration isolation is therefore of significance first when the operating frequency considerably exceeds the natural frequency. transmission of vibration forces can be affected in three ways: 1) Rigidly mounted machines transmit vibration forces in unchanged form to the base.e. This is exactly what happens when vibration isolation is achieved i. 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.1. through many interacting movements. that is. grows to a very high value in the resonance point. and thereby influence Z. in theory. but in favorable circumstances can be as low as 2 %.04 . which means that 88 and 96 % of interference forces are eliminated. B (at Z=D√2) has dropped to 100% and when Z is further increased. the change in force acting on the base of a machine supported on springs compared with a machine which is rigidly mounted. To sum up. In rubber D has the value 0. they can vibrate in many different ways. Typical reductions can be from 100 down to 10 %. and the vibrations will be infinitely large if there is no damping. a rubber spring has a distinct advantage over a steel spring. If the system's natural vibration coefficient can be altered in any way. but the methods of calculation are often 141 . If Z is high.0.

000 * 0. the tensile strength of rubber is rather low. This does not apply to rubber under tension or compression.Vibration Analysis on Rotating Equipment very difficult.200.429) where G = torsional modulus (typically 11. according to Hooke's law./inch ωn = (317. is supported by by four springs. on the other hand.192” Number of turns (spring) = 6 so: k = (11. This means that rubber does not have any constant tensile or compression module of elasticity. rubber is very ductile. To calculate the natural frequency of a spring: Equation 15 Natural frequency of a spring Where k = the spring constant k (linear spring) = Gd4 8nD2 (refer to Hartog p. Metals. The maximum level that can be achieved with rubber is 25-30 MPa. Calculate the natural frequencies of the unit using the following assumptions: Weight of motor and fan = 500 lbs. however. In some cases.089 lbs. If a material is subjected to a load below the elastic limit. have very small strains below the elastic limit. Metals will normally be softer towards the end of a tensile test. and the modulus is increased until there is abrupt failure. Experience has shown that all resonance frequencies that can arise do not need to be clarified.200. To make the calculation easy we will assume that the weight of the unit is equally distributed over the four springs. However. the deformation will. It is usually quite sufficient to calculate the most significant ones which can be determined easily.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. rubber has a very large work absorption capacity compared with the best grade of steel. and by far the highest proportion of this strain is elastic. The desired level of isolation and the interference frequency determine where the resonance frequency shall be. in turn. Compared with metals. 142 . the strain may be higher than 1000 %. Rubber does not have any yield point. Spring coil diameter = 1” Wire diameter = 0.5/2π = 0.1924)/(8*6*12) = 317. be proportional to the load.2535 Hz Rubber Compared with other engineering materials.089*4/500)0. because of the high strain. while the opposite is often the case with rubber.

**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 hardness and shear modulus

rubber

Figure 20 illustrates the relationship between rubber hardness and shear modulus. Figure 21 shows the dependence of the bulk modulus on the shape factor. The latter curve applies at 10% deformation.

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 Where Ra = k (Forcing_Function/Damping) 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

you could end up with a totally new set of resonance problems. Examining the formula it is obvious that increasing the stiffness will increase the natural frequency of the beam. Additional stiffness can often be added by using turn-buckles at a very small mass penalty. In the first place.l 4 Remember that µ is the mass per unit length and l is the length. If stiffness is added to the machine.Vibration Analysis on Rotating Equipment amplitude response. E. If you think that adding stiffness is the correct solution for the resonance problem. E. I µ . 146 .l 4 The combination of E and I may be considered to be the stiffness of the beam. increasing the stiffness by 10% will increase the natural frequency by only 3. only stiffen in the direction which has a problem with the natural frequency being co-incident with the exciting force. 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. 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. great care should be exercised. In small systems it is sometimes feasible to add a damper such as a shock absorber. But additional mass reduces the natural frequency. But the ratio of increase is not linear. • • If you stiffen in places where it is not required. By increasing the value for I we can increase the natural frequency of the system. Changing the Mass Remember the formula for natural frequency? ωn a 1 . Take care when adding the finished mass (often in the form of steel) that you do not increase the stiffness of the structure. you will add mass. Changing the stiffness ωn a 1 . I µ . In other words you have done a lot of work for no change in the vibration characteristics of the machine. but usually we cannot change the damping of the system significantly. In other words.16% (√10). be very careful about the mass that is added. As you stiffen the structure. 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).

