100% found this document useful (1 vote)
239 views32 pagesSKF FFT
SKF - Fault Diagnosis using Spectrum Analysis
Uploaded by
Mohammadreza HadipourCopyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
100% found this document useful (1 vote)
239 views32 pagesSKF FFT
SKF - Fault Diagnosis using Spectrum Analysis
Uploaded by
Mohammadreza HadipourCopyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
/ 32
Spectrum Analysis
The key features of analyzing spectra
By Jason Mais * SKF USA Inc
Summary
This guide introduces machinery maintenance workers to condition monitoring analysis methods used to detect and analyze machine com-
ponent alures.Itinforms the reader about common analysis methods. It intends to lay the foundation far understanding machinery analysis
concepts and show the reader what s needed to perform an actual analysis on specific machinery.
tet Contents
1. Introduction
Common steps in a vibration monitoring program.
3. Step
3.1, Collet useful information
3.2 Identify components ofthe machine that could cause
vibration,
3.3. Identify the running speed
3.4, Other key considerations —
35. Identify the type of measurement that produced the
FFT spectrum
4. Step2
44, Analyze spectrum
4.2. Comman components of vibration spectrums.
4.3, Identify and verify suspected fault frequencies
4.4, Determine fault severity
5. Misalignment.
6. Unbalance
7.
8.
Mechanical looseness
Bent shaft.
9. Rolling element bearing defects
10. Gears.
11, Blades and vanes,
12. Electrical problems
13. Step 3
13.1. Muti-parameter monitoring
14, Conclusions .
15. Further reading
11
16
23
25
27
28
28
29
29
SKF 1. Introduction
A vibration FT (Fast Fourier Transform) spectrum isan incredibly useful tool for machinery vibration analysis, Ia machinery problem exists,
FFT spectra provide information ta help determine the source and cause ofthe problem and, with trending, how long until the problem
becomes critical
FFT spectra allow us to analyze vibration amplitudes at various component frequencies onthe FFT spectrum. In this way, we can identity
and track vibration occurring at speific frequencies. Since we know that particular machinery problems generate vibration at specific fe
|quencies, we can use this information to diagnose the cause of excessive vibration.
“The key focus ofthis article hinges onthe proper techniques regarding data collection and common types of problems diagnosable with
vibration analysis techniques. This article can be used asa reference source when diagnosing vibration signatures.
Fig 1. Example of velocity spectrum that contains running speed (at F = 2 700 RPM or 45 Hz}, harmonics of running speed (at F = & 500 RPM or
75 He) and bearing defect frequencies (at F= ~34 000 RPM (526 Hz) and ~39 000 RPM (650H.) marked with bearing overlay markers).
SKF 3 2. Common steps in a vibration monitoring program
“There are several steps to follow as guidelines to help achieve a successful vibration monitoring program. The following is a general ist of
these steps:
1 Callet useful information —Look, listen and feel the machinery to check or resonance. Mentfy what measurements are needed point and
point type). Conduct addtional testing if further data is required,
2 Analyze spectral data — Evaluate the overall values and specific frequencies corresponding to machinery anomalies. Compare overall values
in diferent directions and current measurements with historical data
3 Malt-parameter monitoring -Use addtional techniques to conclude the faut type. (Analysis tools such as phase measurements, current
analysis, acceleration enveloping, il analysis and thermography can be used.)
4 Perform Root Cause Analysis (RCA) ~ In order to dent the real causes of the problem and to prevent it from occurring again,
5 Reporting and planning actions ~Use a Computer Maintenance Management System (CMMS) to rectify the problem and take action to
achieve a plan,
In this article, only stops 1 through 3 are investigated, The readers refered to ather SKF @ pttudeXchange articles on RCAand CMMS to
‘explain these additional monitoring technologies,
3. Step1
3.1. Collect useful information
When conducting a vibration program, certain preliminary information is needed in order to conduct an analysis. The identification of
‘components, running speed, operating environment and types of measurements should be determined iil to assess the overall system,
3.2. Identify components of the machine that could cause vibration
Before a spectrum can be analyzed, the components that cause vibration within the machine must be identified, For example, you should be
familar with these key components:
+ Ifthe machines connected toa fan or pump, tis important to know the numberof fan blades or imple.
+ Ifbearings are present, know the bearing identification number ors designation.
+ Ifthe machine contains, ors coupled, toa gearbox, know the number of teeth and shaft speeds.
+ Ifthe machines driven with belts, know the belt lengths
‘The above information helps assess spectrum components and helps identify the vibration source. Determining the running speeds the
initia task. There are several methods to help identify this parameter.
