Sie sind auf Seite 1von 12

Tribology International 32 (1999) 469480 www.elsevier.

com/locate/triboint

A review of vibration and acoustic measurement methods for the detection of defects in rolling element bearings
N. Tandon
b

a,*

, A. Choudhury

a ITMME Centre, Indian Institute of Technology, Hauz Khas, New Delhi 110016, India Department of Mechanical Engineering, Regional Engineering College, Silchar 788010, India

Received 5 May 1999; received in revised form 1 September 1999; accepted 12 October 1999

Abstract A review of vibration and acoustic measurement methods for the detection of defects in rolling element bearings is presented in this paper. Detection of both localized and distributed categories of defect has been considered. An explanation for the vibration and noise generation in bearings is given. Vibration measurement in both time and frequency domains along with signal processing techniques such as the high-frequency resonance technique have been covered. Other acoustic measurement techniques such as sound pressure, sound intensity and acoustic emission have been reviewed. Recent trends in research on the detection of defects in bearings, such as the wavelet transform method and automated data processing, have also been included. 2000 Elsevier Science Ltd. All rights reserved.
Keywords: Rolling element bearings; Bearing defects; Condition monitoring; Vibrations; Acoustics

1. Introduction Rolling element bearings nd widespread domestic and industrial applications. Proper functioning of these appliances depends, to a great extent, on the smooth and quiet running of the bearings. In industrial applications, these bearings are considered as critical mechanical components and a defect in such a bearing, unless detected in time, causes malfunction and may even lead to catastrophic failure of the machinery. Defects in bearings may arise during use or during the manufacturing process. Therefore detection of these defects is important for condition monitoring as well as quality inspection of bearings. Different methods are used for detection and diagnosis of bearing defects; they may be broadly classied as vibration and acoustic measurements, temperature measurements and wear debris analysis. Among these, vibration measurements are the most widely used. Several techniques have been applied to measure the vibration and acoustic responses from defective bearings; i.e., vibration measurements in time and frequency

* Corresponding author. Tel./fax: +91-11-685-7658. E-mail address: ntandon@itmmec.iitd.ernet.in (N. Tandon)

domains, the shock pulse method, sound pressure and sound intensity techniques and the acoustic emission method. A lot of research work has been published, mostly in the last two decades, on the detection and diagnosis of bearing defects by vibration and acoustic methods. Some of these works have also been reviewed by researchers. Tandon and Nakra [1] presented a detailed review of the different vibration and acoustic methods, such as vibration measurements in time and frequency domains, sound measurements, the shock pulse method and the acoustic emission technique, for condition monitoring of rolling bearings. Some of the vibration and wear debris analysis techniques such as vibration, shock pulse, spike energy, spectrographic oil analysis, ferrography and chip detection have been reviewed by Kim and Lowe [2] with a particular reference to railway freight cars. A brief review of vibration monitoring techniques in time and frequency domains and their results on rolling element bearings have been presented by Mathew and Alfredson [3]. McFadden and Smith [4] and Kim [5,6] have also presented reviews on some specic techniques for condition monitoring of rolling bearings. Therefore, the objective of the present study is to update these reviews by incorporating recent works and the advanced techniques adopted in bearing defect detection.

0301-679X/99/$ - see front matter 2000 Elsevier Science Ltd. All rights reserved. PII: S 0 3 0 1 - 6 7 9 X ( 9 9 ) 0 0 0 7 7 - 8

470

N. Tandon, A. Choudhury / Tribology International 32 (1999) 469480

2. Vibration and noise generation in bearings

3. Vibration response due to localized defects Two approaches have been adopted by researchers for creating localized defects on bearings to study their vibration response. One is to run the bearing until failure and monitor the changes in their vibration response [5,1822]. Usually the failure is accelerated by either overloading, overspeeding or starving the bearings of lubricants [5,21,22]. The other approach is to intentionally introduce defects in the bearings by techniques such as acid etching, spark erosion, scratching or mechanical indentation, measure their vibration response and compare it with that of good bearings [2334]. In some of these studies the size of the simulated defects has been quantied and varied [26,2933]. The former approach of life tests is quite time-consuming. On the other hand, the testing of bearings with simulated defects is much quicker but preparation of the defective bearings requires special techniques. Several techniques have been applied to measure and analyse the vibration response of bearings with localized defects. These techniques are not totally independent; rather, in many cases, they are complementary to one another. 3.1. Time-domain approach The simplest approach in the time domain is to measure the overall root-mean-square (RMS) level and crest factor, i.e., the ratio of peak value to RMS value of acceleration. This method has been applied with limited success for the detection of localized defects [30,32]. Some statistical parameters such as probability density and kurtosis have been proposed for bearing defect detection [35,36]. The probability density of acceleration of a bearing in good condition has a Gaussian distribution, whereas a damaged bearing results in non-Gaussian distribution with dominant tails because of a relative increase in the number of high levels of acceleration [35]. However, Mathew and Alfredson [3] have reported obtaining a near-Gaussian distribution for some damaged bearings also. Instead of studying the probability density curves, it is often more informative to examine the statistical moments of the data, dened as

Several studies [713] have been conducted to explain the mechanism of vibration and noise generation in bearings. Bearings act as a source of vibration and noise due to either varying compliance or the presence of defects in them. Radially loaded rolling element bearings generate vibrations even if they are geometrically perfect. This is because of the use of a nite number of rolling elements to carry the load. The number of rolling elements and their position in the load zone change with bearing rotation, giving rise to a periodical variation of the total stiffness of the bearing assembly. This variation of stiffness generates vibrations commonly known as varying compliance vibrations [7,8]. When the bearing races are assumed as continuous systems, the changing direction of the contact forces applied by the rolling elements may cause exural or ring-mode vibration of the races even if they are geometrically perfect [9,10]. However, the presence of a defect causes a signicant increase in the vibration level. Bearing defects may be categorized as distributed or local. Distributed defects include surface roughness, waviness, misaligned races and off-size rolling elements [811]. The surface features are considered in terms of their wavelength compared with the Hertzian contact width of the rolling element raceway contacts. Surface features of wavelength of the order of the contact width or less are termed roughness, whereas longer-wavelength features are termed waviness [11]. Distributed defects are caused by manufacturing error, improper installation or abrasive wear [12,13]. The variation in contact force between rolling elements and raceways due to distributed defects results in an increased vibration level. The study of vibration response due to this category of defect is, therefore, important for quality inspection as well as condition monitoring. Localized defects include cracks, pits and spalls on the rolling surfaces. The dominant mode of failure of rolling element bearings is spalling of the races or the rolling elements, caused when a fatigue crack begins below the surface of the metal and propagates towards the surface until a piece of metal breaks away to leave a small pit or spall. Fatigue failure may be expedited by overloading or shock loading of the bearings during running and installation [13]. Electric pitting or cracks due to excessive shock loading are also among the different types of bearing damage described in the literature [1317]. Whenever a local defect on an element interacts with its mating element, abrupt changes in the contact stresses at the interface result which generates a pulse of very short duration. This pulse produces vibration and noise which can be monitored to detect the presence of a defect in the bearing.

