Online Condition Monitoring of Induction Motors through Signal Processing

Online Condition Monitoring of Induction Motors through Signal Processing S. H. Chetwani, M. K. Shah & M. Ramamoorty Electrical Research and Development Association ERDA Road, GIDC, Makarpura, Vadodara-10, India Abstract- Induction motor is critical component in many power plants & industrial processes and is frequently integrated in commercially available equipment. Safety, reliability, efficiency and performance are some of the major concerns of induction motor applications. Due to high reliability requirements, and cost of breakdown, the issue of condition monitoring of induction motors and diagnosis is of increasing importance. For these reasons, there has been a continually increasing interest and investigations into the fault detection and diagnosis of induction motors. This paper describes the utility of online monitoring technique for detection of various faults that can be applied to existing motors without dismantling or shut down. The technique presented here is based on the monitoring of the current when the machine is normally operated and analyzing the same in frequency domain for detection of the faults. The technique can detect online the presence of various faults such as broken bar in the rotor cage of induction motor, Bearing faults, Eccentricity faults and stator turn to turn short, by monitoring and analyzing the line current. These different faults within induction motor are simulated in the laboratory and they are detected by online monitoring of current and analyzing same in frequency domain. This technique was also applied for condition monitoring of motors at a nuclear power station. The technique can be used as a diagnostic tool for condition monitoring of motors. I. INTRODUCTION Motors are critical components for electrical utilities and process industries. A motor failure can result in the shutdown of a generating unit or production line. The operators of motors or electrical drive systems are under continual pressure to reduce maintenance costs and prevent unscheduled downtimes, which result in lost production and financial income Early detection of fault within a motor prior to complete failure provides an opportunity for maintenance to be performed on scheduled routine without loss of production. This arises the need for condition based maintenance strategies i.e. monitoring the condition of motors and planning the maintenance based on an indication that a problem is about to occur. Condition monitoring implies monitoring various parameters of a machine in order to assess the health of the machine. Condition monitoring is akin to cardiogram analysis of the human heart. A cardiogram assesses the state and health of the human heart. Similarly the various parameters measured during the condition monitoring of the electrical equipment assess the health of the machine. Many operators now use online condition based maintenance strategies in parallel with conventional planned maintenance schemes. This has reduced unexpected failures, increased the time between planned shutdowns and reduced operational cost. During past fifteen years there has been substantial amount of research into the creation of new condition monitoring techniques. New methods have been developed which are now being used by the operators and research is continuing with the development of new and alternative online diagnostic techniques. This basic objective of this work is to diagnose the different types of motors faults online through processing and analysis of motor current. This technique is often called "current signature analysis" 11. CURRENT SIGNATURE ANALYSIS Signature analysis is the procedure of acquiring the motor current and voltage signals, performing signal conditioning and analyzing the derived signals to identify the various faults. Motor current acts as an excellent transducer for detecting fault in the motor. Spectrum analysis of the motor's current and voltage signals can hence detect various faults without disturbing its operation. Current signature analysis involves the measurement of electric current around any one phase either through clamp on meters or through CT's. This current is then transformed into its frequency spectra and analyzed for detection of fault. Information on the application of specific condition monitoring techniques in industry is not available and evidence of diagnosing problem via reference to actual on site case histories are also not considered. There is no clear distinction made between monitoring techniques, which are at the R & D stage in comparison to those which are being successfully applied in industry. It is a fact that the operator of induction motor requires evidence of the successful application of monitoring systems to assist him in their selection of appropriate systems. An operator must treat each induction motor drive as a unique entity and the potential failure modes, fundamental causes, mechanical load characteristics and operational conditions have all to be taken into consideration when a condition monitoring system is being selected. The focus has been on the use of online condition monitoring to detect degradation processes and failure mechanism. Fig. 1 2175