3.3. Identify the running speed
Knowing the machine's running speed is critical when analyzing an FFT spectrum. Running speeds related to most components within the
machine and therefore, ads in assessing overall machine health. There are several ways to determine running speed
* Read the speed from instrumentation at the machine or from instrumentation in the control raam monitoring the machin.
* Look for peaks inthe spectrum at 1 800 or 3 600 RPM (1 500 and 3.000 RPM for SO Hz countries) ifthe machine an induction electric
‘motor, as electric motos usually run at these speeds. Ifthe machine is variable speed, look fr peaks inthe spectrum thatare close to the
running speed ofthe machine during the time at which the data is captured,
+ An FFT’srunning speed peakis"typically” the frst significant peakin the spectrum when reading the spectrum from left to right. Look or
this peak and check for peaks at two times, three times, four times, et. Multiples ofthe running speed frequency can be an indication of
machine heath
4 SKF 3.4. Other key considerations
‘There are many other considerations to take inte account when analyzing a machine. For example:
« Ifthe machine operates inthe same vicinity as another machine, itis important to know the running speed ofthe adjacent machine.
Occasional, vibration from one machine can travel through the foundation or structure and affect vibration levels on an adjacent machine
+ Know f the machine is mounted horizontally o vertically. Mounting orientation affects machine response to vibration.
* Know ifthe machines overhung or connected to anything thats overhung. Machine support can affect the response of the vibration
3.5. Identify the type of measurement that produced the FFT spectrum
bration monitoring programs use many ype of measurement oGeterine the condon of machinery tis importantto determine which
type of measurement esplaysthe required
+ Was the measurement displacement, velocity, acceleration, acceleration enveloping, et? Depending upon the information needed, a
particular measurement shouldbe tailored to capture the proper results,
‘How was the sensor positioned: horizontal, vertical, axial inthe load zane, e.? Sensor response varies depending upon maunting
orientation.
«+ fre previously recorded values, FFTs or overall trend plots avalable? History can help determine a machines normal vibration level, or how
auickly a machine is degrading,
4. Step 2
4.1. Analyze spectrum
Once machine vibration identfeation and callectonis completed, the process of analyzing the specrum canbe conducted, Aas usualy
follows a proces of elimination lminate the components or sues that donot contrite tothe syst, From the remaining components
identify whats the contributing factor affecting the machine heath,
4.2. Common components of vibration spectrums
The most common components of vibration spectrum shouldbe analyzed inal to determine whether ornotthe spectrum indicates a
posible prabiem
* Compare overall measurement values to prior measurements to determine if a significant increase has occurred.
‘Evaluate the alarm status of a measurement point. overall alarms are set property, this can help indicate when a measurement needs
further evaluation,
+ Identify the typeof measurement that indicates a problem. For example, enveloped signals can indicate bearing damage or gear tooth
damage, while velocty measurements relate more to overall machine health
Once an assessment ofthe measurement is conducted, specific frequencies should be identified
SKF 5 4.3. Identify and verify suspected fault frequencies
Spectra may produce peaks at identified fault frequencies, These peaks may or may not represent the indicated fault. By looking for harmon
ics ofthe fault frequency, aditonal information can be assessed as to whether the generated frequencies are an indication ofthe fault, For
example
* Ifa peak appears atthe fundamental faut frequency and another peak appears a two times (2x) the fundamental faut frequency, itis a
very strong indication thatthe faults real.
* Ifna peak appears a the fundamental fault frequency, but peaks are present at two, three and maybe four times the fundamental fault
‘frequency, there isa strong indication the faults valid
Identifying any harmonics of running speed (2x, 3x, etc) helps determine ifafaultis present.
dentifying any bearing fault frequencies helps determine if a faults present
dentifying fan or vane pass frequencies if applicable, helps determine if faults present.
dentifyng the numberof gear teeth and the shaft on which the gears mounted, if applicable, helps determine if a faults present.
Morecver this helps determine if there is a problem with a particular gear,
* Identifying pum impeller frequencies, f applicable, helps determine faults present.
‘+ As mentioned inthe prior section, identifying adjacent machinery vibration, if applicable helps determine if faults present.
Once the vibration source is determined, the level of severity must be assessed to evaluate whether corectve action shouldbe taken,
4.4. Determine fault severity
Great importance should be placed upon determining the severity of particular fault. Some components of a machine may vibrate at very
high levels and stil be operating within acceptable imi. Other components maybe vibrating at very low levels and be outside acceptable
limits. Thus, amplitude is relative, so the entire syster should be evaluated, not just the amplitude.
* Compare the amplitude wth past readings taken while operating under the same consistent conditions to determine the severity.