impact tools such as chippers (425 m/sec) and riveters. 1993) suggest that hand vibration in excess of 12 mm/sec should have exposures of less than 1 hour. This can contribute to fatigue and a reduced ability to perform certain tasks. pneumatic and electric rotary tools such as grinders [380 m/sec]). higher frequency vibration (15 to 40 Hz) can interfere with vision.. rail vehicles. sea vessels (hovercraft. such as power hand tools. •Holding vibrating controls (e. The initial effects of whole body vibration are discomfort and irritation. heavy machinery. In many cases. Exposure to whole-body vibration can also occur while standing or sitting in vibrating environments or objects. posture) and non-occupational factors (age. forestry or mining equipment). and buildings (generators. work environment. ships). neural disorders.g. Sources of whole body vibration include virtually all modes of transportation (the jarring and jolting associated with off-road vehicles. Prolonged or extreme exposure to whole body vibration has been associated with gastrointestinal disorders. Guidelines established by the American Conference of Government Industrial Hygienist (ACGIH. dozers and construction. fitness). lawn mowers).. pumps. frequencies between four and eight Hz are likely to affect the gastrointestinal and spinal systems. Prolonged exposure to whole-body vibration has been associated with back and neck musculoskeletal disorders. Some examples of whole body vibration sources include the following: • • • • • on-road and off-road vehicles (buses. Whole body vibration usually originates from a moving vehicle or from the operation of heavy machinery. vibration-related syndromes are aggravated by other occupational factors (work history. tractors. and gasoline powered tools such as chain saws (70 m/sec). trucks.g.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. fixed-wing aircraft). Much can be gained by routine maintenance of the 147 . aircraft (helicopters. Very low frequency vibration (less than one Hz or cycle per second) can result in motion sickness. for example). •Holding vibrating work pieces such as when grinding or polishing. low back pain and spinal degeneration. ACGIH and other agencies that recommend vibration guidelines update their guidelines when new information becomes available. health. (The tool manufacturer should be contacted to obtain tool operation characteristics. Exposure can occur while holding: • • • vibrating tools (e. ventilation systems). such as trucks or heavy machinery. Frequency ranges The specific effects also depend on the vibration frequencies to which the worker is exposed. Even a building's vibration may reach levels of concern (such as in process control rooms or at machine operators' control platforms). skill level.) The ACGIH guidelines are based on symptoms of white finger or Raynaud's phenomenon. Localized vibration from hand-held power tools can increase grip force and symptoms of hand-arm vibration syndrome (HAV) and carpal tunnel syndrome.

Anti-vibration suspension seats include a damper and a spring in the seat system. Table 6 Whole body vibration (frequency ranges) Frequency Range < 1Hz 4 – 8 Hz 15 – 40 Hz Physiological Result Motion sickness Gastrointestinal & spinal problems Vision difficulties Whole body vibration transmitted to vehicle operators may be reduced by proper maintenance of roadways.Vibration Analysis on Rotating Equipment vibration source and existing isolation systems. isolate the system from the vibration source (mount the system on anti-vibration mounts).1. Control room vibration may result from machinery operation or heavy vehicles traveling in the vicinity. There are large differences between seats and the performance of an individual seat may change during its lifetime. Generally. this may be accomplished by • 2. or • 2. 2. reduce vibration at the source (in the case of heavy machinery this may be accomplished by balancing moving parts) modify a vibrating system to reduce the likelihood of excessive vibration due to resonance (in the case of a cover panel. Vehicle cab isolation is available in some vehicles and can effectively reduce vibration exposure.2. For more persistent problems. as well as seating and engine mounts should be routinely checked and serviced when required. track and suspension systems.1. stiffening the system by welding a steel section to it).1. Tire pressure. vibration measurements will likely be necessary to make an educated selection of appropriate vibration control strategies. If vibration control available through vehicle and roadway maintenance is insufficient. Conventional foam seats tend to amplify vibration frequencies in the range of greatest susceptibility for the human body. The spring may be constructed of steel or it may be a column of air.1. increase the damping in the system (damping materials may be coated on the system). improved seating selection may provide further protection for the worker. vehicle maintenance and control of vehicle speed. 2. 148 . It is important to know the vibration levels and frequencies in the vehicle in order to select an appropriate suspension seat. This vibration may be addressed in a number of ways: 1. two types of seating systems are considered when addressing a vibration problem: conventional foam seats or suspension seats.

