Before any company investigates electrical predictive maintenance (PdM) instrumentation, it should know the strengths of its equipment's insulation, the voltages its motors are exposed to daily, how a motor typically fails and where these faults typically exist. Only then can you really make a decision as to which electrical PdM equipment is the most appropriate for your operations.
How a motor typically fails
The motor stator has two main insulating systems that include the ground wall and turn-to-turn insulation. When this insulation is in a good condition it can withstand the normal day-to-day voltage spikes that exist during starting and stopping. Over time, this insulation will deteriorate as a result of mechanical movement of the windings, torque transients, heat, contamination, and other environmental contaminates. Once the dielectric strength of this insulation falls below the incoming voltage spikes, another failure mechanism is introduced: ozone.
Ozone is a very corrosive gas that will quickly deteriorate insulation. Although the motor will continue to run when this failure mechanism is introduced, as it sees continual voltage spikes, the deterioration rate will accelerate. Eventually, the dielectric strength of the insulation will fall below operating voltage or deteriorate to the point that copper wire will touch turn-to-turn. At this point a turn-to-turn or hard welded short has developed.
According to "Transient Model for Induction Machines with Stator Winding Turn Faults" written for IEEE by Rangarajan M. Tallam, Tom G. Habetler and Ronald G. Harley, when a hard welded turn-to-turn short develops, the shorted windings will develop high circulating currents. These currents, which can be in the order of 16–20 times full-load amps, create excessive heat that the insulation cannot withstand. This intense amount of heat will burn quickly through the insulation-causing motor failure within minutes.
A study performed at Oregon State University, by Dr. Ernesto Wiedenbrug, looked at a motor specially designed with a turn-to-turn fault by installing two wires connected to turn one and turn two of the same phase. These wires were then brought out to a switch. The motor was placed on a dynamometer and run at about 80% load. When the turn-to-turn short was engaged through the switch, the motor began visibly smoking within 45 seconds. While most motors will not run for long with a turn-to-turn short, some exceptions do exist. A motor with a high resistance or floating ground will run with a shorted phase, but once a second phase shorts, the motor will fail catastrophically.
Recommended tests The tests listed on the next page are recommended in off-line field testing:
Each of these test methods evaluates a different section of the motor. Brief descriptions of the first three tests are given in order to offer a complete array of testing information. The nature of high-voltage testing and the necessity of the Step-Voltage and Surge methods, however, remain the main focus of this article.
Kelvin Method Winding...
The Kelvin Method Winding test measures the resistance of the copper wire of the motor circuit. If tested in a PdM application, the test is typically performed from the Motor Control Center (MCC). This test finds issues with miss connections, shorts, opens, unbalanced turn count in one phase to another and different size diameter copper in one phase to another. This test is very valuable and should be performed for predictive maintenance, troubleshooting and quality assurance.
The Meg-Ohm Test applies a DC potential (typically operating voltage) to the windings while holding the case to ground. Table I shows the recommended test voltages for different voltage class motors. Meg-Ohm testing is typically utilized to find grounded motors. It also is a very valuable PdM tool for finding wet and dirty motors. It's not typically used for quality assurance because of the low voltage level at which the test is performed.
Polarization Index (PI) Test...
This test is much like the Meg-Ohm Test, but it is performed for 10 minutes. Over this time period, the molecules in the slot liner paper polarize. When the molecules polarize, the insulation resistance values should increase over the10- minute period. If the resistance increases during this time, it's an indication of good ground wall insulation with no moisture or contamination.
Until now we have only discussed the low-voltage tests. Upon successfully completing these tests the following is known: the winding resistance is balanced. That means the motor has no shorts, opens or miss connections and the Meg-Ohm and PI indicate that the motor is both clean and dry. These tests, however, still have not confirmed that the motor is capable of starting or running for any length of time. The main reason for performing predictive maintenance on a motor is to learn if it will continue to provide uninterrupted service. Because low-voltage testing is not performed at the voltage a motor typically sees, it can't provide this information.
Many articles have discussed the voltage spikes motors see during starting and stopping. As stated in "Turn Insulation Capability of Large AC Motors, Part I – Surge Monitoring," by B.K. Gupta, B.A. Lloyd, G.C. Stone, and S.R Campbell (IEEE Transactions on Energy Conversion, Vol. EC-2, No. 4, December 1987), these voltage spikes can be in the order of 5 PU (Per Unit):
Calculating this formula for a 480V three phase motor, the PU would be 391.9 volts, or approximately 1960 volts on startup. Logically, if the motor is tested to only operating voltage or below the operating voltage, the user can not be sure if the spikes have caused damage to the motor's insulation that will interrupt service. The other issue is that the turn-to-turn insulation has not been evaluated. In addition, the Meg-Ohm and PI do not evaluate the ground wall insulation for strength or the ability to withstand the high voltages it sees during daily operation. The winding resistance test is only evaluating the motor circuit and not the insulation.