* Compare the amplitude ofa particular reading withthe same type of reading from a similar machine. higher than normal reading on one
‘ofthe machines may indicat a problem in that particular machine.
+ Obtain prior history onthe machine to help identify the various levels at which the machine has operated and ads in assessing machine
health atts currentstate
+ Determine whether or not a baseline measurement (a measurement taken upon installation ofa new or reconditioned machine) was
taken. Ifso, compare the new reading tothe baseline reading to help indicate the severity of vibration.
(Once the information is collected and components are identified, you can begin to use the collected spectra to diagnose machinery problems.
“The following sections help evaluate common machinery probleme and identify their associated causes and effets. In addition, examples of
resulting spectra are included to use as templates when identifying these common issues. Issues such as misalignment, unbalance,
looseness, bent shafts and bearing defects are discussed,
6 SKF 5. Misalignment
Misalignmentis created when shafts, couplings and bearings are not propery aligned along ther centertines. The two types of misalignment
are angular and parallel, ora combination of bath
Angular misalignment
‘Angular misalignment occurs when two shafts are joined at a coupling in a manner that induces a bending force on the shaft (-> fig. 2).
Fig. 2. Angulor misalignment
Parallel misalignment
Parallel misalignment occurs when the shaft centertines are parallel but displaced or offset (— fig. 3)
Fig, 2. Parallel misalignment.
Causes
Common causes of misalignment are:
‘+ Thermal expansion: Expansion or growth of @ component due tothe eating and cooling ofthat companent.
Cold alignment: Most machines are aligned cold and heat as they operate, Thermal growth causes them to grow misaligned.
‘Alignment of components during coupling isnot correctly achieved. Therefore, misalignments introduced into the system during
installation.
+ Improper alignment due to imparted forces from piping and support members.
‘Misalignment due to uneven foundation, shifting in foundation or settling
SKF Effects
Misalignment usually causes the bearing to carry a higher load than ts design specication, which may cause bearing fallure du to early
fatigue. Fatigue isthe result of stresses applied immediatly below the load carrying surfaces and i observed as spalling of surface metal.
Effects on coupling inthe form of damage tothe coupling or excessive heat due to fiction can also be seen. Fig indicates misalignmentin
the system.
‘Fig 4, High 2x running speed peak at 3600 RPM or 60H (running speed s 4 800 RPM or 30H) isan indication of misalignment. The fst peok is
‘most ely abet frequency due too worn or lose drive belt. The second peaks the running speed af the machine (2800 RPM).
Note: 2X amplitude Is net aways present.
8 SKF Diagnoses
The most effective analysis techniques commonly use overall vibration values and a phase measurement that help distinguish between
Various types of misalignment or unbalance
‘Acommon practice when analyzing misalignments to look atthe ratio between 1x (unbalance) and 2x (misalignment), and compare the
values. When analyzing an FTT spectrum where misalignment is indicated, a higher than normal 1x amplitude divided by 2x amplitude may
‘occur. The indication of amplitude can vary from 30% of the 1x amplitude to 100 to 200% ofthe 1x amplitude. An example of this is seen in
fig. 5. The 2x amplitude (0,90 mmfsec)s almost twice that of 1x(0,45 mmised).
Fig. 5. FFT spectrum showing severe misalignment (the second peok in the spectrum at ~8 500 RPM (144 H)indcaes severe misalignment, as itis
‘atmos twice the amplitude of the running speed te peak marked wih the marker running speed (4237.5 RPM (74H2)).
‘With concern to couptng, some general rules are applied:
+ Couplings with 2x amplitudes below 50% of 1x are usualy acceptable and often operate fora long perio of time.
‘+ When the vibration amplitude at 2xis 50 to 150% of fx, itis probable that coupling damage will occur
‘+ Amachine exhibiting vibration at 2x running peed that is greater than 150% ofthe 2xindicates severe misalignment. The machine should
be scheduled for repair as son as possible.
‘As mentioned earlier in this section, phase readings are another factor that help determine the precise problem.
SKF 9 Phase analysis
Phase measurements area very useful ool for diagnosing misalignment. possible, measure the phase shift between aval readings on
epposite ends of the machine
Note: All phase values are #30° because af mechanical variance,
‘Angular misalignment: Inthe axial position, a phase shift of 180° avstsacrassthe coupling or machine
+ Paralle! misalignment Inthe radial direction, a phase shift of 180° exists across the coupling or machine, AO® or 180° phase shift occurs as
the sensors moved from the horizontal to the vertical position onthe same bearing
+ Combination angular and parallel misalignment: In the radial and axial postions, a phase shift of 120° exists across the coupling or
machine
Note: With severe misalignment, the spectrum may contain multiple harmonics from 2x to 10x of running speed, If vibration amplitude in the
horizontal plane increased two or three times, misalignments indicated again.