Mx

xnP(x) dx n1, 2, 3, , m,

(1)

where P(x) is the probability density function of instantaneous amplitude x. The rst and second moments are well known, being the mean value and the variance, respectively. The third moment normalized with respect to the cube of standard deviation is known as the coefcient of skewness. The fourth moment, normalized with respect to the fourth power of standard deviation,

N. Tandon, A. Choudhury / Tribology International 32 (1999) 469480

471

is quite useful. This is called kurtosis and is given by the expression

kurtosis, b2

(xx )4P(x) dx s4 , (2)

where x is the mean. Dyer and Stewart [36] rst proposed the use of kurtosis for bearing defect detection. For an undamaged bearing with Gaussian distribution, the kurtosis value is close to 3. A value greater than 3 is judged by itself to be an indication of impending failure and no prior history is required. However, one disadvantage is that the kurtosis value comes down to the level of an undamaged bearing (i.e., 3) when the damage is well advanced. Therefore, it has been suggested to measure kurtosis in selected frequency bands [36]. White [37] studied the effectiveness of this method under a simulated condition. Several other studies [27,28,34,36,38] have also shown the effectiveness of kurtosis in bearing defect detection but in some cases [3,5,40] the method could not detect the incipient damage effectively. Kurtosis has not become a very popular method in industry for the condition monitoring of bearings. Local defects can also be detected in the time domain by displaying the vibration signal on an oscilloscope or plotting it on a chart recorder and observing the presence of periodic peaks due to impact of the rolling element with the defects [26,29,40,41]. Gustafsson and Tallian [40] proposed a method of defect detection based on the number of peaks crossing a preset voltage level. Some bandpass ltering techniques have also been proposed in the time domain. The principle is based on the fact that structural resonances are excited in the highfrequency zone due to impulsive loading caused, for example, from spalling of the races or rolling elements and can be detected by a transducer whose resonant frequency is tuned to it. The shock pulse method [42], which works on this principle, uses a piezoelectric transducer having a resonant frequency based at 32 kHz (some instruments based on resonant frequency around 100 kHz have also been used). The shock pulses caused by the impacts in the bearings initiate damped oscillations in the transducer at its resonant frequency. Measurement of the maximum value of the damped transient gives an indication of the condition of rolling bearings. Low-frequency vibrations in the machine, generated by sources other than rolling bearings, are electronically ltered out. The shock pulse value generated by good bearings due to surface roughness has been found empirically to be dependent upon the bearing bore diameter and speed. This value, called the initial value, is subtracted from the shock value of the test bearing to obtain a normalized shock pulse value. The maximum

normalized shock value is a measure of the bearing condition. Shock pulse meters are simple to use so that semiskilled personnel can operate them. They give a single value indicating the condition of the bearing straightaway, without the need for elaborate data interpretation as required in some other methods. The shock pulse method has gained wide industrial acceptance and has been reported to be successful in the detection of rolling element bearing defects [3,5,34,39,43]. Some investigators [31,44,45] have reported that the method could not effectively detect defects at low speeds. However, Butler [43] has reported success of the shock pulse method in the detection of defects in low-speed spherical roller bearings in a paper production line. An on-line bearing condition monitoring technique based on the shock pulse method has been suggested by Morando [46]. 3.2. Frequency-domain approach Frequency-domain or spectral analysis of the vibration signal is perhaps the most widely used approach of bearing defect detection. The advent of modern fast Fourier transform (FFT) analysers has made the job of obtaining narrowband spectra easier and more efcient. Both lowand high-frequency ranges of the vibration spectrum are of interest in assessing the condition of the bearing. The interaction of defects in rolling element bearings produces pulses of very short duration whenever the defect strikes or is struck owing to the rotational motion of the system. These pulses excite the natural frequencies of bearing elements and housing structures, resulting in an increase in the vibrational energy at these high frequencies. The resonant frequencies of the individual bearing elements can be calculated theoretically [1,47,48]. It is difcult to estimate how these resonances are affected on assembly into a full bearing and mounting in a housing. However, it is indicated [24] that resonances are not altered signicantly. These natural frequencies are usually more than 5 kHz [1]. Therefore, monitoring the increase in the level of vibrations in the high-frequency range of the spectrum is an effective method of predicting the condition of rolling element bearings and has been used successfully by several investigators [3,21,23,24,26]. Catlin [49] has indicated that the natural frequency for which the wavelength closely matches the pulse length, is most strongly excited. Several parameters such as arithmetic mean, geometric mean and correlation have been suggested [50,51] to quantify the differences in spectra for good bearings and damaged bearings. Each bearing element has a characteristic rotational frequency. With a defect on a particular bearing element, an increase in vibrational energy at this elements rotational frequency may occur. These characteristic

472

N. Tandon, A. Choudhury / Tribology International 32 (1999) 469480

defect frequencies can be calculated from kinematic considerations; i.e., the geometry of the bearing and its rotational speed [1,23,26,40,52,53]. For a bearing with a stationary outer race, these frequencies are given by the following expressions: cage frequency, wc ws d 1 cos a , 2 D

ball spinning frequency, wb

Dws d2 1 2 cos2 a , 2d D

(3)

(4)

outer race defect frequency, wodZwc Zws d 1 cos a , 2d D

(5)

inner race defect frequency, widZ(wswc) Zws d 1 cos a 2 D

(6)

and rolling element defect frequency, wre2wb d2 D ws 1 2 cos a2 , d D

(7)

where ws is the shaft rotation frequency in rad/s, d is the diameter of the rolling element, D is the pitch diameter, Z is the number of rolling elements and a is the contact angle. For normal speeds, these defect frequencies lie in the low-frequency range and are usually less than 500 Hz. In practice, however, these frequencies may be slightly different from the calculated values as a consequence of slipping or skidding in the rolling element bearings [53]. Several researchers [24,26,27,32,54,55] have reported success in bearing defect detection by identifying these rotational frequencies. It has also been observed [24,26,27,32] that, in case of a defect on a moving element such as the inner race or a rolling element, the spectrum has sidebands about the components at characteristic defect frequencies. A typical spectrum [32] due to an inner race defect is shown in Fig. 1. The sidebands have been attributed to the time-related changes in defect position relative to the vibration measuring position [26]. Tandon and Choudhury [56] have derived expressions for frequencies and relative amplitudes of the various spectral lines based on the exural vibration of races due to a localized defect on one of the bearing elements. In some work [23,57,58], it has been mentioned that it is difcult to obtain a signicant peak at these frequencies in the direct spectrum obtained from a defective bearing. This is due to the fact that noise or vibration from other sources masks the vibration signal from the bearing unless the defect is sufciently large. Tandon and Nakra [32] have also found that direct spectral