The most effective way to ensure the motor will start and continue to provide reliable service is to test it at the voltages the motor sees during normal operation-which includes starting and stopping. This is accomplished with two tests: Step-Voltage and Surge. These methods evaluate the ground wall and turn-to-turn insulation respectively.
This DC Test is performed to a voltage that a motor typically sees during starting and stopping. The test voltages, governed by IEEE, are reflected in Table II.
The DC voltage is applied to all three phases of the winding and raised slowly to a preprogrammed voltage step level and held for a predetermined time period. It is then raised to the next voltage step and held for the appropriate time period. This process continues until the target test voltage is reached. Typical steps for a 4160V motor are 1000-volt increments, holding at minute intervals. For motors less than 4160V, the step voltages should be 500 volts (see Fig. 1).
Data is logged at the end of each step. This is to ensure the capacitive charge and polarization current is removed and that only real leakage current remains, thus providing a true indication of the ground wall insulation condition. If, at this point, the leakage current (IμA) doubles, insulation weaknesses are indicated and the test should be stopped. If the leakage current (IμA) rises consistently less than double, the motor insulation is in good standing.
The Step-Voltage Test is necessary to ensure that the ground wall insulation and cable can withstand the normal day-to-day voltage spikes the motor typically sees during operation. If a DC Step-Voltage Test is not performed, the operator cannot be assured that the motor will start and operate without failing in service.
The Surge Test is highly important. That's because 80% of all electrical failures in the stator begin at weak insulation turn-to-turn. These types of catastrophic failures are why NFPA 70 B recommends that Surge and HiPot testing be performed. Regardless of an individual's personal view of Surge testing, knowing that a motor's turn-to-turn insulation is sound is crucial for safety and motor reliability.
During a Surge Test, the equipment will charge up a capacitor inside the unit and dissipate it into one phase while holding the other two phases to ground. Then, automatically, the test unit will slowly increase the voltage from 0 volts to the target test voltage. This generates a waveform, in a shape based upon the inductance of the coil that is displayed on the test equipment screen. If the target test voltage is attained without any frequency change in the waveform, the turn-to-turn insulation integrity has been realized. Fig. 2 is a graphical representation of the waveform at one-third, two-thirds and full voltage of one phase. This is what a waveform will look like when the insulation is in a good condition.
If, at any time, the test equipment sees weak insulation between the turns, the waveform will shift to the left as shown in Fig. 3. The white line on the graph shows the failed waveform at about 1000 volts.
Surge testing theory
When the capacitor is discharged into the winding, it is performed at a very fast rise time (.1 micro second). This produces a nonlinear voltage drop across the turns, producing a potential difference between the turns in succession. As the rise time slows, the operator will notice that the voltage potential difference between the turns is dramatically reduced. This is in contrast to any other signal utilized to diagnose motor issues. No DC test (or AC tests such as an inductance, capacitance, impedance, phase angle or HiPot) will produce this potential difference between the turns.
Physics provides us with Paschen's Law, which states that two bare wires placed next to one another just a thickness of a hair away need a minimum of 325 volts to jump the air gap between the two conductors. These two concepts are the core reason why Surge testing is the natural choice for testing turn-to-turn insulation. The main reason is that if the test equipment doesn't produce a potential difference between the turns above Paschen's Law, the current cannot flow through the fault. If current can't flow through the fault, it will continue through all the coils and not show a difference.
When Surge Testing a coil with weak insulation turn-to-turn, the voltage applied can jump across the weak insulation. Removing these bypassed turns from the circuit reduces the inductance of the circuit and causes the waveform frequency to ring faster. This will produce the frequency shift to the left in the waveform. Fortunately, advancements in technology have led to refinements in the analysis of waveforms, to the point that some test units automatically recognize failures (see sidebar).
Surge comparison In the past the Surge Test has been called a "Surge Comparison Test." Although some individuals believe the Surge Test still needs to be performed in this manner, it really depends on what is being analyzed.
For finding weak insulation, surge comparison is not necessary. As previously noted, weak insulation is diagnosed by a frequency shift to the left and is compared to successive waveforms within one phase. If, however, the following list reflects problems you're seeking to uncover and eliminate, a comparison of each phase is recommended.
