Curso Peakvue Emerson CSI2130

8/14/2019 Curso Peakvue Emerson CSI2130

1/19

www.assetweb.com

White Paper August 2011

PeakVue Analysis for

Antifriction Bearing Fault Detection

Peak values (PeakVue) are observed over sequential discrete time intervals,captured, and analyzed. The analyses are the (a) peak values (measured ing's), (b) spectra computed from the peak value time waveform, and (c) theautocorrelation coefficient computed from the peak value time waveform.

Case studies of various classes of faults are presented to illustrate thePeakVue methodology. The classes of faults are (a) inner race defects, (b)outer race defects, (c) rolling element defects, and (d) cage related defects. Allthree analysis tools enable the identification of the defect and often theseverity of the defect.

2011 Emerson Process Management. All rights reserved. The contents of this publication are presented forinformational purposes only, and while effort has been made to ensure their accuracy, they are not to be construed aswarranties or guarantees, express or implied, regarding the products or services described herein or their use orapplicability. All sales are governed by our terms and conditions, which are available on request. We reserve the rightto modify or improve the designs or specifications of our products at any time without notice.
http://www.assetweb.com/http://www.assetweb.com/http://www.assetweb.com/


2/19

White Paper August 2011 Page 2

2011 Emerson Process Management. All rights reserved.

www.assetweb.com

Introduction The peak value analysis (PeakVue) methodologyintroduced by Emerson for the analysis of impact-likeevents is proven to be an effective tool for identifyingbearing defects.

Overview of Signal Processing for Vibration Analysis The analog signal from the vibration sensor is generallyrouted through some analog signal processing,converted into a digital format and then furtherprocessed digitally. The vibration sensor often is anaccelerometer whose output is expressed in g units. Thesignal processing may include conversion of the signalfrom acceleration to velocity units employing an analogintegrator. The analog signal (g or velocity units)generally is passed through a high order low pass filterimmediately before the analog-to-digital converter toremove any signal components which may be present atfrequencies greater than the Nyquist frequency definedas one half of the sampling rate.

This provides assurance that the digital representationof the analog signal is correct, i.e., the band limitedanalog signal existing prior to digital conversion could be

reconstructed from the digital signal.

Once a block of digital data is acquired at a constantsampling rate of desired length, typically a block size of2n where n is an integer, the digital data are furtherprocessed. By far the most common processing foranalyzing rotating equipment is the Fourier Transform,using a FFT algorithm to construct the spectrum either inacceleration or velocity units. The spectral analysis ishelpful in separating the band-limited signal into periodiccomponents related to the turning speed of the machine.

In addition to spectral analysis, auto-correlation analysiscan be applied to the digital block of data representingthe time waveform. These additional correlationanalyses have not proven to be helpful to the normalspectral analysis, but it can be beneficial for analysis oftime waveform acquired when employing PeakVueanalysis.


3/19



www.assetweb.com

In PeakVue analysis, no low pass filter at or slightlybelow the Nyquist frequency is employed. Instead, ahigh pass (or band pass) filter greater than or equal tothe nyquist frequency is employed. The digital block ofdata consists of absolute maximum values, which thetime waveform experiences over each time incrementdefined by the sampling rate. Hence the analysis of thisrepresentative time waveform is the analysis of peakvalues.

The analysis of this block of data consists of the peakvalues themselves and an identification of periodicitythat is best accomplished using spectral analysis. Theautocorrelation analysis has also been found to be verybeneficial for the peak value time waveform.

Case Studies

These case studies demonstrate the signatureaccompanying various antifriction bearing faults for awide variation in machine speed. The PeakVue timewaveform will also be emphasized relative to trendabilityand fault severity assessment. The primary emphasiswill be placed on the peak values in the PeakVue timewaveform, the spectral peaks and presence or absenceof sidebanding, and the periodic activity in theautocorrelation coefficient function.