analysis can detect defects of comparatively larger sizes only. Ray [45] highlighted the condition under which bearing defect detection becomes difcult. Osuagwu and Thomas [57] have suggested an explanation for the absence of defect frequencies in the spectrum in terms of the average and shift effect produced by the variation of the impact period and the intermodulation effect. In this study the power cepstrum was shown to be an effective diagnostic technique. Power cepstrum is dened as the power spectrum of the logarithmic power spectrum [35]. Tandon [33] has reported that cepstrum can detect outer race defects effectively but failed to detect inner race defects. In order to improve the signal-to-noise ratio and make the spectral analysis more effective, some signal processing techniques have been reported. Braun and Datner [59] have suggested signal decomposition into periodic components and a processing method based on averaging techniques. The adaptive noise cancelling (ANC) technique has also been proposed to improve the signalto-noise ratio in bearing fault diagnosis [60]. Envelope detection or the high-frequency resonance technique (HFRT) is an important signal processing technique which helps in the identication of bearing defects by extracting characteristic defect frequencies (which may not be present in the direct spectrum) from the vibration signal of the defective bearing. A review of the technique has been presented by McFadden and Smith [4]. Each time a defect strikes its mating element, a pulse of short duration is generated that excites the resonances periodically at the characteristic frequency related to the defect location. The resonances are thus amplitudemodulated at the characteristic defect frequency. By demodulating one of these resonances, a signal indicative of the bearing condition can be recovered. In practice, the signal is bandpass-ltered around one of the resonant frequencies, thus eliminating most of the unwanted vibration signals from other sources. This bandpass-ltered signal is then demodulated by an envelope detector in which the signal is rectied and smoothed by low-pass ltering to eliminate the carrier or bandpass-ltered resonant frequency. The spectrum of the envelope signal in the low-frequency range is then obtained to get the characteristic defect frequency of the bearing. The process of extraction of demodulated spectra by the high-frequency resonance technique is shown in Fig. 2. This technique has been used extensively and its success has been demonstrated by several investigators [23,25,3032,6180]. A single-mode vibration model has been developed by McFadden and Smith [6264] to explain the appearance of various spectral lines owing to different defect locations in the demodulated spectrum. They have suggested that the sidebands around the defect frequency are a result of the modulation of carrier frequency by loading and transmission path. This model

N. Tandon, A. Choudhury / Tribology International 32 (1999) 469480

473

Fig. 1.

A typical spectrum obtained from a rolling element bearing with an inner race defect.

Fig. 2.

The process of extraction of demodulated spectra by HFRT.

474

N. Tandon, A. Choudhury / Tribology International 32 (1999) 469480

has been extended by Su and Lin [79] to characterize the vibrations of bearings subjected to various loadings. Martin and Thorpe [80] have suggested normalization of the envelope-detected frequency spectra of the faulty bearing with respect to the healthy bearing to give greater sensitivity to the detection of defect frequencies. The limitation of this technique is that, with advanced damage, the defect frequencies may become submerged in the rising background level of the spectrum [4,74]. This may happen due to the reduced severity of impacts which are generated so frequently that the leading edge of the impact is buried in the decay of the previous impact [4]. However, Burgess [74] argues that this is because of the reduced difference in amplitudes of the random noise oor and the defect peak heights, which also become random as the defect progresses. In recent years, the wavelet transform method has been suggested by some researchers [8183] to extract very weak signals for which FFT becomes ineffective. Wavelet transform provides a variable-resolution time frequency distribution from which periodic structural ringing due to repetitive force impulses, generated upon the passing of each rolling element over the defect, are detected. The adaptive timefrequency resolution makes ` -vis FFT. Mori et al. [83] proposed a it superior vis-a discrete wavelet transform theory which can detect a defect even at pre-spalling stage. 3.3. Proximity transducer technique The literature discussed so far has mostly considered casing-mounted transducers such as accelerometers and velocity pick-up for the measurement of vibration. Some researchers have also used non-contact type displacement or proximity transducers for condition monitoring of rolling element bearings. In these studies, the transducer senses the displacement of the outer race directly as the rolling elements pass under it. Thus the extraneous vibrations of the housing structure are reduced or eliminated and the signal-to-noise ratio is improved. However, the installation of these probes is difcult as it not only involves drilling and tapping of the bearing housing but also ne adjustment of the gap between the probe and the outer race, which can change due to such conditions as vibration, dirt and thermal expansion. This monitoring technique is popularly known as REBAM (Rolling Element Bearing Activity Monitor). Philips and associates [8486] rst reported the use of this technique. They used a bre optic sensor to measure the elastic deformation of the outer ring. Philips [86] has also introduced the concept of ball speed ratio (BSR), which has been found to be a good parameter for bearing performance monitoring. Harker and Hansen [87] used a high-gain eddy current transducer and reported case histories for rotating machinery health monitoring using this technique. REBAM has been

found to be effective in bearing defect detection in some other studies as well [34,88]. Kim [6,8991] has applied this technique to a radially loaded deep groove ball bearing. He has reported an effective diagnostic method for outer and inner race damage to predict the angular location of the damaged spots. He also suggested that this technique, coupled with HFRT, will be very effective for fault detection in rolling element bearings [91]. 3.4. Automated diagnostic system The techniques discussed above need a human interpreter to analyse the results. However, the amount of data collected from modern instruments may be so vast that an automated system providing concise and reliable assessment of the machinery condition becomes necessary. This is equally true for rolling element bearings. In 1989, Li and Wu [92] proposed a monitoring scheme based on pattern recognition. The technique employs short-time-signal processing techniques to extract useful features from bearing vibration to be used by a pattern classier to detect and diagnose the defect. To overcome the shortcomings of this method, bispectral analysis of vibration has been applied to detect the presence of the characteristic defect frequency and its harmonics [93]. The pattern recognition technique is then applied to classify the condition of a bearing based on its bicoherence. An automatic fault diagnosis system for ball bearings, based on processing of time-domain signatures and a pattern recognition technique, has also been reported [94]. In recent times, articial neural networks have emerged as a popular tool for signal processing and pattern classication tasks, and are suitable for condition monitoring programs. An articial neural network can be dened as a mathematical model of the human brain and has the ability to learn to solve a problem, rather than having to be preprogrammed with a precise algorithm. Baillie and Mathew [95] proposed a model based on a neural network for fault detection in rolling element bearings. The system consists of a collection of parametric time-series models, one for each class of bearing fault to be identied, based on a back-propagation neural network. This time-domain-based model has the advantage that the diagnosis can be performed using very short data lengths and is suitable for application in slow- and variable-speed machinery. However, the main drawback is that the model cannot handle very large data without misclassifying the fault. To overcome this shortcoming, they have proposed to present the vibration data to the neural network in the frequency domain [96] and have also proposed that characteristic defect frequencies can be used for classication of faults. In separate research [97], a neural bearing analyser model has been

N. Tandon, A. Choudhury / Tribology International 32 (1999) 469480

475

developed taking only certain areas of the vibration spectrum and using a back-propagation network.