presents an overview of problems and possible online monitoring techniques....irgap MONOR MECHANICAL CU?r ECC~ThICJTY CCENMRCtTY VIBRATION PROBLEMS VIBJATION P YES ARWING #W FLUX.L M IGNMEN Fig. 1 Problems, failures and possible online monitoring Technique for Induction motors At this stage it is suffice to state that a user of critical induction motor drives should select a condition monitoring system based on evidence of its reliability to diagnose problem in industrial drive and on its applicability to the particular industrial installations. III. METHOD OF MEASUREMENT The method of measurement involves measurement of electric current around any one phase or all the three phases, if the line current is too high then measurements may be made on the secondary of current transformer. The current signal is conditioned. A/D board digitizes the signal to permit the time and frequency spectra for analysis. The time varying current signal is then transformed into frequency domain through FFT algorithm. IV. FAULTS IN INDUCTION MOTOR The most prevalent faults in Induction Motor are briefly categorized as Rotor faults Bearing faults Eccentricity faults Stator faults The surveys indicate that in general, failures in electrical machines are dominated by bearing and stator faults with rotor winding problems being less frequent. Fig. 2 shows the statistical spread in the various dominant mechanisms. Dominant failure mechanism V. ROTOR FAULTS Each individual rotor bar can be considered to form a short pitched single turn, single-phase winding. The air gap field produced by a slip frequency current flowing in a rotor bar will have a fundamental component rotating at a slip speed in the forward direction with respect to rotor and one of equal magnitude that rotates at the same speed in the backward direction. With symmetrical rotor, the backward component sums to zero. For a broken bar rotor, however the resultant is non-zero. The field, which rotates at slip frequency backward with respect to the rotor, will induce EMF on stator side that modulates the main frequency component at twice slip frequency [1],[2],[3]. The side band components around the fundamental of the line current spectrum are usually measured, detecting broken bar faults given by fb 1± 2s)f Hz The measurements on totally enclosed IHP motor with cast rotor was undertaken to verify the presence of broken bar signals predicted by the above analysis. The motor was fed with 3ph-balanced supply through measurement panel. The line current was recorded through the noninductive shunt. The signal was then given to Oscilloscope for the processing. Initially the no load current and no load losses were also calculated and then the motor was operated at rated output. The slip and efficiency of the motor were calculated. The same experiments were repeated first for motor with one broken bar, two broken bar and three broken bars. The results are tabulated in Table I. Fig. 3, 4, gives the spectral plot of current for the no broken bar, one broken bar respectively. From Table I it can be seen that with no broken bars there is appearance of lower side band LSB 1 but the magnitude is very less and the ratio of slip frequency amplitude to line frequency amplitude is around 48 Db i.e. greater than 45 Db and hence no fault exists within the rotor. With one broken bar, there is increase in the magnitude of LSB 1. The ratio also drops down to 31 indicating the fault within rotor. The constant losses and no load current of the motor also increases. There is also decrease in the efficiency of the motor. With increase in the bar failure there is increase in magnitude of LSB 1 and ratio and efficiency goes on decreasing. The slip, constant losses and no load current go on increasing. VI. BEARING FAULTS Fig 2 Different faults within Induction motors *Bearing EStator DRotor MEccentricity and others The relationship of the bearing vibration to the stator current spectrum can be determined by remembering that any air gap eccentricity produces anomalies in the air gap flux density. As with the air gap eccentricities, these variations generate stator current at predictable frequencies [6]. Fbng, is related to the vibration and electrical supply frequencies by 2176