a (Newton’s 2nd law).Vibration Analysis on Rotating Equipment Glossary A Accelerometer Transducer for measuring vibration in the form of acceleration. Attenuation The weakening of a signal by distance from the source or a mechanical interface. Anti Aliasing Nyquist's theorem says that as long as the sampling rate is greater than twice the highest frequency component of the signal. ASCII American Standards Code for Information Interchange. AC Literally Alternating Current but often used to imply a time-fluctuating signal. then the sampled data will accurately represent the input signal. averaging many spectra together improves the accuracy and repeatability of measurements. Bin See “resolution” Blackman-Harris Window 149 . Acceleration The rate of change of velocity in inches/second/second or G’s (acceleration due to gravity is 1 G). When an amplitude reaches or exceeds the alarm amplitude then the software automatically generates an exception report. Averaging In general. This is a good indicator of the forces inside a machine since F = m. 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. It is one of the most rugged transducers for vibration measurements and also has the widest frequency range. B Band Pass Filter A measurement filter that removes data below the low cutoff frequency and above the high cutoff frequency. This is the antialiasing filter. 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. The band pass filter only passes the data between the cutoff frequencies.

all the power of the output signal is due to the input signal.the most common format for displaying frequency in vibration analysis.7 dB) than the Hanning. The phase is the relative phase between the two channels. the input and output are completely random with respect to one another. 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. 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. Coherence Coherence measures the percentage of power in channel 2 which is caused by (phase coherent with) power in the input channel. Frequencies where signals are present in both spectra will have large components in the cross spectrum. 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. This allows signals close together in frequency to be distinguished. Correlation The two channel analyzer may also compute auto and cross correlation. 150 . 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. The filter is steep and narrow and reaches a lower attenuation than the other windows. If the coherence is 1. 100 mV/G for an accelerometer. CPM Abbreviation for Cycles per Minute . If the coherence is 0. e. Although correlation is a time domain measurement the some analyzers use frequency domain techniques to compute it in order to make the calculation faster. and the cross-correlation is a measure of how much overlap a signal has with a delayed-by-t version of the other channel. very good selectivity and the fastest filter rolloff. The magnitude is simply the product of the magnitudes of the two spectra. It has better amplitude accuracy (about 0.Vibration Analysis on Rotating Equipment The Blackman-Harris window is a very good window to use with the spectrum analyzer. 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. Coherence is a unitless quantity which varies from 0 to l.g.

New Average = (New Spectrum .g. Data Collector Hardware device for collecting vibration data off line.806 m/S2 E Envelope Measurements A. e. Decibels are usually referenced to a standard. 151 . remember 1G = 386 in/s2 or 9.f) = Displacement /(2. A database allows rapid access of the database files. In PdM terms this may be used to check the accelerometer or for a process measurement. Averaging takes place according to the formula.π.1 GdB re 0. dB = 10 log (X/Xο) where X or Xο is a power or squared quantity or where Y or Yο is a linear quantity.π. 0.b.001G. Differentiate Differentiation changes displacement to velocity and velocity to acceleration relative to time using the formulae: Acceleration Velocity or Acceleration = Velocity /(2. Decibel (dB) A convenient unit for displaying logarithmic data. Data is outside the defined “safe” area. demodulation (see appendix) Exception To have an alarm condition. Exponential Averaging Exponential averaging weights new data more than old data. DC Literally Direct Current.I/N) +(Old Average) .π.f) = Displacement /(2.f)2 dB = 20 log (Y/Yο) where f = frequency n.Vibration Analysis on Rotating Equipment D Database A collection of information files that are ties together by a common topic.k.(N-l)/N where N is the number of averages.a.

His experiments with these electromagnetic waves led to the development of the wireless telegraph and the radio. The FFT spectrum analyzer samples the input signal. H Hanning Window The Hanning window is the most commonly used window. which travel at the speed of light and which possess many other properties of light. such as operating speed. signals appear wide but do not leak across the whole spectrum. Make sure that the number of averages is not so large as to eliminate the changes in the data that might be important. Thus. like you would see on an oscilloscope trace. Heinrich Rudolf Hertz (1857-94). High Pass Filter 152 . Its between-bin amplitude variation is about 0. further changes in the spectra are detected only if they last sufficiently long. Frequency The rate at which periodic events happen.5 dB (for signals between bins) and provides reasonable selectivity. and displays the spectrum of these measured frequency components. Hertz (Hz) Common frequency units in cycles per second. F Flattop Window The Flattop window improves on the amplitude accuracy of the Hanning window. 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. Hertz clarified and expanded the electromagnetic theory of light that had been put forth by the British physicist James Clerk Maxwell in 1884. Unlike the Hanning. and compute its frequency spectrum. Its filter rolloff is not particularly steep. it is commonly abbreviated Hz. Typical units are Hertz (Hz). As a result. It has an amplitude variation of about 1. German physicist born in Hamburg and educated at the University of Berlin. Once in steady state. Harmonic Frequencies at direct multiples of a fundamental frequency. computes the magnitude of its sine and cosine components. The fundamental is not necessarily the run speed. the Flattop window has a wide pass band and very steep rolloff on either side. Cycles per Minute (cpm) or orders (multiples of run speed). However. the Hanning window can limit the performance of the analyzer when looking at signals close together in frequency and very different in amplitude. 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.Vibration Analysis on Rotating Equipment Exponential averages "grow" for approximately the first 5N spectra until the steady state values are reached. the selectivity is a little worse.02 dB. Named after. Fundamental Frequency Primary frequency. The unit of frequency that is measured in cycles per second was renamed the hertz. Hertz proved that electricity can be transmitted in electromagnetic waves. FFT Spectrum Analyzer FFT Spectrum Analyzers take a time varying input signal. to which other frequencies may be referred back.