4. Vibration response due to distributed defects As discussed earlier, distributed defects include surface irregularities like roughness, waviness or off-size rolling elements. The vibration response for these defects has been studied mostly in the frequency domain. Vibration produced by surface roughness or shorterwavelength features has been studied by Sayles and Poon [98] and is found to be signicant only when the asperities break through the lubricant lm and contact the opposing surface. However, longer-wavelength features (waviness) or varying rolling element diameter have a more dominant effect on the vibration level. The frequency limit in which waviness produces a signicant vibration level has been assessed [11,99] and is found to be below 60 times the rotational speed. Systematic studies of vibration produced by waviness were rst made by Tallian and Gustafsson [8]. The vedegree-of-freedom motion of the outer race considered as a body which is rigid except for the contact deformations was analysed with a linearized dynamic model. The waviness orders of major importance have been assessed and are found to have a simple numerical relationship with the number of rolling elements. Meyer et al. [9] proposed an analytical model for exural vibration of the stationary race under axial load due to waviness on the moving race or unequal ball diameter. The amplitudes of the important frequency components have also been predicted in the model. The model was extended later by Choudhury and Tandon [10] for radially loaded rolling bearings. Vibration forces produced by axially loaded angular contact ball bearings under the inuence of waviness on various bearing elements have been studied by Wardle [100]. He considered an additional mass (housing) attached to the stationary outer ring. Analytical results for radial and axial vibrations have been veried experimentally [101]. A linear model has been proposed by Yhland [102] for vibration of the shaft bearing system due to form errors. Waviness orders of major importance for the vibration response of a tapered roller bearing have been investigated experimentally by Ohta and Sugimoto [103]. Sunnersjo [12] has studied the vibration of a radially loaded bearing due to inner race waviness or varying roller diameter, and found the signicant peaks to occur at harmonics of shaft and cage speed, respectively, with a sideband spaced at roller passage frequency in the case of inner race waviness. Similar cage harmonics have been observed by Hine [104] for circumferential ball diameter variation in the bearing of a turbopump in the main engine of a space shuttle. The effects of surface irregularities have also been investigated [105,106] to explain the spectral lines

observed in the demodulated spectra of a normal tapered roller bearing. Since many of the frequencies resulting from distributed defects coincide with those due to localized defects, it becomes difcult to identify from frequency information alone whether a peak at a particular frequency is due to a localized or a distributed defect. Therefore, it has been suggested [107,108] that, in addition to frequency information, the amplitudes of the spectral components should also be studied for defective bearings. Some studies [109111] have also been carried out to estimate the roughness or waviness of the races or the balls in bearings by vibration analysis. Laser Doppler vibrometry has been proposed for surface prole measurements in rolling bearings [112].

5. Acoustic emission response from defective bearings Acoustic emission (AE) is the phenomenon of transient elastic wave generation due to a rapid release of strain energy caused by a structural alteration in a solid material under mechanical or thermal stresses. Generation and propagation of cracks, growth of twins, etc. associated with plastic deformation are among the primary sources of AE. Hence it is an important tool for condition monitoring through non-destructive testing. AE instrumentation consists of a transducer, mostly of the piezoelectric type, a preamplier and a signal-processing unit. The transducers, which have very high natural frequency, have a resonant-type response. The bandwidth of the AE signal can also be controlled by using a suitable lter in the preamplier. The most commonly measured AE parameters are ringdown counts, events and peak amplitude of the signal. These are demonstrated on a typical AE signal shown in Fig. 3. Ringdown counts involve counting the number of times the amplitude exceeds a preset voltage level (threshold level) in a given time and gives a simple number characteristic of the signal. An event consists of a group of ringdown counts and signies a transient wave. The advantage of acoustic emission monitoring over vibration monitoring is that the former can detect the growth of subsurface cracks, whereas the latter can detect defects only when they appear on the surface. It is also important to note that the energy released by neighbouring components in the vibrational frequency range (up to 50 kHz), which often masks the vibrational energy released from a defective rolling element bearing, do not affect the AE signal released in the very high frequency range. Several studies have been conducted to investigate the AE response of defective bearings. In the case of AE monitoring of local defects also, two approaches of life

476

N. Tandon, A. Choudhury / Tribology International 32 (1999) 469480

Fig. 3.

A typical acoustic emission burst signal.

tests [113117] and simulated defects [118120] have been adopted by researchers. In 1979, Rogers [121] suggested the application of acoustic emission as a measure of the condition of slow-speed anti-friction bearings of slewing cranes in offshore gas production platforms. Yoshioka and Fujiwara [113,114] have shown that AE parameters can detect defects before they appear in the vibration acceleration range and can also detect the possible sources of AE generation during a fatigue life test of thrust ball bearings. They also measured the propagation initiation time of cracks and the propagation time until aking occurs by a combination of AE parameters and vibration acceleration [115]. The source locator system was also improved later by introducing two AE sensors in the system and measuring the difference of arrival times of acoustic emission signals at the sensors [116,117]. Acoustic emission signals have been shown to detect defects in the form of a ne scratch on the inner race of axially loaded angular contact ball bearings but at low speeds only [44,118]. Tandon and Nakra [119] have demonstrated the usefulness of some acoustic emission parameters, such as peak amplitude and count, for the detection of defects in radially loaded ball bearings at low and normal speeds. The distribution of events by counts and peak amplitude has also been used for quality inspection of bearings to judge whether the bearing is a new or regenerated one [122]. Tan [123] has suggested that the measurement of area under the amplitudetime curve is a preferred method for defect detection in rolling element bearings. He also applied the adaptive noise

cancellation technique to lter out background AE noise emitted by other machine components [124]. The usefulness of demodulated AE signals in detecting defects in rolling element bearings has been demonstrated by some researchers [120,125127].