Bearing cocked on a shaft
Like misalignment, a cocked bearing (algned improperly in the housing) usually generates considerable axial vibration, However, phase mea
‘surements from the axial position help differentiate the two.
Fig. 6. Four sensor locations.
Ifthe phase readings among the four sensor locations in ig. 6 (12 veloc, 3 o lock, 6 atlack and 9 oclock) vary considerably, a cocked
bearingis indicated
Summary
+ thers abnormal high 2xampltde dvdedby 2x amptude andthe tem contains acuplng or ether may be misalgnment
+ theradial2xamptudes abnormally gh andthe system contains coupling ore, thee may be malgnment
* Heal xamplue is abrormaly igh ad the system contains a coupling or be, there may be misalignment.
10 SKF 6. Unbalance
Another common indication of poor machine health is unbalance inthe system. Unbalance can cause excessive forces that affect the machine.
Unbalance occurs when the shafts mass centerline does not coincide with ts geometric centerline, In general, there are three types of
Lunbalance:
* Static unbalance
* Couple unbalance
* Dynamic unbalance
Fig. 7. Point isthe unbalance weight a
distance fram the center ofthe rato or disk
This typeof station in a machines considered
Ainbolane.
Static unbalance
With static unbalance, only one force sinvolved. For example, if you have a bicycle ire that has mud buildup on one area or portion of the tire,
when stopped the wheel naturally settles with the clump of mud atthe bottom of the wheel. Similarly, arotor turns untilthe heavy spot is
located at 6 olock(- fig. 8). The term "static" implies that this typeof unbalance can be observed at rest
Fig. 8. Static unbolance.
SKF a Couple unbalance
Unlike static unbalance, couple unbalance cannot be measured at rest. With couple unbalance, two equal forces (weights) are 180° from each
ther whic eauseste rotor to appear balanced atrest(-> fig 9), However, when the rotor rotates these forces mave the rtorin oppose
tirecion at ther respective ends ofthe shat. This causes the atria wobble, which produces a 180° out-of-phase reading from opposite
ends ofthe shaft.
Fig. 9. Couple unbalance.
Dynamic unbalance
In realty, most unbalance is dynamic. Dynamic unbalance isthe combination of static and couple unbalance, On simple machines, theres,
usually more static unbalance than cauple unbalance. On more complex machinery, with more than ane coupling or several areas onthe rotor
where unbalance can occur, couple unbalance is usually more prominent nthe system.
‘When balancing a machine, always balance aut static unbalance first, then couple unbalance. When balancing couple unbalance, itis
important to nate that balancing must occur within several planes.
Cause
Unbalance can be caused by a numberof factors, Several examples include:
+ Improper component manufacturing
* Uneven buld up of debris on the rotors, vanes or blades
+ The addition of shaft fitings without an appropriate counter balancing procedure
* Vane/Blade erosion or thrown balance weights
Key characteristics of vibration caused by unbalance:
+ Itisa single frequency vibration whose amplitude isthe same in al-ratial directions
+ Itis sinusoidal, occurring ata frequency of once per revolution (x)
+ The spectrum generally does not contain harmonics of x running speed, unless the unbalance is severe
+ Amplitude increases with speed
Effects
Just tke misalignment, unbalance usually causes bearings to carry a higher dynamic oad than their design specifications, which causes.
bearing ofall from early fatigue. Fatigue, ina bearing, isthe result of stresses applied immediately below the load carrying surfaces ands.
‘observed as spalling away of surface metal
Diagnoses
‘The use of overall vibration, FFT spectra and phase measurements aidsin diagnosing unbalance problems,
FFT spectrum analysis
Vibration caused by pure unbalance i @ once per revolution sinusoidal waveform. On an FFT spectrum, this appears as a higher than normal
‘Lx ampltude. While other faults can produce high 1x amplitude, they usualy also produce harmonics. In general, ifthe signal has harmonics,
_above once per revolution, the fault is not unbalance. However, harmonics can occur as unbalance increases or when horizontal and vertical
‘support stfness differs by a large amount,
2 SKF Phase analysis
The use of phase measurements isin the agnoss of unbalance probe
Note: All phase readings are *30° due to mechanical variance,
“The sensor shows a 90° phase shift between the horizontal and vertical positions when readings are compared, When the system involves
predominantly static unbalance, there is usually ne phase shift across the machine or coupling in the same measurement position.
