PROPOSING A NEW ACOUSTIC EMISSION PARAMETER FOR BEARING CONDITIONMONITORING IN ROTATING MACHINES

Seyed Ali Niknam1,2, Victor Songmene1 and Y.H. Joe Au21Department of Mechanical Engineering, École de Technologie Supérieure (ÉTS), Montreal, QC, Canada

2School of Engineering and Design, Brunel University, Uxbridge, U.K.E-mail: seyed-ali.niknam.1@ens.etsmtl.ca; victor.songmen@etsmtl.ca; joe.au@brunel.ac.uk

Received August 2012, Accepted February 2013No. 12-CSME-87, E.I.C. Accession 3407

ABSTRACTBearings are important machine parts and their condition is often critical to success of an operation orprocess, hence there is a great need for periodic knowledge of their performance. According to reportedresearch works in the past several years, it is believed that the extracted information from acoustic emission(AE) signals can be used for bearing condition monitoring. In this work, a novel parameter based on usingthe ratio of AE mean (µ) and AE standard deviation (σ ), formulated as µ/σ is proposed to distinguishbetween lubricated and dry bearings. A heavy duty test rig was used in experimental work. Various levelsof radial loads and rotational speed (ω) were applied to rotating shaft, which is connected to rolling elementbearings. It was found that, except few cases, regardless of various levels of radial loads used, at higherlevels of rotational speed, dry and lubricated bearings can be clearly distinguished when using proposedparameter.

Keywords: acoustic emission; condition monitoring; bearing; low rotating speed machine.

PROPOSER UN NOUVEAU PARAMÈTRE D’ÉMISSION ACOUSTIQUE PORTANT SUR LASURVEILLANCE DE L’ÉTAT DE LA MACHINE TOURNANTE

RÉSUMÉLes roulements sont parmi les pièces critiques de la machine et leur état est indispensable à la réussited’une opération ou d’un procédé, d’où la nécessité de connaître périodiquement leur performance en coursde fonctionnement. Les travaux de recherche effectués au cours des dernières années ont montré que lesinformations émises sur les signaux acoustiques (AE) peuvent être utilisées pour la surveillance de l’état despaliers à roulements des machines. Dans ce travail, un nouveau paramètre basé sur l’utilisation du rapportdes AE moyenne (µ) et AE écart-type (σ ), formulée comme µ/σ est proposé. Durant les expériences,plusieurs niveaux de charges radiales et de vitesse de rotation (ω) ont été appliqués sur l’arbre qui est reliéaux paliers à roulements. Il a été constaté que le paramètre proposé a permis de distinguer les deux typesde paliers testés (à sec et lubrifié) à des niveaux élevés de vitesse de rotation indépendamment de différentsniveaux de charges radiales utilisées.

Mots-clés : émission acoustique ; surveillance ; palier à roulement ; machine à faible vitesse de rotation.

Transactions of the Canadian Society for Mechanical Engineering, Vol. 37, No.4, 2013 1105

1. INTRODUCTION

Rolling element bearings are perhaps the most omnipresent machine elements that can be found in almost allrotating machines. If a manufacturing plant performs efficiently, the bearings need to be kept in acceptableconditions. Considering that the bearing failure may occur from a variety of causes [1], having periodicknowledge of their health status is desirable. There are two approaches for bearing maintenance: (1) statis-tical bearing life estimation and (2) bearing condition monitoring [2]. The latter approach is more reliable,because it provides up to date information of bearing condition. The most popular bearing condition mon-itoring techniques use vibration and AE signal analysis. Comparative studies of various vibration and AEsignal acquisition methods were reported in [3–5]. Previous studies [6–8] confirmed that AE monitoring ismore reliable than vibration monitoring, as the former can detect the subsurface crack growth, whereas thelatter can at best detect a defect only when it emerges on the surface of a structure. Yashioka [9] describedan AE system for source location. Bashir et al. [10] studied the variation of AE energy during fatiguecrack growth. Further investigations reported in [11–16] stated that AE is an effective condition monitoringmethod for early fault detection in bearings.