it is the best window to use for measurements requiring a large dynamic range. Small signals may be hard to detect with linear scaling but become visible with logarithmic scaling. This type of averaging is useful for eliminating transients. The Kaiser window has the lowest side-lobes and the least broadening for non-bin frequencies. Linear Linear scaling displays all data on an equal basis.Vibration Analysis on Rotating Equipment A measurement filter that removes data below its low cutoff frequency. Because of these properties. which is available on IRD analyzers. Use DC coupling for process measurements and use AC for vibration measurements.It turns out that the log display is both easy to understand and shows features which have very different amplitudes clearly. This avoids the problem of taking the log of negative voltages. DC coupling includes both DC and AC signals whereas AC coupling does not include any DC offsets.π. Line of Resolution see “resolution” Logarithmic Although the linear magnitude scale is used most often for displaying spectra. below full scale.01% of full scale would look like on a linear scale. If we wanted it to be 1 inch high on the graph.8 dB for signals between exact bins). the top of the graph would be 833 feet above the bottom . Imagine what something 0. This effect is reduced by using windows such as Hanning or Flat Top. Linear Averaging Linear averaging combines N (number of averages) spectra with equal weighting in either RMS. The Log Mag display graphs the magnitude of the spectrum on a logarithmic scale using dBEU (Engineering Units) as units. another way of displaying amplitude is the Log Magnitude. Integrate Integration is the opposite to differentiation and changes acceleration to velocity and velocity to displacement where A = V x (2. 153 . Vector or Peak Hold fashion. L Leakage Errors resulting from Fourier transforming nonperiodic time domain data. The real and imaginary parts are always displayed on a linear scale. I Input Couple Measurements may be AC or DC coupled.f) K Kaiser The Kaiser window. combines excellent selectivity and reasonable accuracy (about 0. Low Pass Filter A measurement filter that removes data above the cutoff frequency. The 16 bit analyzer has a dynamic range of about 90 dB.

measurements this value includes vibration at all frequencies. Orbit The orbit is simply a two dimensional display of the time record of channel 1 vs. With narrow spans. When RMS averaging narrow spans. the points at the ends of the time record do not contribute much to the FFT. The first order corresponds to the operating speed. The orbit display is similar to an oscilloscope displaying a "Lissajous" figure. The bands are arranged in octaves with either 1 or 3 bands per octave (1/1 or 1/3 octave analysis). these points are “re-used" and appear as middle points in other time records. the second order is two times the operating speed and so on.11 (1986). 0% is no overlap and 99. this can reduce the measurement time by a factor of two. The amount of overlap is specified as a percentage of the time record. 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.8% is the maximum (511 out of 512 samples re-used). Remember. handles input and output operations and accepts and executes commands issued by the user. Thus. the analyzer does not wait for the next complete time record before computing the next FFT. 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. Octave analysis computes the spectral amplitude in logarithmic frequency bands whose widths are proportional to their center frequencies. or 1/12 octave. 1/3. With overlap processing. This speeds up the processing rate.Vibration Analysis on Rotating Equipment O Octave Analysis The magnitude of the normal spectrum measures the amplitudes within equally divided frequency bins. 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. in octaves. Overall A single value representing the vibration or some other measurement parameter. Typically the shape of each band is a third-order Butterworth filter whose bandwidth is either a full. The update rate would be no faster than one spectra per time record. that is. Orders A frequency axis scale which is useful for viewing data as a function of the operating speed. With overlap. most window functions are zero at the start and end of the time record. the time record of channel 2. Octave analysis measures spectral power closer to the way people perceive sound. It loads programs. Typically. this could be quite slow. For vibration P 154 . And what is the processor doing while it waits? Nothing. Instead it uses data from the previous time record as well as data from the current time record to compute the next FFT. This is why overlap effectively speeds up averaging and smoothes out window variations. The center frequency of each band should be calculated according to ANSI standard S1. time records with 50% overlap provide almost as much noise reduction as non-overlapping time records when RMS averaging is used.