Summary
Sane general idles arphae ratios lon
« Ifthe racial measurement’ 1x amplitude is high and harmonics (except vane passing) are less than 15% ofthe 1x, there may be unbalance,
«+ Ifthe majority of vibration is inthe radial plane, the 1x armplitude fs medium tohigh in amplitude and the phase from the vertical and
horizontal measurements difes by 90°, there may be unbalance.
« Ifthe primary vibration plane/s oth axial ad radial, the machine has an overhung mass and the aval phase measurements across the
‘machine aren phase, there may be unbalance.
The spectrum on the following page (-> fig. 10)is an indication of unbalance ina machine.
Note: Increasing unbalance forces place increasing loads on nearby bearings the bearings specified loa is exceeded, damage can occur
and the bearing’ ies drastically reduced.
Fig. 10. FFT showing unbalance inthe spectrum (at F= 4237.5 RPM or 70 He) Additionally, ther isan indication of sight misalignment (the smaller
peak) a 475 RPM or 161 He.
SKF B 7. Mechanical looseness
‘Along string of rotating frequency harmonics or 3/2 rotating frequency harmonics at abnormally high amplitudes generally characterizes
‘mechanical looseness or an improper ft between component parts (-> Fig. 14}.
Note: These harmonics may be random and unorganized. For example, looseness may display peaks at 2x, 3, 6x, 5x, 6X, ete, or at 3X, 3.5%
4, 5.5x, 6x ote
Causes
Possible causes of wear/loosenes are:
+ The machine came loose from its mounting
+ Mounting is cracked or broken
+ Amachine component came loose
The bearing developed a faut, which wore down bearing elements, or the bearing seat caused excessive clearance in the bearing
Effects
Ifloosenessis generated from a component, there’ a possibilty the part wll become detached and cause secondary damage.
Diagnosis
Loaseness can be exhibited in varying amplitudes, both overall bration and individual frequency amplitudes. Loosenessis best diagnosed
using FFT spectra and phase
Spectrum analysis
Fig, 14 deplays vtaton signatures asaciated with components or ystems. Typcaly looseness dente by abnormal igh runing
speed amplitude followedby mutles 3/2 mules, Harmonic peaks may decrease in ampitude as they increase in frequency (except at
2 which, when measure inthe vertical peston, canbe higher in amp)
Summary
* Hftere area series of three oF more synchronous or /esynchronous multiple of running speed (range 2xto 10x), and their magnitudes
are greater than 208 of the 1x, there may be mechanical looseness.
+ Ifthe machines rigidly connected (no coupling or belt), and the radial 2xis high there may be mechanical looseness.
4 SKF Fig. 14. FFT spectrum indicating looseness in the machine, Ntie all ofthe repeating multiples of running speed or4/2 of runing speed.
8. Bent shaft
‘With overall vibration and spectral analysis, bent shaft problem usually emits a vibration signature that appears to be identical toa
‘misalignment problem. The use of phase measurementsis needed to distinguish between the two,
Causes
There are several causes that can resultin a bent shat
+ Cola bow: shaft with high lenath-to-width rato can, atrest, develop a bend
+ Improper handling during assembly or transportation
+ High torque
Effects
‘As with unbalance, a bent shaft usually causes the bearing to carrya higher dynamic lad than ts design specification, which causes the
bearing to fal
Diagnosis
“The use of veal bration measurement, spec naj and phase measurements can be eectnve to analy abent shat.
Spectrum analysis
[Abentsbat pica produces spectra that have misalignment ype characteristics. higher than normal x divided by 2x amplitude may
cut High 2x amplitude can vary rom 30% ofthe 2x amplitude to 100 to 200% of the x amp.
SKF 5 Phase analysis
Phase measurements are essential when diagnosing a hen shat.
Note: All phase values are #30°, Radial phase measurements (vertical and horizontal) typically appear in phase” withthe shaft
‘Axial phase measurements are typically 180° out of phase with the shaft. IFboth ofthe prior conditions are true the problem s most Ukely a
bent shat.
Summary
the primary wan pane isn tea cretion, theres a dominant Ixpeak nd there 2180" phase dferencein the aval drecton
oss the machine, there may bea bent shat
9. Rolling element bearing defects
Mastoften the bearing defects nat the source ofthe problem, Usually; some other machinery component or lubrication problem causing
the bearing defect. When a bearing defects detected, you should automatically look for other roat cause problems such as misalignment and
unbalance. Then schedule the repair ofboth the defective bearing and the fault causing the bearing defect
Bearing defects
‘To understand how te monitor bearings, an understanding of how a bearing defect progresses shouldbe achieved,
Note:
he following discussion relates to typical spall or crack type bearing defects on reling element bearings
Bearing failure may be caused by:
Ineffective lubrication
* Contaminated lubrication
+ Heavier loading than anticipated
+ Improper hanaling or installation
+ Old age (subsurface fatigue)
* Incorect shaft or housing fits
+ False bineling due to external vibration sources waile machine stands till
+ Passage of current through bearing
Often, inital bearing fatique results in shear stresses cycically appearing immediately below the load-carrying surface. After time, these
stresses cause racks that gradually extend tothe surface. As a roling element passes over these cracks, fragments break away. This is known
2 spalling or flaking. The spalling progressively increases and eventually makes the bearing unusable. This ype of bearing damages a
relatively long process and makes is presence known by increasing noise and vibration
Figure 12. Spllng or flaking on the outer rng ofa bering.
