Você está na página 1de 14

M.

Elforjani
e-mail: elforjani@gmail.com

D. Mba
e-mail: d.mba@craneld.ac.uk School of Engineering, Craneld University, Craneld, Bedfordshire MK43 0AL, UK

Monitoring the Onset and Propagation of Natural Degradation Process in a Slow Speed Rolling Element Bearing With Acoustic Emission
The monitoring and diagnosis of rolling element bearings with the high frequency acoustic emission (AE) technology has been ongoing since the late 1960s. This paper demonstrates the use of AE measurements to detect, locate, and monitor natural defect initiation and propagation in a conventional rolling element bearing. To facilitate the investigation a special purpose test rig was built to allow for accelerated natural degradation of a bearing race. It is concluded that subsurface initiation and subsequent crack propagation can be detected with the AE technology. The paper also presents comparative results between AE and vibration diagnosis. DOI: 10.1115/1.2948413

Introduction

Slow speed rotating machines are the mainstay of several industrial applications worldwide. They can be found in paper and steel mills, water industry, wind turbines, etc. The operational experience of such machinery has not only revealed challenging design issues but has also presented opportunities for further signicant improvements in the technology and economics of such machines. Failures associated with bearings represent the cause of extended outages and are typically caused by gradual deterioration and wear 1 . Such slow degradation processes can be identied if a robust online monitoring and predictive maintenance technology is used to detect impending problems with obvious economic advantages. Slow speed rotating machinery generates relatively reduced energy loss rates from damage related processes, and therefore conventional condition monitoring technologies e.g., vibration analysis tend to be more difcult to apply. Jamaludin et al. 2 summarized the limitations in applying vibration to slow rotating machines. However, this is not the case for the acoustic emission AE technology, which is well suited to detecting very small energy release rates. As a result AE is able to detect subtle defect related activity from machinery 3,4 . To date most published works on the application of the AE to monitoring bearing mechanical integrity have been on articially or seeded damage, which is generally induced with an electrical discharge system and engraving machine or by introducing debris into the lubricant 4 . This paper presents the experimental results of an evaluation of AE technology in detecting and diagnosing the onset of subsurface cracks and their propagation to spalls. Comparisons between simple AE and vibration parameters are presented for all tests. AE can be dened as the class of phenomena whereby transient elastic waves are generated by the rapid release of energy from localized sources within a material. A tremendous amount of work has been undertaken over the last 20 years in developing the application of the AE technology for bearing health monitoring 4 . Jamaludin et al. 2 conducted an investigation into the applicability of stress wave analysis for detecting early stages of bearing
Contributed by the Technical Committee on Vibration and Sound of ASME for publication in the JOURNAL OF VIBRATION AND ACOUSTICS. Manuscript received October 23, 2007; nal manuscript received April 30, 2008; published online July 15, 2008. Review conducted by Michael Brennan.

damage at a rotational speed of 1.12 rpm 0.0187 Hz . Attempts had been made to generate a natural defect on the bearing components by fatiguing. However, after allowing the test bearing to operate for a period of 800 h under conditions of grease starvation, no defect and/or wear was visually detectable on any of the bearing components. In a further study, Morhain and Mba 5 examined the application of standard AE characteristic parameters on a radially loaded bearing. The use of typical AE parameters such as root mean square rms and count values was validated as a robust technique for detecting bearing damage and was shown to correlate with increasing speed, load, and defect size. Al-Ghamdi and Mba 6 conducted a comparative experimental study on the use of AE and vibration analysis for bearing defect identication and estimation of defect size. It was concluded that AE offered earlier fault detection and improved identication capabilities than vibration analysis. Furthermore, the AE technology also provided an indication of the defect size, allowing the user to monitor the rate of degradation on the bearing, unachievable with vibration analysis. Miettinen and Pataniitty 7 described the use of the AE in the monitoring of faults in an extremely slow rotating rolling bearing 5 rpm . Prior to testing the test bearing had been naturally damaged on its outer race. It was concluded that the AE measurement was a very sensitive method for fault detection in an extremely slowly rotating bearing. Choudhury and Tandon 8 applied the spark erosion method for seeding defects in bearings. AE measurements from bearings without defect and with defects of different sizes were undertaken. It was shown that the use of AE parameters such as ring-down counts and peak amplitudes could identify bearing defects. Price et al. 9 employed a four-ball lubricant test machine to simulate pitting fatigue and scufng wear commonly experienced by gear and bearing components. The principal monitoring technique utilized in this investigation was AE. The study concluded that scufng wear and pitting was detectable with AE. To date the only investigation on the identication of the onset of natural degradation in bearings involves the work presented by Yoshioka 10 . This focused on the detection of a rolling contact subsurface fatigue crack using AE technology. An AE source locating system was developed, and it was reported that the system was able to locate the AE source based on an analysis of the time delay associated with AE events acquired simultaneously from different sensors. Yoshioka stated that cracks AUGUST 2008, Vol. 130 / 041013-1