The case studies presented are:Outer Race Defects

Pinion Stand GearboxCrowd Motor

Inner Race DefectsCrusher gearboxPrecision13 drill head spindle

Ball or Roller DefectChipperRougher gearbox

Cage and othersLubricationSingle stage rotary air compressor


4/19



www.assetweb.com

Outer Race Defects

The first case study is from a pinion stand gearbox ata steel producing facility. The normal spectral data takenfor a 1000 Hz bandwidth had g levels of 2 g's and noindication of a bearing problem. The input shaft speedwas about 360 RPM.

The PeakVue spectra and time waveform data arepresented in Fig. 1. A high pass filter of 1000 Hz wasused. Significant activity at the outer race defectfrequency with many harmonics is present. The impactlevels are as high as 37 g's.

Although many harmonics are present in the spectraldata, it is obvious from the PeakVue time waveform thatthe time between impacts corresponds to the outer racefault frequency (the harmonics are of no physicalsignificance). This is also demonstrated in theautocorrelation coefficient function presented in Fig. 2.

Figure 1: PeakVue Spectraand time waveform frominput shaft of pinion standgearbox


5/19



www.assetweb.com

The impacting levels trended from 18 g's in July to ahigh of 37 g's in September. The levels had decreasedto 14 g's in October prior to a bearing replacement.

The second case study is from a centrifugal servicewater pump rated at 12000 gpm driven by an 8 pole700 hp motor.

Normal route-based vibration monitoring identified anouter race defect of reasonably low level (the peak-to-peak g level was 1.5 g's). The fault was identified as"alert" and hence was to be trended. On December 8,1997, PeakVue was employed and showed both outer

Figure 2: Autocorrelationcoefficient from PeakVuetime waveform in Fig 1

Figure 3: Photograph ofdefective bearing from theinlet of pinion stand gearbox


6/19



www.assetweb.com

and inner race defects with a signal level of 10 g's. Asecond PeakVue reading was taken on December 22,1997 and is presented in Fig. 4. The impacting levelshad increased to 33 g's. The pump was then placed onthe "fault" level and the bearing was replaced. Theautocorrelation coefficient computed from the PeakVuetime waveform is presented in Fig 5. The autocorrelationcoefficient data shows the only activity of significance isthe impacting from the outer race defect.

Figure 4: PeakVue spectraand time waveform fromoutboard on service waterpump

Figure 5: Autocorrelationcoefficient from PeakVuewaveform in Figure 4.


7/19



www.assetweb.com

In this case, normal vibration data did identify the fault;however, the low levels observed did not place the faultat a level of significant concern. The impacting levelsidentified in PeakVue, excess of 30 g's, raised the

concern level and initiated planning for replacement.

Inner Race Defects

The first inner race defect case involves a crusher gearbox. The input shaft of this gear box has three bearings.The output of the gear box turns a crusher used at amining site. The gear box, approximately 8 ft. 10 ft 20 ft,is powered by an eight pole 2000 hp motor.

The trend observed on this bearing using PeakVue was:

February 13, 1997: Cage defect dominant but no otherbearing fault. Peak impacting 2.5 g's.

February 28, 1997: Inner race dominant with peakimpacting of 5.5 g's.

March 20, 1997: Cage and inner race defeat with peakimpacting of 6.6 g's.

March 27, 1997: Cage and inner race defeat with peakimpacting of 3.5 g's.

The PeakVue data acquired March 20, 1997 arepresented in Fig. 7. The cage activity at 6.23 Hz (0.42orders) is identified. The inner race defect (129.75 Hz) at8.72 orders is being sidebanded by running speed. Theinner race defect was not detected in the normal

Figure 6: Photograph ofdefective service water pumpbearing, showing inner and

outer race spalling.


8/19



www.assetweb.com

spectral analysis.

The autocorrelation coefficient function computed formthe PeakVue time waveform in Fig. 7 is presented in Fig.8. The inner race defect is clearly evident with amplitudemodulation at running speed. The peak g levels for aninner race are generally not as large as for thoseaccompanying an outer race defect. This is expectedsince the stress waves accompanying an inner racedefect will experience attenuation in reaching the outer

Figure 7. PeakVue spectraand time waveform from inletof crusher gear box.

Figure 8. Autocorrelationcoefficient from PeakVuetime waveform of Figure 7.