16 SKF Another type of bearing failures intiated by surface distress. Surface distress causes cracks to form on the surface and grow into the mate-
rial. Surface distress is usually caused by excessive load or improper lubrication. In both cases, the fling bearing produces noise and vibra
tion signals, thatf detected, give the user adequate time to correct the cause ofthe bearing problem or replace the bearing before complete
failure
‘Acceleration enveloping is an effective tool to detect and moritor the early stages of bearing failure caused by local defects. Again this pro-
vides enough pre-warning time ta possibly correct the cause of the bearing problem and potentially extend the bearing’ life. Acceleration and
velocity vibration measurements are also useful tools for measuring the final stages of a bearings fe. These measurements typically provide
indications of imminent bearing failure (less than 10% of residual bearing lif).
Velocity measurements
‘The prior examples and many other types of problems can cause bearings to fll. Its important to assess and understand the proper types of
‘measurements to take and ther results. One of the mast common measurements used in vibration analysis is velocity These measurements
are very useful for detecting and analyzing low frequency rotational problems such as unbalance, misalignment, looseness, bent shat, ee
The following section describes velocity measurements and provides an 'SO classification to help determine severity levels
‘Table 1 illustrates the $0 2372 Standard for an overall severity of vibration. Please keepin mind that the levels are machinery and
environment dependent, and have tobe fine tuned in practice.
Velocity Range Limits and
Vibration ee aes
Severity (50'Standard 2372-197
ean Tage Mecines
a Mtoe Fee
Suppers
on ‘beri
Table 4.150 2372 Standard.
SKF a Vibration — spectral analysis
Due to the nature of bearing defect frequencies, they occur at much higher frequencies and much lower amplitudes than frequencies related
to unbalance and looseness, ISO severity charts were not developed to aid in setting parameters for detecting early bearing degradation. For
hearing relatedissues, itis important to evaluate the bearing’s FFT spectrum and its related defect frequencies.
‘To help determineif machine problems include a faulty bearing, hearing defect frequencies can be calculated and used as overlays to aidin
siagnosis. There are several naming conventions that were adapted for use when discussing frequency analysis. The two most common
‘conventions are listed below. The four primary bearing frequencies:
Fug~Frequency Outer Race Defect
Fia~ Frequency nner Race Defect
Fyg~ Frequency Ball Defect
Fr=Frequency Cage
or
+ BPFO ~ Bal Pass Frequency Outer Race
+ BPFI~ Ball Pass Frequency Inner Race
+ BSF Ball Spin Frequency
+ FTF—Fundamenta Train Frequency
[When the defeet frequencies (Fas Fea Faas) align with peak amplitudes inthe vibration spectrum itis commonly accepted that there are
defects within that particular component of the bearing, Notice that a ball defect frequency is by definition twice the ball spin frequency, as the
ball defect hits the inner and outer race during one rotation
Note: In many condition monitoring pragrams,the following are interchangeable. The use of one set or the ather sets suggested, but do not
interchange them,
If bearing analysis software isnot available, bearing defect frequencies should be mathematically calculate.
Fg (n/2} RPM/60) (1 = (Bd/Pd) (cos 0}
Fra (V2) [RPMI60) (1+ (Bd/Pd) (cos o))
Fua= 2* (/2)(Pa/8d) (RPM/60) [1 ~(Bd/Pd}?cos?o}
F, = (1/2) (RPM/60) (1 -(Bd/Pd) (cos 0)
Where:
= number of balls
Bd = ball diameter
Pa « pitch diameter
‘9 contact angle
18 SKF Fig. 13 shows a typical bearing defect nits Final stages. The size and width of the hump at ~9x running speed indicates thatthe defects
approaching failure. n early stages, this hump may appear as non-synchronous peaks, or may not exist.
Fig. 13. Velocity measurement with typical bearing frequencies indicated asa “hum inthe spectrum at approximately x running speed. The other
aks tothe left side ofthe spectrum are unbalance, misalignment and same laseness due to the loss of lading properties.
