Surge testing electric motors - cooling the debate (Part 2)

Mike Teska - Engineering Manager

In this, the second and final part of our article that seeks to settle once and for all the debate about whether surge testing for motors is beneficial or harmful, we look at how surge tests should be performed, the ways in which motors fail, and the benefits that modern techniques of surge testing can provide. The first part of the article, which discussed the principles and history of surge testing, is still available online at the Megger website.

A Surge Test Done Properly 

If one is to benefit from surge test results, several things must be understood. First, the operating voltage must be known for the motor under test. Second, the person conducting the test needs to understand the surge test procedure and the appropriate voltage to apply to the motor. Several industrial organizations have written standards regarding the application of the surge test. Table 1 below lists recommended test voltages for a number of types of motors and covers both new and in-service machines. 

When determining a proper test voltage, it must be understood that there is a minimum voltage that must be applied across a gap before there is any possibility of an arc forming in a gaseous dielectric. This is called the Paschen’s Minimum. Figure 1, called Paschen’s Curve, shows the breakdown voltage versus air spacing at standard temperature and pressure for a uniform field. The minimum voltage is approximately 350 volts. This means it takes 350 volts to excite an arc across a fault. Below this minimum voltage, no reliable findings will be made. 

Figure 1

How a Motor Fails 

The motor stator has two main insulating systems: the ground wall insulation and the turn-to-turn insulation. When the insulation is in a good condition, it can withstand the normal day-to-day voltage spikes that occur during starting and stopping. Over time, this insulation will deteriorate due to mechanical movement of the windings, torque transients, heat, contamination and other environmental influences. 

In 1987, Gupta, Lloyd, Stone, Campbell, Sharma and Nilsson published an article in IEEE Transactions on Energy Conversion on a three-year study of motors in power plants. Their conclusion was that motors see approximately 3 p.u. (per unit) surges in normal operation during startup. Some motors may see as high at 4.6 p.u. The formula to calculate the per-unit voltage is: 

Figure 2 displays the effects that starting and stopping has on electric motors by showing the relationship between dielectric strength and daily operations of an electric motor. When the motor is new, the voltage spikes at startup do not seriously affect the dielectric strength of the insulation. Over time, as the insulation weakens, the voltage spikes begin to cause problems in the turn-to-turn insulation of the motor. When the dielectric strength is reduced to the level of operating voltage, the spikes cause the insulation deterioration to accelerate exponentially. 

Figure 2

As the insulation deteriorates, voltage spikes contribute to development of arcing faults. If a motor is operating with severely weakened insulation, a failure of the turn insulation quickly results in catastrophic failure of the motor and unplanned shutdown of the process the motor is driving. Knowledge of a motor’s turn-to-turn viability is therefore crucial. With periodic maintenance and surge testing, problems of weakened insulation can be identified and tracked, in most cases with time to spare. This allows the plant maintenance team to act before the motor fails. 

Writing for IEEE in “Transient Model for Induction Machines with Stator Winding Turn Faults,” authors Rangarajan Tallam, Thomas Habetler and Ronald Harley state that “a turn fault in the stator winding of an induction machine causes a large circulating current to flow in the shorted turns, of the order of twice the blocked rotor current. If left undetected, turn faults can propagate, leading to phase-ground or phase-phase faults. Ground current flow results in irreversible damage to the core and the machine might have to be removed from service. Detection of incipient turn faults is essential to avoid hazardous operating conditions and reduce down time.” 

Benefiting From Modern Surge Testing 

Since 1961, Baker Instrument Company has manufactured high-voltage test equipment. The basic concept of the surge test has been enhanced to further benefit the user. The surge comparison test that was developed just prior to 1950 still is a viable test today; however, there have been advances in waveform analysis and detection that change the manner in which this test is used. 

If the problems being analyzed are shorts, opens, different-diameter copper between phases, unbalanced turn count between phases, reversed coils or shorted laminations, the comparison test is quite effective. Baker Instrument Company has developed the Line-Line Error Area Ratio (L-L EAR) to automatically detect these issues. This is a comparison of each phase, 1-2, 2-3, and 3-1, at the recommended test voltages. The area under the curve is mathematically calculated and compared to the other windings to see if any of the phases are out of tolerance. If they are, the tester will automatically stop testing and notify the user that a problem exists. 

More advanced than L-L EAR testing is the Pulse- Pulse Error Area Ratio testing. With the P-P EAR, Baker Instrument Company has refined the analysis of the waveform to the point where the test instrument is precise and sophisticated enough to recognize failures. The P-P EAR calculates the arc by comparing the curves for each successive pulse as the voltage is increased to the IEEE-recommended testing levels. If the frequency of the waveform moves more than the allowable tolerance, the unit will again automatically stop the test. 