the display shows the real or imaginary part or phase of the complex peak value.Vibration Analysis on Rotating Equipment Peak The maximum value as seen in the time domain data. If the noise is Gaussian in nature. There is no real or imaginary part or phase. This allows measurements with different spans to be compared. 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. Phase In general. Rectangular Window (Uniform or No Window) 155 . it is possible to compute the spectra for every time record with no loss of data. and Octave Analysis measurements. At larger spans. Thus the PSD is displayed in units of V/√Hz or dBV/√Hz. However. Peak Hold detects the peaks in the spectral magnitudes and only applies to Spectrum. it is the frequency span whose corresponding time record exceeds the time it takes to compute the spectrum. This avoids the messy phase display associated with the noise floor. If the real or imaginary part or phase is being displayed for spectrum measurements. At this span and below. This is done on a frequency bin by bin basis. This is because the line width changes so the frequency bins have a different noise bandwidth. usually from -180 to +180 degrees. Since the PSD uses the magnitude of the spectrum. What good is this? When measuring broad band signals such as noise. and if the new data is larger. The phase is displayed in degrees or radians on a linear scale. the amplitude of the spectrum changes with the frequency span. some data samples will be lost while the FFT computations are in progress. 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.s.414 Peak Hold Averaging Peak Hold is not really averaging. then the new data is stored. then the amount of noise amplitude in other bandwidths may be approximated by scaling the PSD measurement by the square root of the bandwidth. even if a signal is small. phase measurements are only used when the analyzer is triggered. The spectra are computed in "real time". (Remember. 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.). the peak magnitude values are stored in the original complex form. the PSD is a real quantity. The phase is relative to the pulse of the trigger. instead.012% of full scale (-78 dB below f. The resulting display shows the peak magnitudes which occurred in the previous group of spectra. The PSD. on the other hand. For a sine wave peak = RMS x 1. the new spectral magnitudes are compared to the previous data. PSD. normalizes all measurements to a 1 Hz bandwidth and the noise spectrum becomes independent of the span. 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.

5 Hz (the same as the lowest measurable frequency) Everything below 2. displaying the real or imaginary part or phase of an RMS average has no meaning. The output spectrum thus represents the frequency range from DC to 1000 Hz with points every 2. S Sampling Rate The original digital time record comes from discrete samples taken at the sampling rate.5 Hz. For a frequency domain spectrum this refers to the number of lines (or bins) of resolution that are combined to display the spectral data. the spectrum has less than half as many frequency points as there are time points.2. For a sine wave RMS = Peak x 0. RMS (Root Mean Square) The square root of the average of a set of squared values. It takes 0. For example a spectrum of 500 Hz Fmax with 400 bins or lines would have a line resolution of 1.7071 RMS Averaging RMS averaging computes the weighted mean of the sum of the squared magnitudes (FFT times its complex conjugate). It does not reduce leakage errors and should only be used for impulsive or transient data that dies out within the time sample period.56 = No samples / 2. RMS averaging reduces fluctuations in the data but does not reduce the actual noise floor. Resolution The accuracy of a reading based on the number of discrete values used to define it. The RMS average has no phase information. The FFT of this record yields 400 frequency points or lines. With a sufficient number of averages.25 Hz. The lowest frequency is just the F-max divided by the number of lines: F-max No.4 Seconds to take this time record. The corresponding FFT yields a spectrum with discrete frequency samples.56 = F-max / No. a very good approximation of the actual random noise floor can be displayed.Vibration Analysis on Rotating Equipment The rectangular windows applies an even weighting (no window) over the time period.56) / (1024 / 2. In fact. but over what frequency range? The highest frequency will be determined by the in-built ratio of F-max to data sampling rate . Route For off-line data collectors this is a an ordered list of points containing the sequence for collecting data. Suppose that you take 1024 samples at 2560 Hz. Sidebands occur because of a modulation of the fundamental by another frequency. Since RMS averaging involves magnitudes only. The weighting is either linear or exponential. Of Lines Bin resolution = data sampling rate / 2.56. Spectrum 156 . of lines = (2560 / 2. Sideband A frequency which occurs either side of a fundamental frequency.5 Hz is considered to be DC.56) = 2.