SKF 9 Acceleration enveloping spectral analysis
Inthe early stages of degradation a bearing defect may not be detectable on normal acleraton or veld vibration spectra. Thisis due in
pat
‘+ The vibration thas presenti the bearing frequency range may nat be shown by the FFT.
‘+ The vibration’s amplitude isso small that low frequency rotational vibrations maski
Acceleration enveloping measurements moritor bearing frequency ranges at which the defects repetitive impacts occur and fier out all non
‘epetitive impact signals ie. low frequency rotational events}. The repetitive impact signals are enhanced and appear as peaks atthe defects
frequency. To asist in determining fa machine's problems include faulty bearing, bearing defect frequencies can be calculated and overlaid
‘onthe vibration spectra,
‘The enveloped time domain of an acceleration measurement and spectra for an inner ring defect are shown in fig 24. When collecting
acceleration enveloping readings itisimportanto also collect time damain data, Time domain data can be very usefulin the diagnosis of
vibration problems in components such as gears and bearings. Figs. 14 through 19 show examples of spectrum and time waveform data All
‘ofthe ilustrations contain captions to describe each figure ands data,
ig. 14. nner ring defect frequencies displayed in an enveloped spectrum. The ist peo, from lett right, is running speed (5 775 RPM) The arge
‘peoke at "51 000, 125 000 RPM, etc. are peak in the spectrum related tthe defec frequency ofthe ner rng ofthe bearing. These peaks indicate a
Dossible defect on the bearings nner rng.
20 SKF Fig 15. Enveloped spectrum with outer race defect ond bearing frequency overlays. This spectrum indicates a defect present onthe beoring’s outer
Fig 16. Enveloped time waveform defect cuter race). The defects indicated by the modulation of ths signal. The expansion and contraction ofthe
‘peas from a high amplitude (2 gE) then toward the center (0) Indicate that energy is Boing generated as the rolling element over-rlls the defect. Fig. 17. Enveloped time waveform (defect inner race). The defect isindicated by the modulation ofthis signal. The expansion and contraction of the
‘eoks froma high amplitude (0,2 gf) then toward the center (0) indicate that energy i being generated as he rong element over-rals he defect.
Fig. 18. Enveloped spectrum with inner race defect ond bearing frequency overlays. This spectrum indicates a defects present onthe bearing’ inner Journal or plain bearings
Journal, plan and tited pad bearings have infinite fe, provided there is adequate lubrication to support the oil wedge on which the shat
rotates. The shaft performance is normally monitored by two orthagonally positioned eddy current probe transducers that measures the rel~
ative displacement between shaft an bearing block. Displacement warning alarms willindicate the need for possible shaft balancing proce-
dures. The bearing conditions best assessed by sump ol analysis, which confirms excessive metal wear.
‘SEE measurements have been applied to rend lubrication conditions of rotating element bearings, butit has not been reported as being
applied ta journal bearings. SEE transducers are baced on acoustic emissions concept, which provide indications of metal-to-metal contacts.
Acceleration enveloping is normally only applied where impulsive defect forces are measurable, such asin rotating element bearing,
Summary
Figs, 16 through 18 are advanced examples of data from an FFT Analyzer. Time waveform and spectrum analysis are dificult subjects to
explain thoroughly inan article that overviews the key features of spectrum analysis. There are extensive training courses on analyzing vibra~
tion data, Some key isues to consider when using vibration analysis asa method of determining machinery health ar:
+ Callect both spectrum and time waveform data ta complete a thorough data analysis.
' Develop skis around condition monitoring through training and application.