fbng [fe ± mf v ], where m = 1,2,3... and f, is one of the characteristics vibration frequencies calculated based on bearing dimensions. The characteristic frequencies for ball bearings are based upon dimensions and given by fbrg =f± nbf i,,,with fi,o = 2 frf1± RD cos f];where 2L PD j nb is the number of balls, fr the mechanical rotor speed in Hz, PD the bearing pitch diameter, BD the ball diameter, ) dat ccntcmn The Bearing fault was simulated by replacing the driving end bearing with the degraded bearing of same size and number. The fault frequencies were predicted as given in Table II and These frequencies were visible in the frequency spectrum. VII. ECCENTRICITY FAULTS The use of current monitoring for detection of air gap eccentricity to identify the frequency component in the current spectrum [4], [5] is given by f[ec = fl (R ± nd n] where R = no. of rotor bars. P= pole pairs, s= slip nw = 1,3,5..., f, = supply frequency, nd = dynamic eccentricity =1 The eccentricity fault was simulated by inserting the 0.1 mm copper strip in the end shield of driving end. This will create the static eccentricity in the motor. The motor will be allowed to run at rated speed to create the dynamic eccentricity. The fault frequency related to this fault were predicted as given in Table III. These frequencies were visible in the frequency spectrum. Fig. 5 & 6 shows the current spectrum without and with eccentricity fault VIII. STATOR FAULTS Stator winding failures are also a major problem in low and medium voltage induction motors. It should be noted that volume of low voltage motors is much greater than high voltage machines. In motors it is normally the case that insulation degradation cannot be initially diagnosed via on line measurements and the first indication of a problem will be that a fault actual develops. It is important to appreciate that there is clear distinction between insulation degradation prior to a fault and an actual fault. Stator winding faults can be classified as follows 1. Turn to turn short within coil - motor will continue to operate but for how long? 2. Short between coils of the same phase - motor can continue to operate but for how long? 3. Phase to phase short - motor fails and protection equipment disconnects the supply 4. Phase to earth short - Motor fails and protection equipment disconnects the supply 5. Open circuit in on phase (Single Phasing) - motor may continue to operate depending on the load condition and protection equipment. Pre warning of serious problem (3 and 4 above) can only be achieved if shorted turns within coil ( one or two shorted turns) can be initially diagnosed via online diagnostic techniques. This requires continuous online monitoring to diagnose the faults state in 1 and 2 above. There is also the question of how long does it take for shorted turns within coil to develop into phase to phase or phase to earth fault and motor failure? This question has not been resolved and will be function of many variables and will be unique to each motor. Some operators and manufacturers have previously considered that it is not worth diagnosing shorted turns or coils in stator windings since the lead time to failure is too short to merit a continuous online diagnostic system. The concept that the motor has already developed a fault and will need to be repaired has prevailed. This philosophy is generally now considered to be somewhat out dated and defeatist. In modern production process any lead time can be extremely advantageous since unexpected failure of a drive can be very costly and in some industries it can also be a serious safety hazard. If shorted turns in a stator coil can be diagnosed a preplanned shut down can be arranged fo rthe motor to be replaced by healthy one and the faulty one sent for repair With respect to the problem of single phasing it is relatively easy to diagnose the problem provided the correct protection equipment is used to cater for all load conditions. It is also possible to monitor and analyze signals such as current to diagnose single phasing under any load operating condition. The stator winding itself is used as the sensor for the detection of abnormalities in the windings. The harmonics which are expected to vary and which have their origin in the stator currents [8], [9] are given as fsc = fl (jrtr s j± 2 jsa ± ist The stator fault i.e interturn short circuit in the windings was simulated by the rewinding the 1 hp, 4 pole motor. The winding of the motor was rewinded and the tapings at the end of the coils and the tappings at 1,3,5, 20 and 25 turns were taken out on the terminal box. The interturn short circuit was created by shorting the turns through rheostat to limit the short circuit current to rated current of the winding. Fig. 7 shows the arrangement for simulating the stator fault. The fault frequencies related to stator fault were predicted and obtained in the spectrum It was found that for the 2 turn interturn short circuit, frequency component 398 was most sensitive to fault. From no fault to maximum fault current variations in the range of 49% was recorded. Components 498 Hz were found to be less sensitive. The component 198 and 298 2177

