Predictive Maintenance Can Prevent Motor Failures
Motor Predictive Maintenance
Joel Fallbright and David R. David,
Pennsylvania Power and Light Co.
Over the Past Few Years, Two Stator Winding
failures occurred in the 3,500 hp, 13.2 kV circulating
water pump (CWP) motors at Pennsylvania Power and Light Co.`s (PP&L) nuclear plant–the Susquehanna Steam Electric Station. In response to these failures, PP&L began a research and development project with Iris Power Engineering. The program aimed to devise a method of instrumenting the remaining large motors to monitor partial discharge on-line in order to detect future incipient failures.
The first failure occurred when a motor experienced a phase-to-ground fault (Photo 1). Engineers attributed the failure to the abrasion of the high-voltage, stator coil elec trical insulation caused by coil vibration in the stator slot. Upon inspection, technicians found that a large number of stator slot wedges were missing.
The CWP motors are not of the global vacuum pressure im pregnation type; they have magnetic wedges that are supposed to reduce stator slot groundwall potential. However, the wedges tend to become brittle and deteriorate over time, come out of the slot, and are ground to a powder by rotor action. Consequently, enough missing wedges on any stator slot may allow the bar to move, abrading against the core.
Standard electrical tests did not detect the degraded motor condition. Since on-line, partial discharge monitoring would most likely have detected this failure, PP&L had stator slot couplers (SSCs) installed in the stator slots under the wedges during a rewind (Photo 2).1 Technicians also replaced a large number of magnetic wedges with Nema epoxy glass wedges. A second motor was also sent to a vendor`s facility to be rewedged. After cleaning and rewedging the stator, this motor failed the Hi-pot testing. It was then rewound and equipped with SSCs.
Technicians rewound the remaining Unit 1 CWP motors with SSCs. Installing SSCs is expensive, however, and can only be performed during a stator rewind or rewedge. Installation of SSCs during a rewedge of motors with Versa-Pac end rings is particularly difficult, because it is not easy to determine which slots should have SSCs installed. The older-style winding in these motors is a fiberglass circular channel that is placed over the stator winding end-turns, filled with sand aggregate and heated. Its function is to bind the end-turns.
As part of a special project, PP&L had radio frequency current transformers (RFCTs) installed around the ground cables from surge capacitors that are attached to the motor leads (Photo 3). The distance between the breakers and the motors made it unfeasible to set up a method to cancel noise mixed with the partial discharge. Thus, the installation is single-ended.2 The long cable leads from the breakers to the motors, however, dissipate most of the high frequency noise before reaching the motors. Testing confirmed that the RFCT method works. As a continuation of the project, during subsequent outages, PP&L installed RFCTs on the plant`s 20 remaining 13.2 kV motors. This included eight 3,500 hp circulating water pump motors, eight 3,500 hp condensate pump motors and four 9,000 hp reactor recirculation motor-generator (MG) set drive motors.
PP&L has SSCs in place on an indirectly cooled GE unit at the company`s Sunbury 2 plant, a GE water-cooled unit at Montour 2 and a Westinghouse direct gas-cooled unit at BISES 2B. Cable capacitors have also been installed on three hydropower generators at the Holtwood Station. Through a cooperative effort between PP&L`s fossil and nuclear departments, a portable analyzer is used for all sensors installed at each station.
Rewind test success
To gauge the success of the modified equipment, PP&L recorded a partial discharge baseline for all of the 13.2 kV motors. The results were good–meaning the data were consistent, characteristic and reliable. Several motors exhibited high levels of partial discharge. Engineers collected data from these motors every two weeks to determine if there were any adverse trends. It took about 30 minutes to collect the data for each motor. Data on the remaining motors were collected and reviewed every 6 months.
Two condensate pump motors (1P102A, 1P102D) exhibited high levels of partial discharge, especially on the B phase. PP&L decided to have the 1A motor rewound during the next unit outage because its discharge levels were higher than those of the 1D motor. The highest NQN value for 1A was 937 (Figure 1), and the Qm was 413 mV. NQN is proportional to the area under the plotted pulse height analysis line. It is related to the total charge transfer due to partial discharges or the integrated charge and is similar to power factor tip-up. Qm is the peak partial discharge magnitude, a measure of a discharge pulse repetition rate of 10 pulses per second on the pulse height analysis plot. Figures 2 and 3 compare the B phase discharge readings before and after the rewind. The highest values were on the C phase. The NQN was 428 and the Qm was 230 mV. The A and B phases realized even greater reductions in partial discharge activity than the C phase.