This article investigates the effectiveness of proposed AE parameter (µ/σ ) to distinguish between lubri-cated and dry bearings in a rotating machine. The µ/σ is formulated as a function of arithmetic mean (µ)and standard deviation (σ ), where

µ =1N

N

∑i=1

Xi (1)

σ =1N

N

∑i=1

(Xi −µ)2 (2)

where Xi, i = 1, . . . ,N are data points recorded from each AE signal.This article is organized as follows: In Section 2, bearing and AE are briefly introduced. In Section 3,

methodology and experimental devices are described, followed by results in Section 4. Finally, the paper isconcluded in Section 5.

2. BACKGROUND

2.1. BearingBearing are mainly classified into rolling element bearing (see Fig. 1) and plain bearing (see Fig. 2). Theclassification is based on the friction mechanism that occurs in the contacting interface. Rolling elementbearings include all forms of bearing that utilize the rolling action of balls or rollers to provide minimumfriction in the constrained motion of one body relative to another. Both ball and roller bearings can be usedas a part of rotating machine elements.

The common types of bearing failures are flaking (see Fig. 3), fatigue, faulty installation, crack, con-tamination, wear, corrosion, electrical discharge damage and indentations. There are various solutions toimprove the bearing performance, including the use of lubricant in bearing surfaces. Lubrication is mainlyused to reduce the friction between two contacting surfaces sliding friction between the rolling elements onthe one hand and the cage and raceways on the other. Moreover, the lubricant can protect raceways androlling elements against corrosive pitting and rust, and it can also be used as a heat transfer medium to main-tain an even temperature in the bearing. It must also be chemically neutral and non-corrosive. The choice oflubricant lies between grease and oil.

2.2. Acoustic EmissionAcoustic emission is a natural phenomenon of sound generation applied to the spontaneously generated elas-tic waves produced within a material under stress [4]. Nowadays the technology of AE as a non-destructive

1106 Transactions of the Canadian Society for Mechanical Engineering, Vol. 37, No. 4, 2013

Fig. 1. Example of a rolling element bearing [17].

Fig. 2. Example of plain bearing [17].

Fig. 3. Inner rings with initial flaking and spreading flaking [17].

technique (NDT) is very well established and highly sophisticated AE systems are commercially available.A distinct feature of AE is its ability to monitor an entire structure often without taking it out of service. Thismay offer better expense reduction as compared to other NDT approaches, such as ultrasonic. AE signalsare high frequency transients in the range from 20 KHz to 1 MHz. Typically in order to eliminate the noiseeffect, the threshold is used. The AE events that rise above the threshold level are counted for further inves-tigations. In order to capture AE events, an AE sensor is required to convert very small surface displacement

Transactions of the Canadian Society for Mechanical Engineering, Vol. 37, No.4, 2013 1107

Table 1. Experimental parameters.Level Rotational speed (Hz) Pressure (bar) Radial load (N)1 30 20 22682 40 40 45373 50 60 68054 60 80 90735 70 – –6 80 – –7 90 – –8 100 – –

Fig. 4. Schematic diagram of the experimental test rig [20].

(on the order of picometers) into electrical signals that can be amplified and recorded. Most of AE sensorshave a sensing element in the form of a thin disc of piezoelectric material; normally lead zirconate titanate(PZT) that can convert mechanical deformation into electrical voltage [18]. It has an excellent sensitivityand relative immunity to a wide range of industrial applications [19].

3. METHODOLOGY AND APPARATUS

3.1. Experimental PlanIn this study, eight levels of rotational speeds (30–100 Hz), and four levels of radial loads (see Table 1) wereapplied to the test rig rotating shaft (see Figs. 4 and 5), when using dry and lubricated bearings. This led to32 tests with respect to each bearing mode as presented in Tables 2 and 3. The average values of recordedAE parameters are presented in Tables 2 and 3.

3.2. Experimental Test RigA schematic diagram of the experimental test rig is shown in Fig. 4. In order to construct the shaft and baseplate, steel was used as a raw material. The design of the test rig comprises a rotating shaft supported at threepositions: a double-row self-aligned ball bearing SKF 2206 ETN9 (third bearing) at the drive-end, a singlerow self-aligned ball bearing SKF 1206 E (second bearing) near the non-drive, and a spherical roller bearingSKF 22207 E (testing bearing) near to applied load position. They are mounted in the bearing housingsthat in turn were attached to the base plate. The radial load could be applied to the rotating shaft whenusing hydraulic pressure. The shaft rotational speed can be varied by using an inverter and motor controller.