Here again, as referenced in the accompanying sidebar, instrumentation that automatically detects these problems is now available.
Older vs. newer equipment
Just like computers, high-voltage test equipment has changed vastly over the past 20 years.
Today's equipment incorporates modern, high-speed electronic evaluation of changes to resistance, leakage current, leakage current versus time, voltage, step-voltage, dielectricabsorption, frequency response, wave shape, corona inception voltage (C.I.V.) and more to detect faults at or under the levels of energy exposed to the motor during operation. Microprocessor- controlled instantaneous trips allow winding conditions to be evaluated without compromising dielectric integrity. Moreover, the addition of field-developed PASS/FAIL test criteria now makes this testing extremely repeatable.
One of the greatest advances in high-voltage testing has come from via solid-state, highvoltage power supplies replacing the heavy step-up transformer. This has resulted in big improvements to equipment portability. Every test is now digitized and compared to the previously applied pulse. If any weakness is detected, the test is instantaneously stopped, preserving dielectric. The level of weakness is stored for future reference, in the memory bank.
What to look for
When evaluating electrical PdM equipment, keep in mind that every manufacturer is slightly different. Test units, though, should be able to perform the following safety checks to ensure that your motors aren't damaged during testing:
Case study: Step-Voltage testing
Exelon Nuclear, Limerick Station…
The Station Predictive Maintenance program at Limerick routinely performs electrical testing of large motors at a two-year frequency. This testing consists of winding resistance, insulation resistance, PI Capacitance/dissipation factor and DC step-voltage testing to 20kV. The resulting data has been tracked and trended for almost 20 years.
On a few occasions during 2002, Operations personnel reported that an "acrid" odor was present at the 1C Circulating Water Pump Motor. The PdM group had been tracking this motor on a "watch" list that came about as a result of an increasing trend in leakage current detected by DC step-voltage testing from 1997 to 2002 (see Fig. 4).
As part of its increased troubleshooting activities, the Limerick Station PdM team monitored the motor through the summer of 2002, utilizing acoustic monitoring and vibration and winding temperature/RTD monitoring on a monthly basis. In September 2002, an action request was made to replace the motor in the winter, based upon the electrical testing results, increasing vibration at stator slot frequencies and higher acoustic/ultrasonic "noise."
Once the motor was removed, it showed high leakage current on the "A" phase motor winding compared to the other two windings. After cleaning, a visual inspection of the winding identified partial discharge at the junction where the core slot winding tap transitions to the end winding/knuckle tape. Investigation revealed a lack of "proper" corona suppression tape at this critical junction point in the winding.
Among the lessons learned from this event was the fact that tracking and trending leakage current versus applied voltage on a DC Step- Voltage Test, as presented by the Baker AWA offline tester, can and does indicate potential problems in the winding. Furthermore, when this data is combined with other predictive technologies, it will allow for proactive replacement of a motor prior to an in-service failure.
Case study: Surge testing
Pulp & Paper Operation…
A 2300V form wound motor at a pulp and paper plant was found to have weak turn-to-turn insulation. Of all the tests performed on this motor, the only one that found the turn-to-turn weakness was the Surge Test. The controversy around surge testing, though, is that after finding a problem with insulation, could the tester have so degraded the motor that it would not run?
This Pulp & Paper industry case study easily puts this myth to rest. The motor in question was immediately put back in service after testing. It was started up and ran for the four months required until it could be shut down and removed for repair. Again, as noted in Fig. 5, the Surge Test was the only method to identify the insulation weakness. The problem was well above line voltage, so other lowvoltage tests would not have approached this threshold. (The surge summary in Fig. 5 highlights the fault weaknesses found with the tester.)
This particular Pulp & Paper site motor takes about 6-7 hours to change. Thus, it could have cost about $42,000 in downtime had the Surge Test not found the problem. Interestingly, 80% of all electrical motor failures begin with weak insulation turn-to-turn. The Surge Test is clearly the best method available to find this problem. That's why it is so important to perform this type of non-destructive testing on all motors.
The Step-Voltage and Surge Tests are necessary for an effective PdM program. They identify problems that low-voltage tests can't find.
As the case studies in this article have shown, both of these tests are non-destructive in that the tested units were returned to service until the next available time could be scheduled to replace them.
Finally, these tests are performed at voltage levels a motor is exposed to during normal operation. If a motor cannot pass the Step-Voltage and Surge Tests, you can bank on the fact that it is approaching the end of its service life. Consequently, provisions should be made as soon as feasibly possible to have that motor removed before unscheduled downtime occurs.