Journal of Vibration and Acoustics

Copyright 2008 by ASME

Downloaded 22 Jun 2011 to 117.206.44.33. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Table 1 Bearing life calculations for Case I Input data Load applied N Number of balls Ball radius mm Race groove radius mm Mean bearing diameter mm Rotational speed rpm Auxiliary quantities Load on single ball N Curvature sum mm Curvature difference mm Dimensionless semimajor axis of contact ellipse a* Dimensionless semiminor axis of contact ellipse b* Dimensionless contact deformation d* Flat race 3571 0.35 0.00 1.00 1.00 1.00 Groove race 3571 0.18 0.98 6.31 0.32 0.44 50000 14 5.75 5.8 64 72

Fig. 1 Test bearing

were identied parallel to the surface maximum length of approximately 200 m in the rolling direction of ball and were distributed between 50 m and 200 m below the surface. It must be noted that the tests undertaken by Yoshioka were on a bearing with only three rolling elements, which is not representative of a typical operational bearing. Furthermore, tests were terminated once AE activity increased such that the propagation of identied subsurface defects to surface defects was not monitored. This work builds further on the work of Yoshioka by monitoring not only the initiation of cracks, but also its propagation to spalls or surface defects on a conventional bearing with the complete set of rolling elements. Furthermore, the location of the AE source was also monitored throughout the test sequence in order to validate that the AEs generated throughout the test period can be eventually attributed to the surface defect noted at the end of the test; this study is the rst of its kind to date.

Surface stress and deformation Semimajor axis of the ellipse contact mm Semiminor axis of the ellipse contact mm Deformation mm Maximum pressure stress N / mm2 Permanent deformation mm 0.0001D mm Permanent deformation/0.0001D % Plastic/elastic deformation % Flat race 0.513 0.513 0.046 6484 0.0044 0.0012 379 9.5 Groove race 4.063 0.206 0.016 2042 0 0.0012 3.4 0.24

Subsurface stresses Thomas and Hoersh theory Flat race 2215 Maximum shear stress N / mm2 Shear-pressure stress ratio % 34.16 Depth below the surface mm 0.241 Depth/semiminor axis of ellipse 47.00 contact % Lundeberg and Palmgren theory Flat race 2774 Maximum amplitude orthogonal shear stress N / mm2 Shear-pressure stress ratio % 42.78 Depth below the surface mm 0.180 Depth/semiminor axis of ellipse 35.09 contact % von Mises distortion energy theory Flat race 3696 Maximum octahedral shear stress N / mm2 Shearpressure stress ratio % 57 Depth below the surface mm 0.385 Depth/semiminor axis of ellipse 75 contact % Race life prediction Basic dynamic load rating N L10 day Flat race 23,556 1 Groove race 61,874 18 Groove race 628 30.76 0.158 77.00

Test-Rig Design and Layout

A specially designed test rig that encouraged the natural damage condition of a test bearing was employed. To speed up crack initiation, a combination of a thrust ball bearing and a thrust roller bearing was selected. One race of ball bearing SKF 51210 was replaced with a at race taken from the roller bearing SKF 81210 TN of the same size, as shown in Fig. 1. A consequence of this arrangement is that the rolling elements on the at track caused higher contact pressure relative to the grooved race due to the reduced contact area between the ball elements and the at race. For the purpose of this experiment the following procedure was undertaken to determine the subsurface stresses on the test bearing and thereby estimate the time, or number of cycles, toward surface fatigue on a track. Theories employed for this procedure, particularly for the at race, included the Hertzian theory for determining surface stresses and deformations, the Thomas and Hoersh theory for subsurface stress, and the Lundberg and Palmgren theory for fatigue evaluation. For the grooved race the standard procedure, as described by BS 5512, 1991, was employed for determining dynamic load rating. Finally the anticipated life for dened stresses was computed for both the grooved and at races see Tables 1 and 2 . Results clearly illustrated that surface fatigue, such as aking, could be initiated on the at race within a few days depending on the load condition, thereby authenticating the test-rig design. It should be noted that the theoretical estimation of rolling contact fatigue is known to be subject to variability or scatter when compared to experimental results, and this has been attributed to the probability of inclusions in the steel material located in the highest load zones of the race 11 . 041013-2 / Vol. 130, AUGUST 2008

