Industry competitive pressures have forced utilities to optimize operation and reduce maintenance costs, in other words, do more with fewer resources. In 1993, utilities made a major effort to re-evaluate preventive-maintenance activities based on extensive personnel experience, and electrical and diagnostic testing methods available at that time.

Most utilities use a time-based preventive approach to substation maintenance. This approach dictates maintenance scheduling strictly related to time in service rather than equipment condition. Frequently, funds are misdirected for maintenance performed on equipment that does not require maintenance. This misdirection of funds created the need for an online condition assessment of power transformers and associated substation equipment.

The EPRI M&D Center, working in cooperation with predictive-maintenance (PDM)-member utilities, implemented a PDM program that incorporated the following technologies:

• Vibration Analysis
• Ultrasonic Noise Analysis
• Partial-Discharge Detection (PDD)
• Portable Gas in Oil Analysis
• Infrared Thermography
• Sound Level Testing
• Ultrasonic Bearing Wear Detection.

Project Overview

Researchers developed a formal plan, which included the newly proposed preventive-maintenance activities and frequencies combined with PDM condition-monitoring technologies. This maintenance optimization effort included a PDM program development and implementation project containing the following tasks:

• The evaluation and enhancement of known online periodic equipment condition-monitoring tools.

• The demonstration and evaluation of various new research technologies.

• The enhancement of the programmatic aspects of the PDM program.

• Procedure development.

• On-site training for data collection and reduction associated with new technologies.

• Documentation of cost-benefit analysis.

Partial-Discharge Detection

As part of the Substation Predictive Maintenance (SPDM) program, researchers initially performed PDD on transformers, load tap changers and circuit breakers using typical “off-the-shelf” PDD equipment. This type of detector counts activity at approximately 150 KHz. Although this technique is useful for quantifying activity levels, most transformers do not have levels of internal activity high enough to trigger the detectors. The difficulty then is that no data is saved for trending and, therefore, no judgment of transformer condition can be made over time.

Assuming that internal arcing might be low-level random activity containing multiple frequencies, users decided to acquire data in two different frequency bands and listen to the data in addition to logging overall root mean square (RMS) levels. Triple 5 Industries (Yardville, New Jersey, U.S.) developed a sonic/ultrasonic fault detector that measures RMS levels in both frequency bands. Applying this technique provides the ability to document the detection of low-level random events. Off-the-shelf PDD equipment detected this activity, even at very low levels. However, the events were usually less than 100 counts and difficult to detect in any mode except peak-hold because of its randomness. With this in mind, Alliant now uses the PDD (150 KHz) equipment for quantifying levels. Other equipment (Triple 5 Model 5550FD) has been successful in identifying partial-discharge activity. Combining the usage of these devices yields Ultrasonic Noise Analysis/PDD.

Partial-discharge produces acoustic signals that are detectable with an acoustic sensor held against the transformer wall with the transformer in service. The acoustic emission emanating from the partial-discharge sources within a power transformer display a burst-type signal (Fig. 2) with a characteristic frequency of 150 KHz. The amplitude of the stress wave of the partial discharge, which excites the sensor, determines the number of oscillations contained within each burst. The number of oscillations occurring within a one-second interval contains information relative to the number and amplitude of the discharges. Keep in mind the oil of the transformer attenuates the amplitude of the stress wave. The counting technique for these signals is the method used to determine the approximate location of a partial-discharge source as well as its severity.

Transformer Problems

A significant find was on a 138/69-kV 93.3 MVA power transformer. The location of this transformer in the transmission system made this discovery a cost benefit. It is a normal procedure for Alliant PDM technicians to listen to the on-load tap changers while changing its position (stepping). The technicians hear the arcing as the tap changer goes through the step. After completing the step, there should be no arcing. In this case, the arcing continued at a frequent rate and a high level. When opening the tap changer, crews discovered a crack in the bridge casting supporting the drive mechanism. This allowed the gears to slip and the mechanism to get out of time on one phase. This is a potential failure point for the transformer as the steps do not complete and the arcing continues (Fig. 1).

Alliant crews performed a survey on another 138/69-kV 33 MVA power transformer with the Triple 5 fault detector. The instrument detected a partial discharge below the X1 bushing at about the 5-ft (1.5-m) level. The discharge was intermittent (about 2 to 3 minutes apart) and at a high level.

The crews decided to see if the problem was load dependent. While increasing the load, the level and frequency of the partial discharge responded in a linear fashion. Dissolved gas analysis (DGA) tests determined the concentration of acetylene gas was increasing. Crews took the transformer out of service and performed a degasification process on the transformer oil. The transformer was then put back in service with load constraints and monitored by DGA. This transformer remained in service for three more years, allowing Alliant Energy to plan, budget and engineer the substation upgrade.

A DGA performed on a 69/13.09-kV 11.3 MVA transformer showed some increase in combustible gases. A profiling of the transformer with the fault detector revealed some loud corona activity near the top of the tank at the no-load tap changer (NLTC) location. After opening the transformer tank manhole, the crew discovered a black, crusty deposit on the movable and stationary contacts of the NLTC (Figs. 3 and 4). An analysis revealed the deposits to be a sulfur contamination on the copper parts in the transformer. Knowing at least one identical transformer failed because of the NLTC, all the similar transformers within the company were checked. All the transformers had the same problem.

Other Equipment Problems

The monitoring of lightning arrestors with the 5550FD was discovered by chance. While approaching a 138/69-kV transformer for an annual profile survey, an ultrasonic noise was heard in the headphones and seen on the level meter of the fault detector. This noise was heard on the radiators located on the 138-kV (high) side of the transformer, as well as on the transformer side of the oil circuit breaker feeding the transformer. Careful investigation determined the noise was generated externally from the transformer. The bottom of the H2 arrestor on the 138-kV high side was the source of the noise.

The transformer was taken out of service, and the lightning arrestors were power factor tested. The results showed the suspect arrestor had significantly higher current and watt readings than the other two identical arrestors. Replacing the lightning arrestor eliminated the ultrasonic noise.

Alliant Energy has found several more similar problems with lightning arrestors using this method. The real advantages of this method are the online testing of the arrestors at their operating voltage and the ability to detect partial discharging at an early stage before failure mode.

Another practical and proven use for the fault detector is the online monitoring of instrument transformers. On one occasion, the inspector listened to three single phase-to-ground 69-kV potential transformers (PTs). Two of the PTs had ultrasonic counts of 28, which is the base line number seen on a normal operating piece of equipment. The third PT had counts of 1000 and higher. Also, the inspector could hear the partial discharge in the headphones at a constant rate.

The positioning of the PTs created a situation such that a total bus outage would occur if the PT failed. Alliant crews took the PT out of service immediately and changed it out within a couple of hours without an outage. An oil sample revealed very high levels of combustible gases, particularly Acetylene at a level over 900 ppm. The early detection of the partial discharge prevented a potentially hazardous situation and prolonged outage for an entire community.

It is important to give your maintenance program a really good look. Ask yourself these questions:

• Are you comfortable with your current substation reliability?

• Have you had any recent equipment failures?

• Is there some way you could have predicted the probability of these failures?

• Would you like to know where the most “at risk” equipment is on your system?

Mark N. Theyerl has been with Alliant Energy (formerly Wisconsin Power & Light) for 18 years, working in substation construction, maintenance and testing for the last 16 years. Theyerl began working in the field of substation predictive-preventive maintenance in 1996. He is an AVO-certified substation maintenance specialist and infrared thermographer.