Figure 3

To illustrate the power of the P-P EAR, a surge summary from a motor at a pulp and paper manufacturer is displayed in Figure 3. This motor failed on the second lead at 4,760 volts. The target test voltage was set at 5,600 volts. Further examination of the PP-EAR% dialog box clearly shows a jump in the waveform, well above the pre-set 10% threshold. At this point, the tester automatically stopped testing and notified the equipment user that a problem had been found. 

In this case, along with the surge test, a full battery of DC tests were performed, including resistance, insulation resistance, polarization index and HiPot. Of all the tests performed, only the surge test revealed potential insulation weaknesses (Figure 4). This is mainly because the other tests focus strictly on groundwall and not turn-to-turn insulation. It has been stated in studies that upwards of 80% of all electric motor failures begin with weak turn-to-turn insulation. 

Figure 4

After failing the surge test, this motor was immediately put back into service and ran for the remaining four months until the next scheduled outage. This saved the company more than $40,000 in unscheduled downtime. This case study confirms that the surge test is non-destructive to motors. As stated by Westinghouse engineers in 1951, “Because the energy of the surge is extremely limited, the current through the faulty insulation is so small that no severe burning occurs at the point of weakness.” 

To further support this statement, in 2001, a 90-day study was performed at Baker Instrument Company on a 5 HP, 460 Vac, 1,200 rpm motor produced in China. This motor was chosen because it was the cheapest motor available. The motor was subjected to approximately 40 million surge impulses over the 90 days. The first 20 million pulses were applied at steadily increasing voltage levels, and 17 million of the pulses applied to the motor were 50% to 350% greater in amplitude than the 2,000 volts suggested by Baker Instrument Company. 

After each phase winding was subjected to approximately one million impulses at the prescribed voltage, the voltage was increased by 1,000 volts. In other words, the goal was to apply three million pulses at each voltage level – 2,000, 3,000, 4,000, 5,000, etc. Upon reaching the dielectric limit of this motor’s insulation (which was 7,000 volts), the dielectric limit of the remaining insulation was found to be 1,750 volts. The motor was then subjected to an additional 20 million impulses at 1,750 volts. All these pulses were applied to phase one, as it was where the weakness developed (Figure 5). At the end of this sequence, the dielectric limit of phase one was still 1,750 volts – no further deterioration occurred within the insulation. It is important to note that, again, the prescribed test voltage for the motor was 2,000 volts, and even though the insulation was performing at a lower level than acceptable, the motor would still run normally for an indefinite period. 

Figure 5

At this point, it was decided to see how much abuse an electric motor can handle prior to failing. The motor was wired to 460 Vac and rapidly cycled until an observable failure occurred. In this case, the motor was started and rapidly stopped 42 times before suffering turn-to-turn failure under these extreme operating conditions. Failure was evidenced by gross imbalance in the phase currents and an audible difference in the sound of the motor running. After the turn-to-turn failure, or hard short occurred, less than one minute elapsed before smoke was observed coming from the connection box. 

This study shows that even a very cheap motor can withstand heavy abuse and extremely high voltage (350% of the recommended test voltage). After the weakness developed, the surge test did not cause further deterioration of the motor, even after an additional 20 million pulses. It took severely abusive and uncharacteristic rapid starting and stopping of the weakened motor to precipitate failure. This again demonstrates that proper surge testing is not destructive. This is because of the high-voltage, low-current characteristics of the test. On this motor, during testing at 1,700-1,800 volts, the peak observed current was 1.8 amps. The current was present for 1-2 microseconds, 5 times per second with 200 milliseconds between pulses, which equates to an instrument duty cycle of approximately 0.001%. The power generated is extremely tiny. In layman’s terms, it’s like throwing a tennis ball against a brick wall. It simply does not have enough power to burn the insulation. 

Summary 

The predictive maintenance and motor reliability fields have debated the validity of the surge test for many years. However, the benefits of this test can stand on their own merits. It is the most efficient way to find turn-to-turn weakness within motors prior to breakdown, giving maintenance professionals the opportunity to plan the repair or replacement, which saves money. It has been proven through multiple studies and through continuous use that the surge test does not burn or damage insulation further upon finding a fault. 

Since its development, the instrumentation and the test itself have changed dramatically. In 1926, when J.L. Rylander introduced the test, equipment was heavy and large. Instrumentation now is portable and computer controlled, making it easy to use, highly accurate, and able to produce results quickly and automatically.