Groove race 1018 49.88 0.102 49.63

Groove race 1164 57 0.154 75

A specically designed test rig, as shown in Fig. 2, was employed for this investigation. It consisted of a hydraulic loading device, a geared electrical motor Motovario-Type HA52 B3B6-B7 j20,46-lubricated: AGIP , a coupling, and a supporting Transactions of the ASME

Downloaded 22 Jun 2011 to 117.206.44.33. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Table 2 Bearing life calculations for Case II Input data Load applied N Number of balls Ball radius mm Race groove radius mm Mean bearing diameter mm Rotational speed rpm Auxiliary quantities Load on single ball N Curvature sum mm Curvature difference mm Dimensionless semimajor axis of contact ellipse a* Dimensionless semiminor axis of contact ellipse b* Dimensionless contact deformation d* Flat race 2500 0.35 0.00 1.00 1.00 1.00 Groove race 2500 0.18 0.98 6.31 0.32 0.44 35000 14 5.75 5.8 64 72

Fig. 2 Test-rig layout

Surface stress and deformation Semimajor axis of the ellipse contact mm Semiminor axis of the ellipse contact mm Deformation mm Maximum pressure stress N / mm2 Permanent deformation mm 0.0001D mm Permanent deformation/0.0001D % Plastic/elastic deformation % Flat race 0.455 0.455 0.036 5757 0.0021 0.0012 186 6 Groove race 3.608 0.183 0.013 1813 0.0000 0.0012 2 0.1

allow for the placement of AE sensors and thermocouples directly onto the race see Fig. 4 . The thrust shaft was driven by a hydraulic cylinder Hi-Force Hydraulics Model No. HP110-HAND pump-single speed-working pressure: 700 bars , which moved forward to load the bearing and backward to allow periodical inspections of the test bearing face. The rotating disk was driven by a shaft attached to a geared motor with an output speed of 72 rpm. A thrust bearing SKF 81214 TN was placed between the coupling and the test bearing to react the axial load. A exible coupling was employed between the shaft and the geared motor.

Instrumentation

Subsurface stresses Thomas and Hoersh theory Flat race 1967 Maximum shear stress N / mm2 Shear-pressure stress ratio % 34.16 Depth below the surface mm 0.214 Depth/semiminor axis of ellipse 47 contact % Lundeberg and Palmgren theory Flat race 2463 Maximum amplitude orthogonal shear stress N / mm2 Shear-pressure stress ratio % 42.78 Depth below the surface mm 0.160 Depth/semiminor axis of ellipse 35.09 contact % von Mises distortion energy theory Flat race 3282 Maximum octahedral shear stress N / mm2 Shearpressure stress ratio % 57 Depth below the surface mm 0.342 Depth/semiminor axis of ellipse 75 contact % Race life prediction Basic dynamic load rating N L10 day Flat race 23556 3 Groove race 61,874 53 Groove race 558 30.76 0.141 77

Groove race 904 49.88 0.091 49.63

A schematic of the data acquisition process is detailed in Fig. 3. The AE acquisition system employed commercially available piezoelectric sensors Physical Acoustic Corporation type PICO with an operating range of 200 750 kHz at temperatures ranging from 65 C to 177 C. Four acoustic sensors, together with two thermocouples RoHS type: J 1 M 455-4371 , were attached to the back of the at raceway using superglue. One accelerometer Endevco-236-M-ISOEASE-PF44 , attached to the housing of the at race, was used to measure the vibration in the axial direction. The acoustic sensors were connected to a data acquisition system through a preamplier, set at 40 dB gain see Figs. 3 and 4 . The system was continuously set to acquire AE wave forms at 2 MHz sampling rate, while AE parameters such as counts, rms, average signal level ASL in decibels, maximum amplitude, and absolute energy joules were recorded over a time constant of 10 ms and a sampling rate of 100 Hz.