6. Acoustic noise response Measurement of acoustic noise can also be used for the detection of defects in rolling element bearings. These measurements are normally carried out in two modes: sound pressure and sound intensity. Sound pressure generated by good bearings has been studied by several researchers [128131], but very little literature is available on sound measurements as a defect detection technique. Igarashi and Yabe [132] have shown the usefulness of sound pressure measurement for the detection of defects in axially loaded ball bearings. The role of surface irregularities in the production of noise in rolling contact has been studied with the help of sound pressure measurement [133]. Sound intensity measurement, a comparatively recent technique, has also been tried successfully for the detection of defects in rolling element bearings [5]. Sound intensity is dened as the time-averaged rate of ow of sound energy through unit area. Unlike sound pressure, it is a vector quantity and the two-microphone intensity probe has directional characteristics. Sound intensity in the frequency domain can be obtained from the imagin-

N. Tandon, A. Choudhury / Tribology International 32 (1999) 469480

477

ary part of the cross-spectrum between the signals of two closely spaced microphones [134]. A special twomicrophone probe is used for intensity measurements. The imaginary part of the cross-spectrum can be obtained directly using a dual-channel FFT analyser. Tandon and Nakra [135] have shown the usefulness of sound intensity measurement as a bearing diagnostic technique and have concluded that, for this purpose, it is more effective than sound pressure measurement. Spectral analysis of demodulated signals received by a sound meter has been suggested as a monitoring tool for wayside detection of railroad roller bearing defects [136]. The detectability of defects by sound measurement may be affected by sources other than bearing noise unless adequate precautions are taken to isolate the latter. A partial or full acoustic enclosure lined with soundabsorbing material is usually constructed around the test bearing for this purpose. Design of such anechoic chambers has also been discussed in the literature [133,137].

technique seems to be better than sound pressure measurements for bearing diagnostics. Acoustic emission measurements have also been used successfully for detecting defects in rolling element bearings. Some studies indicate that these measurements are better than vibration measurements and can detect a defect even before it appears on the surface. Demodulation of AE signals for bearing defect detection has also been suggested. In recent years, attention has also been focused on the automated interpretation of data for bearing diagnostics. The pattern recognition technique and neural networks have been applied to data obtained from vibration measurements in both time and frequency domains for the detection of defects in rolling element bearings. References
[1] Tandon N, Nakra BC. Vibration and acoustic monitoring techniques for the detection of defects in rolling element bearings a review. Shock Vibr Digest 1992;24(3):311. [2] Kim PY, Lowe IRG. A review of rolling element bearing health monitoring. In: Proceedings of Machinery Vibration Monitoring and Analysis Meeting, Vibration Institute, Houston, TX, 1921 April, 1983. p.14554. [3] Mathew J, Alfredson RJ. The condition monitoring of rolling element bearings using vibration analysis. Trans ASME, J Vibr, Acoust, Stress Reliab Design 1984;106:44753. [4] McFadden PD, Smith JD. Vibration monitoring of rolling element bearings by the high frequency resonance technique a review. Tribol Int 1984;17(1):310. [5] Kim PY. A review of rolling element bearing health monitoring (II): preliminary test results on current technologies. In: Proceedings of Machinery Vibration Monitoring and Analysis Meeting, Vibration Institute, New Orleans, LA, 2628 June, 1984. p.12737. [6] Kim PY. A review of rolling element bearing health monitoring (III): preliminary test results on eddy current proximity transducer technique. In: Proceedings of 3rd International Conference on Vibration in Rotating Machinery, York, UK, 1113 September, 1984. p.11925. [7] Sunnersjo CS. Varying compliance vibrations of rolling bearings. J Sound Vibr 1978;58(3):36373. [8] Tallian TE, Gustafsson OG. Progress in rolling bearing vibration research and control. ASLE Trans 1965;8(3):195207. [9] Meyer LD, Ahlgren FF, Weichbrodt B. An analytic model for ball bearing vibrations to predict vibration response to distributed defects. Trans ASME, J Mech Design 1980;102:20510. [10] Choudhury A, Tandon N. A theoretical model to predict vibration response of rolling bearings to distributed defects under radial load. Trans ASME, J Vibr Acoust 1998;120(1):21420. [11] Wardle FP, Poon SY. Rolling bearing noise cause and cure. Chartered Mech Engr July/August 1983:3640. [12] Sunnersjo CS. Rolling bearing vibrations geometrical imperfections and wear. J Sound Vibr 1985;98(4):45574. [13] Washo MW. A quick method of determining root causes and corrective actions of failed ball bearings. Lubric Eng 1996;52(3):20613. [14] Volker E, Martin HR. Early detection of damage in rolling bearings. ISA Trans 1984;23(3):2732. [15] Scheithe W. Better bearing vibration analysis. Hydrocarbon Process July 1992:5764.

7. Conclusions From a review of studies on vibration and acoustic measurement techniques for the detection of defects in rolling element bearings, it is seen that the emphasis is on vibration measurement methods. Although literature is available for the detection of both localized and distributed defects, the former has drawn the attention of a greater number of researchers. Vibration in the time domain can be measured through parameters such as overall RMS level, crest factor, probability density and kurtosis. Among these, kurtosis is the most effective. The shock pulse method has also gained wide industrial acceptance. Vibration measurement in the frequency domain has the advantage that it can detect the location of the defect. However, the direct vibration spectrum from a defective bearing may not indicate the defect at the initial stage. Some signal processing techniques are, therefore, used. The high-frequency resonance technique is the most popular of these and has been successfully applied by several researchers. For this technique, the procedure of signal processing is well established and a good explanation of the resultant demodulated spectrum is also available. The method has a disadvantage that advanced damage is difcult to detect by this method. In recent years, the wavelet transform method has been suggested to extract very weak signals for which Fourier transform becomes ineffective. Very few studies have been carried out on acoustic noise measurements for the detection of bearing defects. Measurements of both sound pressure and sound intensity have been used for this purpose. The sound intensity