9/19



www.assetweb.com

peripherals (where the sensor is attached) of themachine.

The bearing was replaced in June, 1997. The inner racehad deep spalling over an approximate 1 1.5 area.

The second case study deals with a precision 13 drillhead spindle. Bearing defect, especially inner racedefects, detection on these precision multi-drill headspindles have proven to be difficult to impossible usingnormal velocity spectral analysis. PeakVue methodologywas applied to several of these multi-drill head spindlesand demonstrated to be an effective tool for bearingdefect detection. On the class of spindle monitored, thehigh-frequency accelerometer was magneticallyattached to the base of the spindle and data acquired inthe unloaded position. An alert level was set at 2.0 g'son the PeakVue time waveform. When this level wasexceeded, an alert is set signifying a problem is present.PeakVue spectral analysis was then employed toidentify the problem.

Figure 9. PeakVue spectraand time waveform from aspindle on a 13 multi-drillhead spindle.


10/19



www.assetweb.com

The PeakVue spectra and time waveform for a spindlewith a confirmed inner race defect are presented in Fig.9. The inner race defect at 797.86 (10.2 orders) Hz isidentified with side banding (twice running speed isdominant). The autocorrelation coefficient functioncomputed form the PeakVue time waveform ispresented in Fig. 10. The dominant activity here is at theinner race defect frequency. The bearing was replaced.The bearing had deep spalling on the inner race

Ball or Roller DefectFollowing a shutdown where work was done on thischipper machine, the acceleration time waveform for a500 Hz bandwidth spectra indicated some impactingmay be occurring, see Fig. 11. The p-p g levels wereless than 1 g which are not judged to be significant. Thevelocity spectra had a low 1x component but there wereno obvious bearing faults.

Figure 10. Autocorrelationcoefficient from PeakVuetime waveform acquired onthe spindle of Figure 9.


11/19



www.assetweb.com

Figure 11. Velocity spectraand acceleration time waveform from inboard axial on achipper.

Figure 12. PeakVue Spectraand time waveform from

inboard axial on chipper.


12/19



www.assetweb.com

A PeakVue data set was acquired at about the sametime using a high pass filter of 2000 Hz. The resultingPeakVue spectra and time waveform are presented inFig 12. The impacting level is near 20 g's and the cageand twice ball spin frequencies are dominant. Thispattern is indicative of two impacts per revolution of theball that is passing through the load zone at the cagerepetition rate. From the PeakVue spectral data, it s notobvious whether the 38, 40.2, or 42.4 Hz peaks arerepresentative of twice ball spin. The autocorrelationcoefficient function data, presented in Fig. 13 clearlyidentifies the dominant activity is at 40.2 Hz which istwice the ball spin frequency. The autocorrelationcoefficient data also clearly shows the defective ballgoing in and out of the load zone at the cage turningspeed.

The root cause of this problem was found to be impropergrounding of an electrical welder used during the recentoutage.

When data was acquired on the inlet pinion shaftbearing on January 2, 1997, impacts in the 6 g rangewere detected with the PeakVue spectra showing ballspin fault modulated by cage. On February 9, 1997, theimpacting level had increased to 9 g's. The PeakVuedata are presented in Fig 14. The ball spin faultfrequency is at 40.4 Hz and cage at 3.43 Hz. Theautocorrelation coefficient is presented in Fig. 15 which



13/19



www.assetweb.com

emphasizes the cage and ball activity. A photo of thedefective bearing is presented in Fig. 16.

The bearing was replaced and PeakVue data acquiredon February 12, 1997. The impacting level haddecreased to less than 1 g. The normal vibration nevershowed any indication of a bearing problem.

Figure 14. PeakVueSspectra and time waveformfrom inboard pinion roughergear box.



14/19



www.assetweb.com

Cage and Other

When the cage frequency is present in the PeakVuespectral data, it may be indicative of problems other thancage. As seen in the previous section, cage frequencygenerally is present when a rolling element has a defect.This is postulated to be the case since the defectiverolling element will pass through the load zone once perrotation of the cage. Cage frequency may also bepresent with other problems, e.g., lubrication relatedproblems, heavily preloaded bearing, significant axial orthrust forces present, out of round inner race, et al. Allthese sources of cage activity have been observed morethan once, but none have been observed sufficiently tostate the presence of the specific problem will alwaysmanifest itself with cage activity in the PeakVue data. Asexamples, a lubrication problem as well as a suspectedexcessive thrust problem are presented below.