Groove race 1033 57 0.137 75

Experimental Results Observations and Discussions

structure. The test bearing was positioned between the stationary thrust loading shaft and the rotating disk, which housed the grooved race. The at race was tted onto the loading shaft in a specically designed housing. This housing was constructed to Journal of Vibration and Acoustics

4.1 Acquisition System Calibration. Prior to testing, calibration tests were undertaken to understand the attenuation properties of the test bearing. Attenuation can be described as any reduction or loss in the AE signal strength in the form of amplitude or intensity , and it is expressed in decibels 12 . In AE applications, attenuation is a very important property because it determines the signal strength as a function of distance; therefore, it plays a signicant role in specifying locations of AE sensors for purposes of identifying sources of AE events. Bearing attenuation test was carried out prior to laboratory tests. HsuNielsen sources were used for attenuation tests. This test consists of breaking a 0.5 mm diameter pencil lead approximately 3 mm 0.5 mm from its tip by pressing it against the surface of the piece. Two different approaches to ascertaining attenuation were undertaken. The rst involved breaking a 0.5 mm diameter lead pencil of hardness 2H onto the at raceway directly adjacent to AE sensors labeled channels 1, 2, 3, and 4 see Fig. 5 . AUGUST 2008, Vol. 130 / 041013-3

Downloaded 22 Jun 2011 to 117.206.44.33. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Fig. 3 Schematic of the data acquisition systems

A detection threshold was set at 52 dB for the acquisition of AEs generated from the lead breaks, and an average value of maximum signal amplitude of ten pencil breaks from each position was calculated. Signal amplitude and relative attenuation were calculated using A dB = 20 log10 Vmeasured Vreference 1

An analysis revealed that the AE signals on the at ring are attenuated by increasing the distance from the emanating AE source as expected. For instance, lead breaks at channel 1 showed greatest attenuation at channel 3 3 dB , which is the maximum distant from the source channel 1 . A summary of results is presented in Fig. 6. The second approach to understanding the attenuation of the

Fig. 4 Test bearing with attached sensors

041013-4 / Vol. 130, AUGUST 2008

Transactions of the ASME

Downloaded 22 Jun 2011 to 117.206.44.33. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Fig. 5 Breaking lead pencil at four different positions

bearing ring involved breaking lead at another three positions labeled as midpoints between channels 2 and 3, and channels 3 and 4. In-plane tests were also undertaken midpoint between channels 2 and 3. The experimental settings were kept the same as the previous attenuation test. For these tests, ten lead breaks at each position were again performed and the average maximum signal amplitudes and attenuation rate, as described in the rst part, are presented in Fig. 7. During the test, where the source location was at the midpoint between channels 2 and 3, channel 2 recorded the maximum signal strength of 0.209 V, while highest attenuations of 3.13 dB and 3.44 dB were observed at channels 1 and 4, respectively see Fig. 7 . It should be noted that the position 3 in Fig. 7 is not channel 3 but the midpoint between channels 2 and 3 see the gure legend .

4.2 AE Source Location. The capability of AE to determine source locations of signals emanating in real time from materials under load is one of the signicant advantages over other nondestructive test NDT technologies. In AE applications, AE signals traveling through the medium are attenuated and arrive at different sensors with certain time delay. This delay can be attributed to the distance between the source defect and AE sensors, and with knowledge of the signal velocity the location of the AE source can be identied. For this particular investigation efforts were made to identify the defect location AE source location in real time. This was accomplished by identifying the wave velocity on the ring experimentally. At a threshold of 52 dB and with known distances between the AE sensors, the velocity of the AE wave form under such conditions was calculated at 4000 m / s. This velocity was

Fig. 6 Relative attenuation at four different positions

Journal of Vibration and Acoustics

AUGUST 2008, Vol. 130 / 041013-5

Downloaded 22 Jun 2011 to 117.206.44.33. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Fig. 7 Relative attenuation at three different positions

used for all future source locations, and prior to the onset of testing several lead breaks were made at various positions on the surface to establish the accuracy at this velocity and specic threshold level. Results were within 4% of the exact geometric location of the lead break. Figure 8 shows the source location layout used, which essentially unwrapped the bearing race for a linear location 4.3 Bearing Tests. Under normal conditions of load, rotational speed, and good alignment, surface damage begins with subsurface initiation, which gradually propagates to the surface, creating pits and spalls. During testing, AE parameters were recorded in two modes; the rst was a continuous recording of AE absolute energy and rms acquired at a sampling rate of 100 Hz and over a time constant of