There are reasons why the C phase did not have readings as low as the other phases. The rewind for this motor was difficult. The relatively small stator core diameter means there is significant stator slot pitch. Therefore, the coils need to be compressed to enter the slots. Compressing these “hard cell” coils and fitting them into the slot under these conditions can damage the electric field gradient coating on the coils. This operation is made more difficult by the long length of the coil.
The rewound 1A motor failed the Hi-pot test because of coil damage during installation. Enhancements to the installation process and tools made the subsequent rewind successful. However, the partial discharge data plot indicated damage to the gradient coating on the C-phase coil. The positive pulses were much higher than the negative pulses, corresponding to slot discharge or gradient coating damage.
The 1A original stator coils exhibited burn marks across numerous bars. Therefore, the magnitude of the partial discharge readings taken on this motor can be directly related to physical damage. The economic value of preventing this potential failure was calculated at $1.5 million, based on the results of a predictive maintenance calculation. This value is very conservative, since it assumes a full spare set of stator coils would be on hand and a 16-day period would be required to remove, rewind and re-install the motor.
Technicians rewound the 1D motor during a subsequent outage in the fall of 1996. Similar to the 1A machine, the removed stator coils from the 1D motor exhibited distinct signs of degradation from partial discharge activity. The techniques and lessons learned from the rewind of the 1A motor led to a successful rewind of the 1D motor, confirmed by the low levels of partial discharge recorded during the post-rewind testing.
New filter application
As previously mentioned, the installations using RFCTs at Susquehanna Steam Electric Station are single-ended and do not have a means of noise rejection, except for dissipation in the motor cables from the breaker. Noise did not seem to be a problem on Unit 1. Unit 2, however, exhibited pulses in a narrow phase band every 60 degrees. Engineers attributed these pulses to SCR (silicone control rectifier) switching in a battery charger that is unique to Unit 2. Therefore, the partial discharge data became masked because the magnitude of these pulses is on the order of many volts.
Iris Power Engineering de veloped a filter to help eliminate noise in single-ended applications where very high levels of external switching noise occur. Iris engineers visited the Susquehanna site and tested several filters to determine an appropriate cutoff frequency and to take baseline readings with the new filter. In initial tests, the new device filtered out the switching noise on Unit 2 for the 2S134B reactor recirculation MG set motor. The overall reduction in pulses per second and peak magnitude also showed that the filter could possibly eliminate some lower frequency partial discharge indications.
Since economics and time do not allow plant personnel to repair all suspect motors, on-line partial discharge analysis allows users to compare motors and gauge degradation to determine which motors pose the greatest risk of failure. The system also allows users to justify not performing repairs on equipment that does not require repairs, or on equipment where partial repairs have arrested the degradation.
As mentioned earlier, PP&L had both of the 3,500 hp circulating water pump motors rewound, but their replacement coils were produced by two different manufacturers. Post-rewind, partial discharge data on one motor remained extremely low, while the other motor had higher readings. The coil with the higher readings had been suspect in the past, and with this new information, PP&L decided to remove one vendor from its bidder list. S
Authors–Joel Fallbright has worked more than eight years in the nuclear department of Pennsylvania Power and Light Resources. He is currently a senior engineer in predictive maintenance, in charge of the motor program and infrared program at the company, as well as electrical preventive. Fallbright is an electrical engineering graduate of Pennsylvania State University.
David R. David is supervisor of electrical services for Pennsylvania Power and Light Co. He has worked in maintenance since 1989; previous assignments involved plant results, startup and testing. David received a bachelor`s degree in electrical engineering from Lehigh University.
1 Campbell, S.R., G.C. Stone, H.G. Sedding, G.S. Klempner, W. McDermid and R.G. Bussey, “Practical On-Line Partial Discharge Tests for Turbine Generators and Motors,” July 1993 (unpublished).
2 Kurtz, M., J.F. Lyles, H.G. Sedding and G.C. Stone, “On-Line Partial Discharge Measurements: A Powerful Maintenance Tool for Rotating Machines,” Ontario Hydro Research Review, Number 6, June 1992, pp. 18-31.
(Photo 1) Circulating water pump motor at Pennsylvania Power and Light Co.`s Susquehanna Steam Electric Station.
(Photo 2) Stator slot couplers installed in the stator slots under the wedges during rewind.
(Photo 3) Radio frequency current transformers installed around ground cables from surge capacitors that are attached to the motor leads.