T Tachometer A device for measuring the speed of rotation. coherence and orbit. A signal will appear as narrow as a single bin if its frequency is exactly equal to a frequency bin. (Square root of the sum of the real (sine) part squared and the imaginary (cosine) part squared). it will affect every bin of the spectrum. The magnitude is a real quantity and represents the total signal amplitude in each frequency bin. 157 . this window is only useful when looking at transients which do not fill the entire time record. independent of phase. U Uniform Window The uniform window is actually no window at all. the magnitude of the spectrum is displayed. These two cases also have a great deal of amplitude variation between them (up to 4 dB).periodic within the time record). 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. The phase is simply the arc tangent of the ratio of the imaginary and real parts of each frequency component. Two-Channel Measurements Two-channel analyzers offer additional measurements such as transfer function. i. the time record is triggered in phase with some component of the signal. If its frequency is between bins. the input spectrum must have amplitude at all frequencies over which the transfer function is to be measured.e. In general. The phase is always relative to the start of the triggered time record. For the transfer function to be valid. The time record is used with no weighting. then the real (cosine) or imaginary (sine) part or the phase may be displayed. It is simply the complex FFT. the time record is a real quantity. Normally. The magnitude is the square root of the FFT times its complex conjugate. cross-spectrum. If there is phase information in the spectrum. (It is exactly. Time Record The time record measurement displays the filtered data points before the FFT is taken For baseband spans (spans that start at DC).Vibration Analysis on Rotating Equipment The spectrum is the basic measurement of an FFT analyzer. For non-baseband spans (zooms) the heterodyning discussed earlier transforms the time record into a complex quantity which can be somewhat difficult to interpret.

the real and imaginary parts as well as phase displays are correctly averaged and displayed. but it will be periodic (zero at each end). Flattop. a window acts like a filter. BlackmanHarris. 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 the time record does not start and stop with the same data value. With vector averaging. since the filter is narrow. then only frequencies near the bin will contribute to the bin.Vibration Analysis on Rotating Equipment V Vector (Synchronous Time) Averaging Vector averaging averages the complex FFT spectrum. 158 . When the time record is windowed. amplitude accuracy. In the frequency domain. This is true but it's actually a bit worse than that. If the filter is wide. and noise floor. Otherwise. time bin by time bin. Hanning. Windows are functions defined across the time record which are periodic in the time record. A narrow filter is called a selective window . and Kaiser. it falls off from center rapidly. If the filter is narrow. The signal of interest must be both periodic and phase synchronous with the trigger. However. This is because the complex information is preserved. (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. 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. 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. its points are multiplied by the window function. then frequencies far from the bin will contribute to the bin amplitude but those close by will not be attenuated significantly. W Windowing What is windowing? Let's go back to the time record. Vector averaging requires a trigger. The different types of windows trade off selectivity. Several types of window functions are available including Uniform (none). The net result of windowing is to reduce the amount of smearing in the spectrum from signals not exactly periodic with the time record.it selects a small range of frequencies around each bin. the real and imaginary parts of the signal will not add in phase and instead will cancel randomly. the signal can actually smear across the entire spectrum. and the resulting time record is by definition periodic. This means that even frequencies close to the bin may be attenuated somewhat. It may not be identical from record to record. They start and stop at zero and are smooth functions in between.