10 ms. The absolute energy is a measure of the true energy and is derived from the integral of the V2 signal divided by the reference resistance 10 k over the duration of the AE signal. In addition, traditional AE parameters such as counts, amplitude, and ASL were also measured. The ASL is a measure of the continuously varying and averaged value of the amplitude of the AE signal in decibels. The ASL is calculated from the rms measurement and is given as ASL dB = 20 log10 1.4 rms in mV/100 2

The traditional parameters were calculated over an AE event duration of 1500 s and a threshold of 52 dB; the threshold of 52 dB was selected based on numerous rig commissioning tests; at this level a signicant amount of background noise was re-

Fig. 8 Source location layout for linear detection

041013-6 / Vol. 130, AUGUST 2008

Transactions of the ASME

Downloaded 22 Jun 2011 to 117.206.44.33. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Fig. 9 Test conditions run until visually observable surface damage, Case I

jected. With almost all tests it was noted that during the rst 2 h of each test AE activity was present, and this was attributed to a running-in period as after this period 2 h all measured AE and vibration parameters remained constant. For this particular paper two experimental cases are presented that reect the general observations associated with over 18 experimental tests at loads ranging from 20 kN, 35 kN, and 50 kN. Case I is for a load condition of 50 kN, while Case II presents results for a test load of 35 kN. Case I. Observations of continuous monitoring of the AE levels, in addition to traditional AE parameters, for 16 h of bearing operation are presented in Figs. 9 and 10. At the end of the test 16 h there was visible surface damage. It was observed that at approximately 9 h into operation AE emission levels began to

increase steadily. This was not observed on the vibration measurements though vibration levels increased after 13.5 h of operation; much later that was detected by AE reinforcing the widely acknowledged view that AE is more sensitive than vibration for bearing defect identication 4 . The increase in AE energy levels from earlier in the test run between 2 h and 6 h to the condition of surface damage was in the order of 10,000%. Figure 10 shows trends of traditional AE parameters all of which show a signicant increase in AE activity from 9 h of operation. Also worth noting is a small increase in AE levels counts and amplitude at 4 h of operation. Interestingly observations of the AE wave form, sampled at 2 MHz, showed changing characteristics as a function of time. This is presented in Fig. 11 where a typical AE wave form asso-

Fig. 10 Classical AE parameters associated with Case I

Journal of Vibration and Acoustics

AUGUST 2008, Vol. 130 / 041013-7

Downloaded 22 Jun 2011 to 117.206.44.33. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Fig. 11 AE wave form associated with Case I

ciated with spurious AE transient events is presented after 4 h of operation. The wave form at 10 h operations shows a periodicity of AE transient bursts at approximately 18 Hz , while at 16 h operation signicant AE transient events associated with the defect frequency 9 Hz of the bearing are clearly noted see Fig. 11 . It is particularly interesting to note that the frequency of the transient event reduced from 18 Hz to 9 Hz from 10 h operation to the end of the test at 16 h . As surface defects, such as spalls, are continually developing, it is postulated that a newly formed spall will contribute relatively higher AE events as the edges of this newly formed defect will be rougher in comparison to an already existing spall, which becomes smoothened with the passage of time. As such at 16 h operation one of the spalls developed was relatively less mature than others and resulted in high AE levels at the defect frequency. Hence the strong evidence of 9 Hz indicating one defect on the race. This also explains the sharp bursts of AE activity noted during observations of continuously monitored AE energy levels see Fig. 9 . Even though the overall energy levels increase from 9 h operation, relatively large transient rises were noted during the period from 10 h to 16 h. It is postulated that these large transient bursts are attributed to regions that have newly developed surface damage; this is an evolutionary process giving rise to peaks and troughs in AE levels. On termination of the test 16 h a visual inspection revealed surface damage at three locations on the race see Fig. 12 . Thus far the observations have shown AE to monitor the degradation of an accelerated test; the next phase of analysis involved source identication of AE activity throughout the test duration. Figures 1317 highlight the trends in source location throughout the test period; the regions where the surface damage occurred have been highlighted. The location plots show cumulative energy over the test simulation. It is worth noting that only AE events above a threshold of 52 dB contribute to the source location. Whenever the threshold is exceeded the location of the source is computed and identied. The AE energy is assigned to the geometric position source ; this is a cumulative process, and as such a xed source will have the largest contributory energy in a cu041013-8 / Vol. 130, AUGUST 2008