Hz was decreasing in all the phases. But decreasing being more pronounced in the faulty phase. For 4 turn interturn short circuit, frequency component 398 Hz also increased from no fault conditions. The variations of more than 100% was observed from no fault to maximum fault current. For 20 turn interturn short circuit, frequency component 398Hz also increased. The variation of more than 400% was observed from no fault to maximum fault current. From the experiments it was apparent that some of the specified frequency components were redundant, but whether or not this would be the case for other machines. Fig. 8, 9 shows the variations of different frequency components for 2 turn, 4 turn and 20 turn interturn shortciruit. CO)NCLUSION Online current monitoring and diagnostic can be applied for the detection of various internal faults within the machine. Laboratory trails have verified that these problems can be reliably detected in industrial motor drives. TABLE I SUMMARY OF RESULTS FOR ROTOR FAULTS Sr. Particulars W/o One Two Three no. fault bar bar bar failure failure failure 1 No load current, A 0.813 0.916 0.916 0.94 2 Constant losses, w~ 129.5 160.5 170.3 178.5 3 Load current, A 1.71 1.7 1.91 2.06 4 Speed, rpm 2825 2825 2797 2758 5 LSB1, % 0.36 2.7 3.3 4.36 6 Motor current slip 48.8 31.2 29.5 27.2 ratio amplitude, db 7 Starting current, A 10.4 10.3 12.4 17 8. Efficiency, %o 73.4 71.24 68.6 66.4 0 8 30 0 ACKNOWLEDGMENT The investigators are thankful to ERDA management for allowing to undertake the study on the Online Condition Monitoring of Induction motor and permitting to carry out the experiment based on the study Fig. 3 Current Signature with no bar failure REFERENCES [1] G.B. Kliman and R.A. Koegl, "Non invasive Detection of Brokeni bars in operating Iniductioni motors," IEEE Trans. on EC vol.3 No.4 Dec'98, pp. 873 [2] 5. Williamsom and A.C. Smith, "Steady State Analysis of 3 phase cage motors with rotor bar and end ring fault," IEEE Proc. Vol. 129, Pt. B, no. 3 May 82, pp. 93 [3] Nagwa M. Elkasabgy, Anthony R. Easthama, "Detection of Broken bars in the cage Rotors on an Induction machine" IEEE Trans. onlia vol. 28 no. 1 Jan' 92, pp. 165 [4] W.T.Thomson, D Rankin, D.G. Dorrell, "Online Current Monitoring to Diagnose air gap eccentricity in large 3ph induction motors," IEEE Trans. on EC vol. 14 no.4 Dec'99. [5] Alberto Bellini, Fiorenzo Filippetti, Giovanni Franceschini, Carla Tassoni, "On field experience with online Diganosis of Large Induction motors cage failures using MCSA," IEEE Trans. on IA vol. 38, no.4 July! August 2002. [6] Randy R Schoen, Thomas G., Farrukh K., Robert G, "Motor Bearing Damage Detection using Stator current Monitoring," IEEE Trans. on IA vol. 31 no. 6 Dec 1995. [7] Randy R Schoen, Brian K, Thomas G., Jay H., Samir F, "An Unsupervised On Line System for Induction motor fault detection using stator current monitoring," IEEE Trans. on IA vol. 31 no.6 Dec 1995. [8] Mohamed El Hachemi, Michelle Vieira, "Induction Motors' Faults Detection and Localization using stator current Advanced Signal processing Techniques," IEEE Trans. on PE vol. 14 no. 1 January 1999. [9] Andreas Stavrou, Howard G. Sedding, "Current Monitoring for Detecting Interturn Short Cirucits in Induction Motors," IEEE Trans. On EC vol. 16 no. 1 March 2001. Fi.40Curn intr t n a alr -50~~~~ALEI 2 Fo ~~~~~~3 4 8 54 6 6 7 Fi.4 f+foen 134trewt oebralr 6 F+i 180 2178

TABLE III CHARACTERISTIC FREQUENCY FOR ECCENTRICITY FAULT FOR TYPICAL LABORATORY MOTOR Rotor 14 0.00533 Speed 1492 rpm Slip 3 Nw 1-1 3-3 5-5 Se 398 298 498 198 598 98 Nd 1 de 423 323 523 223 623 123 Nd l1 = = I= = de 373 273 473 173 573 73 + 2.OOE+01 + Fig. 7 Schematic Diagram of Laboratory Machine Fig. 5 Current Signature with inherent level of eccentricity 0.1 *) 0.09 A 008 U 0.07-.,0.06- Oa 0.05- al 0.04- E o 0.03-002 co 0.01 2 turn fault * 4 turn fault 20 turn fault 0 0.5 1.5 2 Fault current in Amps Fig. 8 Variation of 398 Hz frequency component with fault current -1.OOE+02 -I I I 9-1.20E+02-I 2.OOE+01 l O.OOE+00-2.OOE+01- -4.OOE+01- -2.0+1 100 200 300 400 500 6 0-6.OOE+01 L) I I -1-8.OOE+01-1.40E+02- Fig. 6 Current Signature with eccentricity fault 'a a 0) 0.16 0.14 0.12 0.1 " 0.08 0 E 0.06 () 0.04 N co 0.02 'I a0 0 *~~~-= -I 0.5 1 1.5 2 A 2 turn fault 4 turn fault 20 turn fault fault current in amps Fig. 9 Variation of 498 Hz frequency component with fault current 2179