* Build a knowledge bankof machinery responses and problems. This helps you apply previously gained knowledge and minimizes repeat
mistakes
10. Gears
Gears are used to transmit power from one system to anather. tis important to understand how gears work and what symptoms ta lok for
when performing an analysis. Moreover, you should full understand the two key elements to consider
* Gear mesh frequency (GMF)
* Sidebands of GMF
{8 montring these two elements you can establish how the gear aes the sytem andthe sigicance of the problem
Gear mesh frequency
Gear mesh frequency equals the number ofteth onthe gear muted by the speed ofthe shaft to which the gears attached
= cM
(Hof tooth onthe gear] (speed of the shaft to which the gears attached)
Example:
(50 teeth) (1 180 RPM)
9000 CPM or 983,3 He
In audition to evaluating GMF, itis important to use the proper span (Fra) regarding frequency range to observe the GMF at higher frequen
cies nthe same vibration signature. To achieve this span, GMF shouldbe multiplied by a factor of 3.25,
Example: Using the above GMF:
25» GM
=(3.25)(59 000CPM)
191. 750. CPM or 3 195,8 Hz
Ifthe GMFis not known, use:
*# Fran # 200.shaft running speed
Fran = 2001 180 RPM
+ Fran = 236000 CPM or 3.933,3 Hz
SKF 2 The factor of 3,25 relates to a gear characteristic that wear problems do nat necessarily occur at fundamental gear mesh frequency (Lx GM),
which gear mesh is detected is 3x GMF.Thisis attributed to
the three motions of gear interaction: engaged sliding, rolling and disengaged siding, Hence, three pulses per revolution, The consideration of
this factor shouldbe evaluated when collecting gear mesh data
bout may accur at 2x or 3x GME. Infact, one af the most common frequencies
Gear mesh frequency sidebands
‘mesh frequency sidebands can be more significant than GME. The sidebands are spaced around the GMF relative to the RMP of each
‘mating gear. When the amplitude ofthe sidebands increases and the numberof sidebands present increases, theres likely a problem with
the gearbox components,
Additionally if one or both interfacing gears have worn teeth, the spectrum also exhibit sdebands around GMF. These sideband are
spaced at a distance equal othe shaft speed
Fig. 19. Spectrum with ger mesh frequency at 402 500 RPM, marked withthe overay. The shafts turning at 7 545 RPM with $3 teeth onthe gear;
therefore, 7595 +53 =402 500 RPM
24 SKF sarteas Fret
years Daag P1308
Jenne Giese Braet
oats uot
vuisus
wero
Fre
Feast
=
Fig, 20. Spectrum containing gear mesh frequency at 378.157 RPM, marked with sideband morkers. Sidebands are spaced ot 7513 RPM, whichis the
‘nominal speed ofthe shaft on which the gear riding.
11. Blades and vanes
Unlike some other types of machine condition vibration, flow-induced vibration can be very dependent on operating conditions. n other
words, depending upon the machines work, or the induced load, the machine can exhibit varying conditions, Flow-induced vibration condi-
tions are as follows
‘Hydraulic or aerodynamic forces
* Cavitation or starvation
+ Recirculation
= Turbulence
* Surging or choking
Pumps, bowers, turbines, etc, inherently produce hydraulic or aerodynamic forces as their impellers impart work ino the fd they are han:
fling. Under normal conditions, such forces are handled rather easy problem arises when these forces excite resonant frequencies and
‘cause prablems such as cavitation or excessively high vibration. The meat commonly generated signal related ta hydraulic ar aerodynamic
forces is Blade Pass Frequency (BPFl
‘© BPF = # of vanes « impeller RPM
Example:
‘+ BPF = (6)(3 600 RPM
+ BPF = 21 600CPM or 360 Hz
SKF 2 “These forces are generated by @ pressure variation or pulse each time a blade loads or unloads ast passes nearby stationary components.
From a vibration signal standpoint, itis common to look for BPF and harmonics of BPF. Ths occurs the impeller isnot properly aligned with
the difusers and centrally located within the housing, Another common blade frequency is Blade Rate Frequency (BRF}
+ BRF = (# impeller vanes) (# difusers)(RPM)/K
+ BRF = [(18 impeller vanes) (24 diffusers) (RPMI / 6
+ BRF=72«RPM
Where:
+ = Highest common factor ofimpeller vanes and number of diffuser vanes
“Thus, BRF (72 RPM)is 4x higher than BPF (18 » RPM) in this case. And, as was pointed out above, this machine wouldlikely suffer much
higher pulsation, as more than one set of impeller and diffuser vanes would ine up with ane another (in this case, six impeller vanes would
‘simultaneously be directly opposite diffuser vanes at angles of O°, 60°, 120°, 180°, 240° and 300°}, which results in pronounced pulsation at
BBRF.f there were ether 17 impeller vanes or 25 diffuser vanes, at no instance in time would more than one set of impellr and diffuser vanes
line up with one another. Therefore high vibration would be unlikely.
Cavitation isa common centrifugal pump problem and can be quite destructive to internal pump components Cavitation most commonly
‘occurs when a pump operating with excess capacity or low suction pressure. Since the pump is actually being starved, the fluids being
pulled apartasittres toil the cavity. This process causes pockets of vacuum that collapse or implode quickly, which creates impact that
‘excite natural frequencies ofthe impeller and nearby components, The most common characteristics of cavitation are:
+ Random, broadband energy between 20kand 120k CPM, which can cause excessive syster, wear
+ The sound of sand or gravel being purnped through system
+ Starvation isthe counterpart to cavitation and also involves insufficient airflow
Asi relates to a pump, recirculation s the opposite of cavitation. tcan occur when the pump is operating at a lower capacity than required or