mulative plot. Evident from these gures was that at the start of the tests Fig. 13 , a concentration of AE source from outside zone 3 195 mm was noted; this is attributed to the running-in AE activity. At about 4 h into operation, relatively early signs of concentrated AE activity from two of the highlighted zones began to appear see Fig. 14 . After 10 12 h operation the concentration of the AE source was clearly located at the three highlighted regions see Figs. 15 and 16 . At this stage the running-in AE related source operational noise , as seen in Fig. 13, was relatively insignicant. Lastly at 16 h operation the location of the AE sources was limited to the three regions where the actual surface damage has occurred see Fig. 17 . These results suggest that the onset of crack development could have been identied as early as 4 h into the operation of the test bearing. Case II. This case presents different trends to that noted earlier in Case I; the applied load on this test bearing was 35 kN. Obser-

Fig. 12 Crack zones on at ring associated with Case I

Transactions of the ASME

Downloaded 22 Jun 2011 to 117.206.44.33. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Fig. 13 Test running-in stage associated with Case I 1 h operation

Fig. 14 Crack onset stage associated with Case I 4 h operation

Fig. 15 Crack propagation stage associated with Case I 10 h operation

Journal of Vibration and Acoustics

AUGUST 2008, Vol. 130 / 041013-9

Downloaded 22 Jun 2011 to 117.206.44.33. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Fig. 16 Crack propagation stage associated with Case I 12 h operation

Fig. 17 Surface damage locations associated with Case I 16 h operation

Fig. 18 Test conditions run until visually observable surface damage, Case II

041013-10 / Vol. 130, AUGUST 2008

Transactions of the ASME

Downloaded 22 Jun 2011 to 117.206.44.33. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Fig. 19 Classical AE parameters associated with Case II

Fig. 20 AE wave form associated with Case II

Journal of Vibration and Acoustics

AUGUST 2008, Vol. 130 / 041013-11

Downloaded 22 Jun 2011 to 117.206.44.33. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Fig. 21 Crack zone on at ring associated with Case II

vations of continuous monitoring of the AE levels, in addition to traditional AE parameters, for 18 h of bearing operation are presented in Figs. 18 and 19. At 4 h of operation relatively high levels of AE activity were noted, particularly the activity associated with AE counts. Also noted on the AE wave form at 4 h operation was the high transient nature of the wave form. At 6 h operation the level of AE is reduced to that prior to the increased AE activity at 4 h see Fig. 18 . The exact reason for this is not known, but the authors postulate that the generation of early subsurface or surface damage resulted in AE activity, which following maturity of the defects worn smooth particularly around the edges overtime reduced the AE activity, as explained earlier. This phenomenon had been noted by Al-Dossary et al. 13 where protrusion at the edge of the articially seeded defects generated high transient AE bursts during the entry and exit only of the roller over the defect. Again, as in the previous tests the wave form at the end of the test highlighted a periodicity of AE transient events associated

Fig. 22 Test running-in stage associated with Case II 1 h operation

Fig. 23 Crack onset stage associated with Case II 4 h operation

041013-12 / Vol. 130, AUGUST 2008

Transactions of the ASME

Downloaded 22 Jun 2011 to 117.206.44.33. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Fig. 24 Crack propagation stage associated with Case II 10 h operation

with the defect frequency of the bearing 9 Hz see Fig. 20 . Interestingly no periodic AE transient event was noted at 10 h of operation; AE transient events remained random. At 18 h of operation the test was terminated and there was now visible evidence of surface damage see Fig. 21 . Source location deployment used in bearing tests provided another simple and rapid means to identify and locate the crack initiation and propagation. As in the previous case the source location over the duration of the tests is presented in Figs. 2226. Again, the start of the tests shown geometric concentrations of AE activity that is attributed to the running in condition see Fig. 22 . The other gures show the growing concentration of AE energy from the defect location over time, again suggesting that the onset of cracking could be ascertained as early as 4 h into operation. It is worth noting that the actual test period leading to visual damage on the race was much faster than the theoretical calculations. This variation was random but always earlier than predicted.