1108 Transactions of the Canadian Society for Mechanical Engineering, Vol. 37, No. 4, 2013

Pressure gauge Hydraulic hand

pump

Supported

frame

Applied load

Fig. 5. Overview of loading system [20].

Hydraulic

cylinder

38 mm

Diameter of area

of contact

Fig. 6. Sketch of applied hydraulic cylinder [20].

The motor is foot-mounted onto the base plate. The connection of the three-stator phase windings of themotor is delta (∆). Therefore, it would accept a three-phase input voltage range of 220–240 Volts for a50 Hz supply frequency. The power of the motor is 0.55 kW (or 0.75 hp). An inverter (or A.C motorspeed controllers) was used to enable the 3-phase induction for voltage deviation by adjusting the rotationalspeed. The control range is 0–10 Volts or 4–20 mA. The power of the selected inverter was 0.75 kW. It canbe set up and operated using the integral keypad. The shaft was driven by a controllable inverter runningat 0.23 rev/sec.

A worm wheel reduction gearbox was used to decrease the shaft speed using a three-phase A.C motor.The flexible mechanical couplings were used on both input and output shafts of the gearbox to incorporatewith the drive shaft from electric motor and the rotating shaft. They are drilled to fix with keyways on bothshafts. At the lower bottom of the base plate, the two U-shape steels are mounted to the bed to preventthe plate’s deformation, when exposed to an excessive load. As depicted in Fig. 5, the hydraulic loadingsystem consists of a hydraulic cylinder, hydraulic hand pump, high-pressure hydraulic hose (1 meter), gaugemounting block and hydraulic pressure gauge. The hydraulics system can provide enough radial force tothe non-drive end of the shaft when using a low height hydraulic cylinder as shown in Fig. 6. Consideringthe radius of the area of contact (r) is 19 mm (see Fig. 6) and the maximum applied pressure is 80 bars, themaximum radial load applied downwards to the non-drive end of the shaft is 9073 N (see Table 1).

Transactions of the Canadian Society for Mechanical Engineering, Vol. 37, No.4, 2013 1109

Fig. 7. Overview of experimental test rig, testing bearing and AE sensor [19].

Table 2. The µ/σ values when using dry bearing.Radial load Average of µ/σ

(N) 30 Hz 40 Hz 50 Hz 60 Hz 70 Hz 80 Hz 90 Hz 100 Hz2268 0.909 0.921 1.061 0.795 0.879 0.369 0.91 0.7774537 0.815 1.205 1.138 0.627 0.868 0.372 0.706 0.7696805 0.828 0.854 0.989 0.884 0.856 0.582 0.677 0.7309073 0.928 1.023 0.898 0.902 0.718 0.523 0.693 0.385

Table 3. The µ/σ values when using lubricated bearing.Radial load Average of µ/σ

(N) 30 Hz 40 Hz 50 Hz 60 Hz 70 Hz 80 Hz 90 Hz 100 Hz2268 0.932 0.941 1.053 1.043 1.042 1.002 0.987 0.8844537 0.965 0.968 0.962 1.106 0.978 0.961 0.998 0.9266805 0.887 0.917 1.054 1.111 0.981 0.934 1.032 0.7319073 0.921 0.985 1.144 1.105 0.949 1.003 0.949 0.941

3.3. Data AcquisitionThe data acquisition system consisted of a Pentium PC with a 1 GHz-CPU that was fitted with a high perfor-mance National Instruments NI 6110 data acquisition card. The NI 6110 card comprises four 12-bit inputresolution channels, simultaneously sampled analogue-to-digital (A/D) input channels with scalable inputlimits and adjustable sampling rates up to 5M samples/s, both controllable by software. The detected sig-nals from AE sensors were recorded on the data acquisition card simultaneously with a sampling frequencyof 3 MHz. The signals were then converted to digital format that can be stored in the PC for further dataprocessing. The software used was LABVIEW (version 5.1). No satisfactory results were obtained duringpreliminary tests when using the threshold levels of 2σ and 3σ . Therefore, 4σ was used as threshold levelto avoid noise effects. In addition, in each test, six trial points were recorded for each AE parameter. Theaverage values of these points were recorded for further analysis. The overview of the experimental testrig, bearings and AE sensors mounting place are shown in Fig. 7. According to pencil lead break tests,the signal attenuation extracted from second and third bearings has insignificant effects on AE power. This

1110 Transactions of the Canadian Society for Mechanical Engineering, Vol. 37, No. 4, 2013

Fig. 8. The µ/σ values when using all levels of rotational speeds and radial load of 2268 N.