478

N. Tandon, A. Choudhury / Tribology International 32 (1999) 469480

[16] Gupta AK, Sharma VK. Failure of ball bearings and its diagnosis by vibration analysis. Chem Eng World 1989;24(5):459. [17] Eschmann P. Ball and roller bearings theory, design and application. Chichester: John Wiley and Sons, 1985. [18] Nishio K, Hoshiya S, Miyachi T. An investigation of the early detection of defects in ball bearings by the vibration monitoring. ASME Paper 79-DET-45. New York: ASME, 1979. [19] Igarashi T, Noda B, Matsushima E. A study on the prediction of abnormalities in rolling bearings (1). J JSLE Int 1980;1:716. [20] Kuhnell BT, Stecki JS. Correlation of vibration, wear debris analysis and oil analysis in rolling element bearing condition monitoring. Maintenance Management Int 1985;5:10515. [21] Reif Z, Lai MS. Detection of developing bearing failures by means of vibration. ASME Design Eng Div (Publ) DE 1989;18(1):2316. [22] Rao BVA, Swarnamani S, Varghese GV. Studies on a test rig to check defective and spurious ball and roller bearings. In: Proceedings of the National Conference on Industrial Tribology, Bombay, India, 1986. p.1.10. [23] Broderick JJ, Burchill RF, Clark HL. Design and fabrication of prototype system for early warning of impending bearing failure. MTI Report MTI-71 TR-1 (prepared for NASA), 1972. [24] Dyer D. Bearing condition monitoring. In: Interim Report 1. Southampton (UK): Department of Mechanical Engineering, Univerisity of Southampton, 1973. [25] Darlow MS, Badgley RH. Early detection of defects in rolling element bearings. SAE Paper 750209. Dearborn (IL): SAE, 1975. 12pp. [26] Igarashi T, Hamada H. Studies on the vibration and sound of defective rolling bearings (rst report: vibration of ball bearings with one defect). Bull JSME 1982;25(204):9941001. [27] Martins LG, Gerges SNY. Comparison between signal analysis for detecting incipient bearing damage. In: Proceedings of the International Condition Monitoring Conference, Swansea, UK, 1013 April, 1984. p.191204. [28] Stronach AF, Cudworth CJ, Johnston AB. Condition monitoring of rolling element bearings. In: Proceedings of the International Condition Monitoring Conference, Swansea, UK, 1013 April, 1984. p.16277. [29] Igarashi T, Kato J. Studies on the vibration and sound of defective rolling bearings (third report: vibration of ball bearings with multiple defects). Bull JSME 1985;28(237):4929. [30] Miyachi T, Seki K. An investigation of the early detection of defects in ball bearings using vibration mnitoring practical limit of detectability and growth speed of defects. In: Proceedings of the International Conference on Rotordynamics, JSMEIFToMM, Tokyo, 1417 September, 1986. p.4038. [31] Tandon N, Nakra BC. Comparison of vibration and acoustic measurement techniques for the condition monitoring of rolling element bearings. Tribol Int 1992;25(3):20512. [32] Tandon N, Nakra BC. Detection of defects in rolling element bearings by vibration monitoring. J Instn Engrs (India) Mech Eng Div 1993;73:27182. [33] Tandon N. A comparison of some vibration parameters for the condition monitoring of rolling element bearings. Measurement 1994;12:2859. [34] Prabhu R. Rolling bearing diagnostics. In: Proceedings of the IndoUS Symposium on Emerging Trends in Vibration and Noise Engineering, New Delhi, 1820 March, 1996. p.31120. [35] Collacott RA. Mechanical fault diagnosis. London: Chapman and Hall, 1977. [36] Dyer D, Stewart RM. Detection of rolling element bearing damage by statistical vibration analysis. Trans ASME, J Mech Design 1978;100(2):22935. [37] White MF. Simulation and analysis of machinery fault signals. J Sound Vibr 1984;93:95116.

[38] Rush AA. Kurtosis a crystal ball for maintenance engineers. Iron Steel Int February 1979:237. [39] Alfredson RJ, Mathew J. Time domain methods for monitoring the condition of rolling element bearings. Mech Eng Trans, IE Australia 1985;ME10(2):1027. [40] Gustafsson OG, Tallian T. Detection of damage in assembled rolling element bearings. ASLE Preprint 61-AM 3B-1. 16th ASLE, Philadelphia, PA, 1961. 39pp. [41] Springer CW. Dont forget the time domain in analyzing bearings: a case study. Tappi J 1988;71(8):1903. [42] The shock pulse method for determining the condition of antifriction bearings, SPM Technical Information. Sweden: SPM Instruments AB. [43] Butler DE. The shock pulse method for the detection of damaged rolling bearings. NDT Int 1973:925. [44] Smith JD. Vibration monitoring of bearings at low speeds. Tribology Int 1982;15:13944. [45] Ray AG. Monitoring rolling contact bearings under adverse conditions. In: Proceedings of 2nd International Conference on Vibration in Rotating Machinery, Cambridge, UK, 14 September, 1980. p.18794. [46] Morando LE. Measuring shock pulses is ideal for bearing condition monitoring. Pulp & Paper 1988;62(12):968. [47] Love AEH. Treatise on the mathematical theory of elasticity. New York: Dover Publications, 1944. [48] Lai HY. Computer aided fault diagnosis of crank systems using engine vibration data. In: Proceedings of 1st International Machinery Monitoring and Diagnosis Conference, NV, 1114 September, 1989. p.24853. [49] Catlin JB Jr. The use of ultrasonic diagnostic techniques to detect rolling element bearing defects. In: Proceedings of Machinery Vibration Monitoring and Analysis Meeting, Vibration Institute, Houston, TX, 1921 April, 1983. p.12330. [50] Alfredson RJ, Mathew J. Frequency domain methods for monitoring the condition of rolling element bearings. Mech Eng Trans, IE Australia 1985;ME10(2):10812. [51] Gudonavicius R, Ziogas V, Kajackas A, Rondomanskas M. On diagnostics of bearings by statistical parameters of their high frequency vibrations. Vibr Eng 1988;2(1):4550. [52] Harris TA. Rolling bearing analysis. New York: John Wiley and Sons, 1966. [53] Prasad H. The effect of cage and roller slip on the measured defect frequency response of rolling element bearings. ASLE Trans 1987;30(3):3607. [54] Taylor JI. Identication of bearing defects by spectral analysis. Trans ASME, J Mech Design 1980;102:199204. [55] McLain DA, Hartman DL. Analysis of defective anti-friction bearings in the paper industry. VirginiaCarolina Section: Tappi Winter Mfg, 1980. p.128. 14 November. [56] Tandon N, Choudhury A. An analytical model for the prediction of the vibration response of rolling element bearings due to a localized defect. J Sound Vibr 1997;205(3):27592. [57] Osuagwu CC, Thomas DW. Effect of inter-modulation and quasi-periodic instability in the diagnosis of rolling element incipient defect. Trans ASME, J Mech Design 1982;104(2):296302. [58] Johnston AB, Stronach AF. Bearing fault detection in a hostile environment. In: Proceedings of International Conference on Condition Monitoring, Brighton, UK, 2123 May, 1986. p.35 44. [59] Braun S, Datner B. Analysis of ball/roller bearing vibrations. ASME Paper 77-WA/DE-5. New York: ASME, 1977. 8pp. [60] Chaturvedi GK, Thomas DW. Bearing fault detection using adaptive noise cancelling. ASME Paper 81-DET-7. New York: ASME, 1981. 10pp. [61] McFadden PD, Smith JD. Information from the vibration of rolling bearings. In: Proceedings of International Condition

N. Tandon, A. Choudhury / Tribology International 32 (1999) 469480

479

[62]

[63]

[64]

[65]

[66]

[67]