The PeakVue data from the inboard bearing on a motor(150 hp) directly driving a centrifugal pump arepresented in Fig. 17.

Figure 16. Photograph ofdefective bearing highlightingthe rollers.


15/19



www.assetweb.com

The impacting levels are 5 g's which are not excessivefor this speed machine but definitely indicative of aproblem. The spectral data is dominated with cage andmany harmonics. This is a NSK NU 318 bearing whichhas 13 rolling elements.

The autocorrelation coefficient data computed from thePeakVue time waveform are presented in Fig. 18. Theonly significant activity is the cage activity.

A photo of the defective bearing is presented in Fig. 19.

The rolling elements show indication of "skidding," butthe dominant problem is the "caked" residue of lubricant(grease).

This bearing is being replaced with a rolling ball elementbearing. The problem of grease being "pushed out" isnot uncommon with cylindrical rollers in low radial loadapplications.

Figure 17. PeakVue spectraand time waveform fromMotor (Inboard Axial) drivingdirect coupled centrifugal pump.


16/19



www.assetweb.com

The compressor from which data was acquired was arecent replacement of a unit that had experiencedcatastrophic failure (rotary had penetrated the housing).PeakVue and normal vibration data was acquired on themotor and on both ends of the two rotary screwelements. The normal data was typical of a smoothrunning rotary screw compressor.

The sensor (accelerometer) used was a 10 mV/g sensorattached to the surface via a rare earth flat magnet. Thesurface was not machined and the paint was notremoved. The lack of surface preparation will attenuatethe higher frequency energy getting to the sensor. Forroutine monitoring, the surface should be prepared atthe places where the accelerometer should be mounted.


Figure 19. Photography ofthe defective bearinghighlighting the "caked"residue of lubricant and skidof roller 5.


17/19



www.assetweb.com

The PeakVue data taken with the sensor mountedvertically over the outbound bearing on the male 4 vanerotary screw are presented in Fig. 20. Given how thesensor was mounted, the 30 g peak impacts seen in thePeakVue time waveform are considered significant. Theactivity are primarily many harmonics of cage frequency.The autocorrelation coefficient data presented in Fig. 21also identifies cage as the dominant source of theactivity.

This type activity has been observed before in caseswhere a) lubrication problems are present, b) whereheavy preloads are present, c) where out-of-round raceways are present, and d) where heavy thrust loads arepresent.

Figure 20. PeakVue Spectraand Time Waveform on theoutboard bearing of the malerotor and simple stagecompressor.


18/19



www.assetweb.com

Given the "newness" of this compressor and lack ofsimilar activity on the female rotary screw, the lubricationproblem is placed low on the suspected source. Thisleads to the conclusion that there could be excessivethrust for the bearing or out-of-round race way. In anyevent, cage activity generally is an indicator of adeveloping problem and should be trended closely.

Conclusion The PeakVue methodology has proven to be a veryuseful tool for bearing defect detection in applicationswhere normal spectral analysis has proven to beineffective e.g., large gear boxes, slow speedmachinery, etc. The PeakVue time waveform providestrendable information that can be classified as to thetype of fault and speed of machinery. For example,detectable impact levels form an inter race defect will bea factor of two or so less than the detectable impactslevels form an outer race defect of the same severitylevel and machine speed. The impacts levels willdecrease with speed of the machine at the rate of thesquare root to linear reduction of speed. As a rule ofthumb, the ratio of speed to 0.75 power providesadequate adjustment for speed changes.

Impacting levels form defective inner race will generallybe modulated with shaft turning speed. In a similarmatter, rolling element defects will be modulated at thecage frequency.



19/19

Date post:	04-Jun-2018
Category:	Documents
Upload:	luisrosav
View:	223 times
Download:	4 times

Curso Peakvue Emerson CSI2130

Documents