Fig. 9. The µ/σ values when using all levels of rotational speeds and radial load of 4537 N.

means that the extracted AE signal information is related to the testing bearing (SKF 22207 E) presented inFigs. 5 and 7.

4. RESULTS

The computed values of µ/σ for dry and lubricated bearings are presented in Tables 2 and 3. As shownin Figs. 8–11, when using rotational speeds 30–50 Hz, regardless of using various levels of radial loads,variation of µ/σ values cannot be interpreted scientifically. In this interval, the proposed AE parameter

Transactions of the Canadian Society for Mechanical Engineering, Vol. 37, No.4, 2013 1111

Fig. 10. The µ/σ values when using all levels of rotational speeds and radial load of 6805 N.

Fig. 11. The µ/σ values when using all levels of rotational speeds and radial load of 9073 N.

cannot be used to distinguish between lubricated and dry bearings. When using a rotational speed of 60–100 Hz (see Figs. 8–11), regardless of any level of radial load used, larger resulting values of µ/σ areobtained for lubricated bearing.

According to confidence interval of 95% and the error bars presented in Figs. 8–11, at the highest ra-dial load used (F = 9076 N) the difference between the computed values of µ/σ from lubricated and drybearings are not within error limits. This exhibits that when the test rig shaft is rotated with 60–100 Hz,the proposed AE parameter can clearly distinguish between dry and lubricated bearings. In this operationalcondition, due to lack of lubrication, more friction occurs in the bearing surfaces which leads to lower valuesof µ/σ in dry bearing. When decreasing radial loads, except few cases as shown in Figs. 8–11, a similar

1112 Transactions of the Canadian Society for Mechanical Engineering, Vol. 37, No. 4, 2013

trend is observed for other cases. According to Fig. 11, at the speed interval 50–100 Hz, the difference be-tween computed values of µ/σ from lubricated and dry bearings are not within error limits. The sensitivityof proposed parameter (µ/σ ) becomes weaker when a lower value of radial load is used (see Figs. 8–10).Based on Fig. 8, within the speed interval of 60–100 Hz, the difference between computed values of µ/σ

from lubricated and dry bearings are not within error limits when using rotational speeds 60 and 80 Hz.It is anticipated that the proposed parameter may show more efficient performance when using higher

levels of radial load and rotational speeds. The results of this study can be useful for subsequent studiesdealing with real time bearing health condition monitoring.

5. CONCLUSION

This study presented the effectiveness of a proposed AE parameter (µ/σ ) that can be used to distinguishbetween lubricated and dry bearings in a rotating machine under similar operating conditions. Irrespectiveto radial load used, at higher levels of rotational speeds (60–100 Hz), larger resulting values of µ/σ wereobtained for lubricated bearing. It was found that the proposed parameter (µ/σ ) can clearly distinguishbetween dry and lubricated bearings at higher levels of radial load and rotational speed (60–100 Hz), whileless sensitivity was observed when smaller radial loads were used. This can be due to higher friction effectsat higher levels of radial load and rotational speed. However irrespective to radial load used, the proposedAE parameter had a poor performance in lower level of rotational speed (30–50 Hz). The experimentaldata in this article are limited to the maximum rotational speed of 100 Hz. More consistence results can beobtained in further experiments at higher levels of rotational speeds. This study can be expanded in widerscopes for real time bearing condition monitoring.

ACKNOWLEDGMENTS

The authors would like to acknowledge Brunel University and École de Technologie Supérieure of Montrealfor providing experimental facilities and financial support in the course of this study.

The constructive comments and contribution of Dr. Chao Wang is also greatly appreciated.