[68] [69]

[70]

[71] [72]

[73] [74]

[75]

[76]

[77] [78]

[79]

[80] [81]

[82] [83]

Monitoring Conference, Swansea, UK, 1013 April, 1984. p.17890. McFadden PD, Smith JD. Model for the vibration produced by a single point defect in a rolling element bearing. J Sound Vibr 1984;96(1):6982. McFadden PD, Smith JD. The vibration produced by multiple point defect in a rolling element bearing. J Sound Vibr 1985;98(2):26373. McFadden PD, Smith JD. The vibration produced by a single point defect on the inner or outer race or rolling elements of a bearing under radial or axial load. Technical Report CUED/CMech/TR34. Cambridge (UK): Engineering Department, Cambridge University, 1983. McFadden PD. Condition monitoring of rolling element bearings by vibration analysis. IMechE Paper, Solid Mechanics and Machine Systems Group Seminar, 9 January 1990. London: IMechE. p.4953. McFadden PD. Advances in vibration monitoring of Gears and rolling element bearings. In: Proceedings of IE Austalia/RAeS Joint National Symposium, Melbourne, 89 August, 1985. p.2732. Burchill RF, Frarey JL, Wilson DS. New machinery health diagnostic techniques using high frequency vibration. In: SAE Paper 730930. Dearborn (IL): SAE, 1973. Winn LW, Bull HL. Diagnostic system for ball bearing quality control. SAE Paper 760910. Dearborn (IL): SAE, 1976. 8pp. Prasad H, Ghosh M, Biswas S. Diagnostic monitoring of rolling element bearings by high frequency resonance technique. ASLE Trans 1985;28(4):43948. Prasad H, Ghosh M, Biswas S. Early detection of defects in anti-friction bearings. In: Proceedings of National Conference on Industrial Tribology, Hyderabad, India, July, 1984. p.1.10. Prasad H. Condition monitoring of anti-friction bearings. J Instn Engrs (India) Mech Eng Div 1988;69:6574. Ghosh M, Rajamani A. Diagnostic monitoring of antifriction bearings. In: Proceedings of International Conference on Noise Control Engineering, Cambridge, MA, July, 1986. p.118790. Brown DN. Envelope analysis detects bearing faults before major damage occurs. Pulp & Paper 1989;63(13):1137. Burgess PFJ. Antifriction bearing fault detection using envelope detection. Trans Instn Prof Engrs, New Zealand 1988;15(2):7782. Voronkin VA, Mikhailov VI, Pavlov KA. Diagnostics of bearings in electric machinery by using high-frequency vibration signatures. Vibr Eng 1988;2(4):40712. Teo S. Condition monitoring of slow speed rolling element bearings in a mechanically noisy environment. Appita 1989;42(3):2068. Gore D, Edgar G. Techniques for the early detection of rolling element bearing failures. RCA Engr 1984;29(5):717. Ratcliffe GA. Condition monitoring of rolling element bearings using the envelope technique. IMechE Paper, Solid Mechanics and Machine Systems Group Seminar, 9 January 1990. London: IMechE. p. 5565. Su Y-T, Lin S-J. On initial fault detection of a tapered roller bearing: frequency domain analysis. J Sound Vibr 1992;155(1):7584. Martin KF, Thorpe P. Normalised spectra in monitoring of rolling bearing elements. Wear 1992;159:15360. Zhengjia H, Jiyuan Z, Qingfeng M, Yiming N. Wavelet transform in tandem with autoregressive technique for monitoring and diagnosis of machinery. In: Proceedings of Condition Monitoring and Diagnostic Engineering Management Conference, New Delhi, 1994. p.20411. Li CJ, Ma J. Wavelet decomposition of vibrations for detection of bearing-localized defects. NDT&E Int 1997;30(3):1439. Mori K, Kasashima N, Yoshioka T, Ueno Y. Prediction of spal-

[84] [85]

[86] [87]

[88] [89]

[90] [91] [92]

[93]

[94]

[95]

[96]

[97]

[98] [99] [100]

[101]

[102]

[103] [104] [105] [106]

[107]

ling on a ball bearing by applying the discrete wavelet transform to vibration signals. Wear 1996;195:1628. Philips GJ, Hirshfeld F. Rotating machinery bearing analysis. Mech Eng July 1980:2833. Philips GJ. Rotating machinery rolling element bearing performance monitoring using the ber optic method. In: Proceedings of ASME Conference on Reliability, Stress Analysis and Failure Prevention Methods in Mechanical Design, San Fransisco, CA, 1821 August, 1980. p.3743. Philips GJ. Bearing performance investigations through speed ratio measurements. ASLE Trans 1979;22(4):30714. Harker RG, Hansen JS. Roling element bearing monitoring using high gain eddy current transducers. Trans ASME, J Eng Gas Turbine Power 1985;107:1604. Sandy J. Monitoring and diagnostics for rolling element bearings. Sound Vibr 1988;22(6):1620. Kim PY. An effective method for bearing health monitoring. In: Proceedings of 7th International Symposium on Air Breathing Engines, Beijing, 26 September, 1985. p.18390. Kim PY. Proximity transducer technique for bearing health monitoring. J Propulsion 1987;3(1):849. Kim PY. An overview of rolling element bearing failure detection techniques. Report I. Canada: National Research Council. Li CJ, Wu SM. On line detection of localized defects in bearings by pattern recognition analysis. Trans ASME, J Eng Ind 1989;111:3316. Li CJ, Ma J, Hwang B. Bearing condition monitoring by pattern recognition based bicoherence analysis of vibrations. Proc IMechE, Pt C: J Mech Eng Sci 1996;210:27785. Wang XF, Shi XZ, Xu M. The fault diagnosis and quality evaluation of ball bearings by vibration signal processing. In: Proceedings of 1st International Machinery Monitoring and Diagnosis Conference, NV, 1114 September, 1989. p.31821. Baillie DC, Mathew J. Fault diagnosis of bearings using short data lengths. In: Proceedings of Condition Monitoring and Diagnostic Engineering Management Conference, New Delhi, 1994. p.4107. Baillie DC, Mathew J. Diagnosing rolling element bearing faults with articial neural networks. Acoustic Australia 1994;22(3):7984. Macintyre J, Smith P, Harris T, Brown A. Neural network architecture and their application in condition monitoring. In: Proceedings of Condition Monitoring and Diagnostic Engineering Management Conference, New Delhi, 1994. p.2207. Sayles RS, Poon SY. Surface topography and rolling element vibration. Precision Eng 1981:13744. Poon SY, Wardle FP. Running quality of rolling bearings assessed. Chartered Mech Engr 1978;(April):747. Wardle FP. Vibration forces produced by waviness of the rolling surfaces of thrust loaded ball bearings. Part 1: theory. Proc IMechE 1988;202(C5):30512. Wardle FP. Vibration forces produced by waviness of the rolling surfaces of thrust loaded ball bearings. Part 2: experimental validation. Proc IMechE 1988;202(C5):3139. Yhland E. A linear theory of vibrations caused by ball bearings with form errors operating at moderate speed. Trans ASME, J Tribol 1992;114:34859. Ohta H, Sugimoto N. Vibration characteristics of tapered roller bearings. J Sound Vibr 1996;190(2):13747. Hine MJ. Absolute ball bearing wear measurements from SSME turbopump dynamic signals. J Sound Vibr 1989;128(2):32131. Su Y-T, Lin M-H, Lee M-S. The effects of surface irregularities on roller bearing vibrations. J Sound Vibr 1993;165(3):45566. Su Y-T, Sheen Y-T, Lin M-H. Signature analysis of roller bearing vibrations: lubrication effects. Proc IMechE, Pt C: J Mech Eng Sci 1992;206:193202. Su Y-T, Sheen Y-T. On the detectability of roller bearing dam-