13. 27 Developed Fatigue. 89. 7. 104. 91 bearing. 60 Hz. 87. 3. 62. 103 Envelope. 66. 3. 3. 6. 93. 94. 59. 153. 158 —C— Campbell. 130 Bearing. 41. 156 Analyzer. 6. 36. 103. 15. 7. 123 —I— Imbalance. 42. 37. 127. 32 Elasto Hydrodynamic Lubrication. 78. 43. 157 Blackman-Harris. 74. 91. 32 BPIR. 2. 3. 27. 2. 88. 93. 33 Impact hammer. 39. 75. 156. 63. 152. 22. 2. 30. 157. 77. 150 Accelerometers. 3. 13. 153 Hydraulic valve. 67. 32 Integrate. 93. 38. 155 Firing Cards. 80. 117. 148. 104 Imbalanced phases. 152 Demodulation. 153 Armature. 100. 93. 31. 10. 59 Air. 111. 70. 154 Frequency. 3. 23. 92 —F— Failure type. 2. 153. 16. 13. 28. 34. 14. 159 Decibel. 127. 3. 91. 108. 71. 22. 21. 64. 7. 22. 87. 19. 132 Boltzman. 102. 130. 100. 106. 6. 153. 37 Flattop. 119 Bin. 152. 101. 93. 4. 50. 158. 45. 3. 41 Compression mode. 23. 110. 38. 127 —E— eccentric rotor. 6. 26. 85 Base cause. 2. 10. 159 Fourier. 103. 41. 67. 26. 124 Choking. 131. 113. 72. 69 FFT. 38. 120 Hertz. 5. 150. 100. 7. 149. 3. 145. 26. 59. 67 Differentiate. 152 Accelerometer. 41. 78. 2. 25. 3. 6. 6. 21. 30. 20. 93. 143. 79. 2. 45. 13 Bump test. 159 Filter. 120 Analysis. 101. 114 Benzene. 120 Induction Motors. 35. 3. 156. 152 Enveloping. 72. 153 —B— Balancing. 155. 118. 28. 61 Coherence. 64. 14. 6. 6. 121 Glycerin. 36. 15. 110. 92. 2. 62 envelope. 92. 40. 18. 146 DC. 151 Comparitor Card. 6. 63. 2. 129. 101. 131. 153 Fundamental. 104. 3. 118. 151 Bode. 24. 9. 109. 154. 82. 159 Harmonic. 5. 60 aerodynamic forces. 154 Displacement. 21. 135. 157. 60. 152 Averaging. 70. 68. 151. 75. 120 hysteresis whirl. 90. 140. 46. 29. 158 159 . 65. 2. 152. 28. 155. 6. 126 Critical speed. 3. 106. 103. 45. 130. 7. 22. 134. 18. 28. 139. 90. 154 acceleration. 6. 24 Aerodynamic cross coupling. 11. 49. 132 Impeller. 80. 152. 27 External manifestation. 106. 9. 151. 69. 26. 47. 151 CPM. 69. 21. 6. 39. 99. 130 —G— Gas. 102. 4. 158. 95. 131. 115. 3. 7. 66. 117. 112. 67.Vibration Analysis on Rotating Equipment Index —A— AC. 103. 78. 151 Cracking. 9. 29. 94. 70. 42. 153. 40. 17. 133. 122. 4. 6. 70. 45. 154. 117. 71. 6. 20. 119 —H— Hanning. 43. 142 fans. 156. 60. 152 displacement. 99. 3. 94. 154 accelerometer. 121 bowed rotor. 34. 62. 6. 74. 17. 100. 40. 13. 7. 153. 65. 154 —D— Damping. 91 fan. 153 Harmonics. 68. 154. 115. 114. 32 Average. 41. 154. 2. 119. 99. 19. 107. 3. 100. 48. 121. 12. 104. 99. 73. 13. 24 Correlation. 13. 150. 69. 2. 7. 71. 157. 9. 43.

6. 114 Mode shape. 4. 94. 91. 130. 20. 7. 115. 65. 104. 113. 103. 4. 92 PEAK. 72. 23. 51. 3. 146 Mercury. 83. 4. 155. 3. 74. 99. 101. 118. 155 —P— Parameter. 144. 100. 101. 45. 91. 157. 73. 2. 118. 111. 120 vector. 156 Orbit. 21. 118. 104. 120. 119. 11. 7. 15. 157. 143. 101. 3. 7. 22. 108. 118. 40. 16. 25. 55. 30. 135. 104. 31. 32. 22. 3. 4. 82. 57. 151.R. 3. 24. 2. 152. 60. 43. 22. 29. 117. 157 resonance. 129. 6. 85. 74. 26. 26. 64. 2. 33. 100. 22. 139. 50. 50. 36. 2. 73. 7. 31. 157 rotor bars. 94. 147. 116. 106. 142. 2. 4. 4. 80. 4. 106. 50. 155 Overlap. 124 160 . 79. 151. 2. 6. 126. 127. 41. 109 —O— Octave. 133. 61. 117. 20. 153. 158. 69. 145 —K— Kaiser. 101. 5. 33. 106. 107. 28. 69. 156. 64. 9. 158 Speed of sound. 24 Sideband. 156. 41. 121. 94. 140. 12. 142 Newton. 106. 99. 10. 158 Triggering. 61 Surging. 18. 33 Mass. 6. 34. 99. 108. 104. 127. 157 —W— waterfall. 53. 131. 94. 50. 7. 104 lubricant wedge. 60 —L— Leakage. 34 Route. 76 Transfer Function. 21. 69. 129. 64. 37. 153. 106. 15 Single Plane. 120 PRELOAD. 130. 153. 99. 133. 91. 154. 19. 20. 13. 7. 33. 131. 30. 21. 154 Looseness. 154 Line. 122. 83. 23. 52. 11. 142. 9. 150. 18. 4. 4. 69. 158. 92 RUB. 114. 50. 158 Simple Harmonic Motion. 152. 6. 100. 13. 124 waveform. 4. 132 —T— Tachometer. 3. 156. 154. 152. 155. 14. 38. 2. 7. 101. 72. 15. 133. 156 Piston pulse. 4. 20. 110 Overall. 92 time. 4. 17. 158. 71. 31. 65. 29. 13. 25. 127 mounting techniques. 89. 158. 93. 109. 117. 11. 123. 154. 62. 27. 116. 154. 34. 113. 26. 10. 18. 94. 69. 132. 101. 14. 19. 4. 130 whirl. 7. 30. 19. 155. 102. 2. 87. 127. 141 RMS. 157. 155. 3. 101. 107. 105 Shear mode. 154 Liquids. 39. 113 Rubber. 139 —R— Resolution. 127. 84. 29. 118. 159 tolerance stack up. 128. 23. 151. 124. 16. 159 Time. 123. 6. 121. 111. 90 —N— Natural frequency. 92. 155 Orbits. 10. 154. 145. 7. 28. 6. 63. 156. 79.C. 114. 59. 92. 2. 19. 110. 121 Spring. 73. 18. 8. 154. 112 —U— Uniform. 91. 6. 141. 54. 92 Severity. 27. 119. 101. 102. 124. 122. 113. 73. 38 Setup. 146. 4. 92. 15. 112. 100. 32. 80. 117. 158 Technology. 33. 64 —M— magnetic center. 59. 140. 84. 72. 142 Stone Walling. 6. 135. 2. 58. 129. 3.Vibration Analysis on Rotating Equipment Interval. 118 Logarithmic. 91. 119 Misalignment. 82 spall. 115. 95 Phase. 33. 81. 157. 159 Spectrum. 101. 8. 103. 94 spectrum. 143. 100. 159 velocity. 31. 112. 18 Peak Hold. 22. 121. 102. 123. 7. 99. 129. 109. 18. 130. 150. 112. 158. 128. 70. 93. 32. 33. 154 Vibration isolators. 107. 159 —V— Vane pass. 66. 2. 139. 150 Nyquist. 159 —S— S. 102. 153 whip. 156 PFA. 117. 112. 94. 6. 15. 149 Resonance. 34 Rotor Bars. 7. 56. 84.