REFERENCES

1. Nisbet, T.S. and Mullet, G., Rolling Bearings in Service: Interpretation of Types of Damage, Hutchinson, 1978.2. Li, Y., Billington, S., Zhang, C., Kurfess,T., Danyluk, S. and Liang, S.., “Adaptive prognostics for rolling element

bearing condition”, Journal Mechanical Systems and Signal Processing, Vol. 13, No. 1, pp. 103–113, 1999.3. Rogers, L., “The application of vibration signature analysis and acoustic emission source location to on-line

condition monitoring of anti-friction bearings”, Tribology International, Vol. 12, No. 2, pp. 51–58, 1979.4. Holroyd, T. and Randall, N., “Use of acoustic emission for machine condition monitoring”, British Journal of

Non-Destructive Testing, Vol. 35, No. 2, pp. 75–78, 1993.5. Choudhury, A. and Tandon, N., “Application of acoustic emission technique for the detection of defects in rolling

element bearings”, Tribology International, Vol. 33, No. 1, pp. 39–45, 2000.6. Tandon, N. and Nakra, B., “Comparison of vibration and acoustic measurement techniques for the condition

monitoring of rolling element bearings”, Tribology International, Vol. 25, No. 3, pp. 205–212, 1992.7. James, Li C. and Li, S., “Acoustic emission analysis for bearing condition monitoring”, Wear, Vol. 185, No. 1,

pp. 67–74, 1995.8. Kaewkongka, T., Au, Y.H.J., Rakowski, R. and Jones, B., “Continuous wavelet transform and neural network for

condition monitoring of rotodynamic machinery”, in Proceedings of 18th IEEE Instrumentation and Measure-ment Technology Conference, Vol. 3, pp. 1962–1966, Centre for Manuf. Metrol., Brunel University, Uxbridge,UK, 2001.

9. Yoshioka, T. and Fujiwara, T., “A new acoustic emission source locating system for the study of rolling contactfatigue”, Wear, Vol. 81, No. 1, pp. 183–186, 1982.

Transactions of the Canadian Society for Mechanical Engineering, Vol. 37, No.4, 2013 1113

10. Bashir, I., Bannister, R. and Probert, D., “Release of acoustic energy during the fatiguing of a rolling-elementbearing”, Applied Energy, Vol. 62, No. 3, pp. 97–111, 1999.

11. Yoon, D.J.I.N., Kwon, O.H.Y., Chung, M.I.H.W.A. and Kim, K.W., “Early detection of damages in journalbearings by acoustic emission monitoring”, Journal of Acoustic Emission, Vol. 13, Nos. 1–2, pp. 1–10, 1995.

12. Miettinen, J. and Andersson,P., “Acoustic emission of rolling bearings lubricated with contaminated grease”,Tribology International, Vol. 33, No. 11, pp. 777–787, 2000.

13. Mba, D. and Marscher, W., “Acoustic emissions and monitoring bearing health”, Tribology Transactions, Vol.46, No. 3, pp. 447–451, 2003.

14. Morhain, A. and Mba, D., “Bearing defect diagnosis and acoustic emission”, IMechE, Part J: Journal of Engi-neering Tribology, Vol. 217, No. 4, pp. 257–272, 2003.

15. Al-ghamd, A.M. and Mba, D.A., “Comparative experimental study on the use of acoustic emission and vibration,analysis for bearing defect identification and estimation of defect size”, Mech. Syst. Signal Process, Vol. 20, No.7, pp. 1537–1571, 2006.

16. Niknam, S.A., Songmene, V. and Au, Y.H.J., “The use of acoustic emission information to distinguish betweendry and lubricated rolling element bearings in low-speed rotating machines”, The International Journal of Ad-vanced Manufacturing Technology, Vol. 69, Nos. 9–12, pp. 2679–2689, 2013.

17. Tonphong, K., Bearing Condition Monitoring Using Acoustic Emission and Vibration, Ph.D. Thesis, BrunelUniversity, UK, September 2002.

18. Swindlehurst, W., “Acoustic emission – 1. Introduction”, Non-Destructive Testing, Vol. 6, No. 3, pp. 152–158,1973.

19. Eissa, S., Condition Monitoring of Pharmaceutical Powder Compression During Tabletting Using AcousticEmission and Vibration, Ph.D. Thesis, Brunel University, UK, March 2003.

20. Niknam, S.A., Bearing Condition Monitoring Using Acousic Emission, M.Sc. Thesis, Brunel University, UK,September 2008.

1114 Transactions of the Canadian Society for Mechanical Engineering, Vol. 37, No. 4, 2013