This is attributed to issues such as misalignment and unbalance, which are not incorporated in theoretical estimates; however, best efforts were undertaken to minimize this.

Conclusion

Bearing run-to-failure tests under natural damage conditions were successfully performed. These tests demonstrated the applicability of AE in detecting and locating crack initiation and propagation on bearing races whilst in operation. The two cases presented are representative of other tests performed in this study and show that there is a clear correlation between increasing AE energy levels and the natural propagation and formation of bearing defects. The study demonstrated that AE parameters such as rms and energy are more reliable, robust, and sensitive to the detection of incipient cracks and surface spalls in slow speed bearing than

Fig. 25 Crack propagation stage associated with Case II 14 h operation

Journal of Vibration and Acoustics

AUGUST 2008, Vol. 130 / 041013-13

Downloaded 22 Jun 2011 to 117.206.44.33. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Fig. 26 Surface damage location associated with Case II 18 h operation

vibration analysis. At the rotational speed on which these tests were employed, this is the rst known attempt at correlating AE and natural defect generation.

References
1 Tandon, N., and Choudhury, A., 1999, A Review of Vibration and Acoustic Measurement Methods for the Detection of Defects in Rolling Element Bearings, Tribol. Int., 32 8 , pp. 469480. 2 Jamaludin, N., Mba, D., and Bannister, R. H., 2001, Condition Monitoring of Slow-Speed Rolling Element Bearings Using Stress Waves, Proc. Inst. Mech. Eng., Part E: J. Process Mech. Eng., 215 4 , pp. 245271. 3 Holroyd, T. J., 2001, Condition Monitoring of Very Slowly Rotating Machinery Using AE Techniques, 14th International Congress on Condition Monitoring and Diagnostic Engineering Management, COMADEM 2001 , Manchester, UK, Sept. 46. 4 Mba, D., and Rao, R. B. K. N., 2006, Development of Acoustic Emission Technology for Condition Monitoring and Diagnosis of Rotating Machines: Bearings, Pumps, Gearboxes, Engines, and Rotating Structures, Shock Vib. Dig., 38, pp. 316. 5 Morhain, A., and Mba, D., 2003, Bearing Defect Diagnosis and Acoustic Emission, Proc. Inst. Mech. Eng., Part J: J. Eng. Tribol., 217 4 , pp. 257 272.

6 Al-Ghamdi, A. M., and Mba, D., 2005, A Comparative Experimental Study on the Use of Acoustic Emission and Vibration Analysis for Bearing Defect Identication and Estimation of Defect Size, Elsevier Science, New York. 7 Miettinen, J., and Pataniitty, P., 1999, Acoustic Emission in Monitoring Extremely Slowly Rotating Rolling Bearing, Proceedings of 12th International Congress on Condition Monitoring and Diagnostic Engineering Management, COMADEM 99, England. 8 Choudhury, A., and Tandon, N., 2000, Application of Acoustic Emission Technique for the Detection of Defects in Rolling Element Bearings, Tribol. Int., 33 1 , pp. 3945. 9 Price, E. D., Lees, A. W., and Friswell, M. I., 2005, Detection of Severe Sliding and Pitting Fatigue Wear Regimes Through the Use of Broadband Acoustic Emission, Proc. Inst. Mech. Eng., Part J: J. Eng. Tribol., 219 2 , pp. 8598. 10 Yoshioka, T., 1992, Detection of Rolling Contact Subsurface Fatigue Cracks Using Acoustic Emission Technique, Lubr. Eng. 49 4 , pp. 303308. 11 Voskamp, A. P., 1985, Material Response to Rolling Contact Loading, Trans. ASME, J. Tribol., 107 3 , pp. 359366. 12 Holroyd, T., 2000, The Acoustic Emission & Ultrasonic Monitoring, 1st ed., Coxmoor, Oxford, UK. 13 Al-Dossary, S., Raja Hamzah, R. I., and Mba, D., Observations of Changes in Acoustic Emission Waveform for Varying Seeded Defect Sizes in a Rolling Element Bearing, Appl. Acoust., in press. Available online March 24, 2008.

041013-14 / Vol. 130, AUGUST 2008

Transactions of the ASME

Downloaded 22 Jun 2011 to 117.206.44.33. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Você também pode gostar