Vibration Analysis on Rotating Equipment Whole body vibration. 159 161 . 149 windings. 99. 32. 33. 148. 71 Windows. 4.

- Level 3 NVQ Refinery Ops Control Room NOS
- Trib Ology
- Vibration Analysis Measurement Manual
- Reference%20Standards%20for%20VibrationMonitoringandAnalysis.pdf
- microlog bumptest.pdf
- Recommendations for the Safe and Reliable Inspections of Atmospheric, Refrigerated Ammonia ST
- Spec - Thermal Imaging System
- Guardian Angel Flowchart
- Body Shop Thermography Examples
- DC Motor Brush Life
- VSD Review
- MV VFD Spec
- RBI and FFS
- Risk Based Inspection Jacketed Platforms
- Integration of safety in machine design
- Smart Refinery
- Pre Shutdown Checklist
- Life Cycle Costing
- Automatic Lubricators
- POF Procedure
- Standard Breakout Insp
- Pressure Vessels Rupture
- Mntce Planning CMMS
- Risk Based Decision Making
- Vortex Shedding in Tubes

Sign up to vote on this title

UsefulNot useful- Vibration Intermediate
- Vibration Adv 0402
- Vibration Analysis
- Vibration analysis lectures
- Vibration Training Quick Reference
- Vibration Basic 0400
- Vibration Diagnostics
- Vibration Analysis Handbook - James Taylor
- Vibrations in Rotating Machinery (7th Int'l Conf.) (IME, 2000) WW
- IntroductiontoBasicVibrationAnalysis
- Level II Metric
- VA II Training
- Rotating Machinery Vibration - From Analysis to Troublesh Ooting
- Vibration
- Basic Vibration Analysis
- Machine Vibration ( a Beginners Guide)
- VIBDI_skriptaEN
- Fundamentals of Rotating Machinery Diagnostics
- Section II - Basic Vibration Theory
- Vibration Analysis Level- 1 Updated [Compatibility Mode]
- Vibration of pumps
- Vibration Diagnostics
- Machinery Failure Analysis & Trouble Shooting
- Machinery Malfunction Diagnosis and Correction Robert Eisenmann (1)
- Vibration Diagonistic Chart.ppt
- Rotating Machinery Practical Solutions to Unbalance and Misalignment - Robert B McMillan,
- Vibration Training Mobius1
- TABLA CHARLOTTE_1
- Basics of Vibration Analysis
- PeakVue Training
- Vibration Analysis Rotating Equipment