480

N. Tandon, A. Choudhury / Tribology International 32 (1999) 469480

[108]

[109]

[110]

[111]

[112]

[113]

[114]

[115]

[116]

[117]

[118]

[119]

[120]

[121]

age by frequency analysis. Proc IMechE, Pt C: J Mech Eng Sci 1993;207:2332. Tandon N, Choudhury A. Prediction of vibration amplitudes for defective rolling bearings. Int Sound Vibr Digest 1997;3(2):20 5 (electronic journal). Kanai H, Abe M, Kido K. Detection of slight defects in ball bearings using the resonant vibration estimated by the two-pulse model. In: Proceedings of International Conference on Noise Control Engineering, Cambridge, MA, July, 1986. p.11916. Kanai H, Abe M, Kido K. Estimation of the surface roughness on the race or balls of ball bearings by vibration analysis. Trans ASME, J Vibr, Acoust, Stress Reliab Design 1987;109:608. Jixi C. Vibration analysis and quality detection of ball bearings. In: Proceedings of 1st International Machinery Monitoring and Diagnosis Conference, NV, 1114 September, 1989. p.5725. Valliapan R, Lieu DK. Defect characterization of roller bearing surfaces with laser doppler vibrometry. Precision Eng 1992;14(1):3542. Yoshioka T, Fujiwara T. Application of acoustic emission technique to detection of rolling bearing failure. In: Dorneld DA, editor. Acoustic emission monitoring and analysis in manufacturing. New York: ASME, 1984. p.5575. Yoshioka T, Fujiwara T. A new acoustic emission source locating system for the study of rolling contact fatigue. Wear 1982;81:1836. Yoshioka T, Fujiwara T. Measurement of propagation initiation and propagation time of rolling contact fatigue crack by observation of acoustic emission and vibration. In: Dowson D et al, editor. Interface dynamics. Amsterdam: Elsevier, 1988. p.2933. Yoshioka T. Detection of rolling contact subsurface fatigue cracks using acoustic emission technique. Lubric Eng 1992;49(4):3038. Yoshioka T, Takeda M. Clarication of rolling contact fatigue initiation using acoustic emission technique. Lubric Eng 1994;51(1):414. McFadden PD, Smith JD. Acoustic emission transducer for the vibration monitoring of bearings at low speeds. Proc IMechE 1984;198(C8):12730. Tandon N, Nakra BC. Defect detection in rolling element bearings by acoustic emission method. J Acoust Emission 1990;9(1):258. Bagnoli S, Capitani R, Citti P. Comparison of accelerometer and acoustic emission signals as diagnostic tools in assessing bearing damage. In: Proceedings of 2nd International Conference on Condition Monitoring, London, 2425 May, 1988. p.11725. Rogers LM. The application of vibration signature analysis and acoustic emission source location to on-line condition monitoring of anti-friction bearings. Tribol Int 1979;12(2):519.

[122] Bansal V, Gupta BC, Prakash A, Eshwar VA. Quality inspection of rolling element bearing using acoustic emission technique. J Acoust Emission 1990;9(2):1426. [123] Tan CC. Application of acoustic emission to the detection of bearing failures. In: Proceedings of Tribology Conference. Brisbane: IE Australia, 1990:1104. [124] Tan CC. Adaptive noise cancellation of acoustic noise in ball bearings. In: Proceedings of AsiaPacic Vibration Conference on Machine Condition Monitoring, 1991:5.2835. [125] Tavakoli MS. Bearing fault detection in the acoustic emission frequency range. In: Proceedings of Noise Conference. USA: INCE, 1991. p.7986. [126] Hawman MW, Galinatis WS. Acoustic emission monitoring of rolling element bearings. In: Proceedings of Ultrasonic Symposium. IEEE, 1988. p.8859. [127] Smulders A, Loob C. Machine condition monitoring using multi-parameter measurement. In: Proceedings of Condition Monitoring and Diagnostic Engineering Management Conference, New Delhi, 1994. p.14753. [128] Lalwani RJ. Some aspects of ball bearing noise. In: Proceedings of Inter-Noise 73, Copenhagen, 1973. p.1715. [129] Jayarasm VD, Jarchow F. Experimental studies on ball bearing noise. Wear 1978;46:3216. [130] Nagamatsu A, Fukuda M. Sound noise generated from ball bearing in high speed rotation. Bull JSME 1978;21(158):130610. [131] Cempel C, Stepaniak Z. Noise and vibration of the new ball bearings. In: Proceedings of 4th International Tribology Conference, PaisleyGlasgow, UK, 1979. p.42533. [132] Igarashi T, Yabe S. Studies on the vibration and sound of defective rolling bearings (second report: sound of ball bearings with one defect). Bull JSME 1983;26(220):17918. [133] Anantapadmanaban T, Radhakrishnan V. An investigation of the role of surface irregularities in the noise spectrum of rolling and sliding contacts. Wear 1982;83:399409. [134] Fahy FJ. Measurement of acoustic intensity using the crossspectral density of two microphone signals. J Acoust Soc Am 1977;62(4):10579. [135] Tandon N, Nakra BC. The application of the sound intensity technique to defect detection in rolling element bearings. Appl Acoust 1990;29(3):20717. [136] Florom RL, Hiatt AR, Bambara JE, Smith RL. Wayside acoustic detection of railroad roller bearing defects. ASME Paper 87WA/RT-11. New York: ASME, 1987. 11pp. [137] Ananthapadmanabhan T, Krishnamurthy K, Radhakrishnan V, Swarnamani S. Design of an anechoic chamber to study acoustic characteristic of contacting surfaces. J Instn Engrs (India) 1983;63(ME5):18995.