Maintenance is necessary to keep cable networks at a desired reliability level. It should be done efficiently, both with respect to cost and to timing. Basically, maintenance can be performed in two ways: through reactive maintenance (or "wait and see") and through active maintenance (or "predictive maintenance").
"Wait and see" means that action is taken when a problem is already discovered, for instance a cable circuit failure. This philosophy can be compared with maintaining health without regularly scheduled physicals. The short-term savings are obvious, but the long-term costs are completely unpredictable and can become catastrophic.
In active maintenance, the necessary actions are taken before problems occur. The overall goal is to reduce costs. Maintenance of distribution cable circuits is mainly required to avoid in-service failures and, thereby, target replacement monies effectively: To replace those parts or components first that have a high probability offailure.
If a failure occurs, the related costs can be divided into materials, labor and customer dissatisfaction and related claims. Labor can become a substantial cost because of time-consuming fault location and necessary overtime paid for test and repair crews. Customer dissatisfaction costs are difficult to quantify at this time, but it is highly likely that such costs will become relevant in the framework of deregulation, as it is already in some European countries.
Active or predictive maintenance needs a tool to look into the future: What is the present condition of the cable circuit and what is going to happen in the next few years? Such a predictive or "diagnostic" method must be allowed to detect and to locate defects and potential failures. Therefore, an essential element of a predictive-maintenance program will be a suitable diagnostic method that will advise the cable user what parts of the system may threaten the system's life and need to be repaired or replaced. A suitable test must be one that can be performed on site, should be nondestructive and should be able to diagnose common types of cable of practical circuit lengths.
The very low frequency partial discharge (VLF PD) diagnostic testing method developed by KEMA has been successfully used in North America for about three years (Fig. 1). The cost-benefit evaluations show that predictive diagnostic maintenance testing will save the user money and improve the cable circuit reliability, which will improve customer satisfaction.
Test Locates Weak Sections The VLF PD diagnostic test method locates weak sections in medium-voltage cable circuits. These sections are detected by measuring partial discharges. Partial discharges of sufficient energy levels gradually affect the insulation material and eventually deteriorate the insulation, often resulting in an electrical failure. To prevent an electrical failure, partial discharges are needed for early detection of weak sections. Figure 2 shows the connection scheme that enables PD testing to be performed.
To be able to generate the partial discharges, a 0.1-Hz high-voltage generator is used typically at 1.5-2.0 times phase-to-ground voltage. At the point of measurement, the discharges are directed via a coupling circuit to a digitizer. The distance between the first discharge pulse and the second pulse is a measure for the location in the cable where the discharge took place. Signal recognition plays an important role here. Figure 4 shows the generator control panel and oscilloscope used for PD tests. Correct interpretation not only requires a trained eye, but also knowledge of deterioration of cables and accessories.
The weak sections in a cable length, currently 15,000 ft, can be recognized by concentrations of partial discharges presented in a discharge diagram as a function of the cable length. The decision whether or not to replace a cable section or a splice is based to a large extent on experience.
Selecting the Circuits PECO Energy's initial step in selecting circuits to be tested was to identify all cable in the operating region that would fit the testing methodology. KEMA-Powertest defined limiting criteria for circuits to be tested. Circuits that were predominately made up of PILC (paper and lead) cables were selected if they were less than 15,000 ft long (4600 m) and did not have any branches. In Philadelphia this reduced the testable cable to 5% of the total possible circuits because most of the circuits were branches, i.e. three or more terminal points.
Scheduling of the cable outages proved to be critical to the test program's efficiency. In prioritizing the test program, we considered how the testing would affect the customer.
We also considered the cable's past failure history. Circuit availability was a continuing concern, and the ability to switch load to other sources was a key factor, especially during warm weather.
The test program's goal was to test two three-phase cables per day. This was readily achieved when circuits had two cables in parallel; however, in other cases we had to complete switching the night before a test in order to expedite the day's work.
One concern is that there is always a possibility of cable failures during the test program. These failures could be normal in-service failures or could, in remote circumstances, occur due to the application of test voltages during the test. Consequently, alternative circuits in different areas of the city were planned as back-up sites in the event of cable failures or other system limitations.
Field walk-downs were completed at all test locations before bringing the test equipment on site. PECO field personnel, engineering and representatives of KEMA inspected switching and blocking arrangements, equipment and vehicle access and equipment security. They also inspected the impact to the public with respect to the noise from the portable generators and placement of test leads, which could be a tripping hazard in areas open to the public.
Test-set location also was a consideration. The equipment was mounted in a 2 1/2-ton customized van (Fig. 5). Maneuvering the test van close to buswork and other substation structures was a concern. The van needed to be within 150 ft of the cable terminations. Portable generators supplying the power to the test equipment were located as far from customers as possible to minimize noise impact. For example, whenever it was possible, the generators were placed in transformer compartments.
Results of Testing During the 10 weeks of testing in 1997, we tested 52 three-phase circuits, including 256 conductor miles of cable and approximately 3000 splices. Our testing efforts yielded a wide range of discharge magnitudes and frequency of discharges. These magnitudes and discharges were prioritized, and the 25 worst discharge sites were prioritized for inspection and possible repair. We gave special consideration to circuits supplying critical customers. This work is now in the planning and scheduling phase.
Several discharge sites of varying severity will not be repaired so that we may obtain better correlation between measured discharge magnitudes and time to failure. These results will aid us in future decision making when we need to determine the time available to repair defects before an in-service failure.
Tests were performed both at the phase-to-ground operating voltage and at 1.5-2.0 times the phase-to-ground operating voltage of the circuit. Five out of 156 conductors did fail during the application of the test voltage. Consequently, we were unable to be re-energize these conductors until after they were repaired. One failure occurred in a termination and the remainder of these failures occurred in lengths of extruded solid dielectric cables that were part of the predominantly PILC underground circuits. This disrupted the testing schedule but did not affect any customers.
The only negative consequence of the test failures was that the failures had to be immediately repaired. This was in contrast to the test program's goal, which allows identification of incipient failures and repair of these failures when convenient. Any cable that failed during the test voltage application was obviously near failing in service. The cables that were broken down during testing and not when in service reduced the number of customer interruptions for the year.
Cost Benefit Is Positive An economic analysis compared the cost of partial discharge testing to doing no proactive testing. In this study, the cost of repair was assumed to be identical for a proactive repair and an in-service failure, even though in many cases there may be higher costs for emergency cable repairs for an in-service failure. The only cost benefit credited to the testing was that no fault locating was required to identify the repair site. We did not assign a cost to service interruption for our customers. The preliminary results of our analysis yielded a positive benefit-to-cost ratio of 1.5 for the test program.
The initial project was too manpower intensive. We can reduce costs by training personnel to perform several tasks. For example, mechanics that disconnect the cable from the system could be qualified to do switching and thus reduce the people required to perform the testing. We did not quantify the value in the economic evaluation for improving system performance or for reducing the repair cost of the cable system before a failure.
Our evaluation found that the project was economical even without included several of benefits that we found difficult to quantify in a dollar amount. We expect the cost of diagnostic testing will decrease substantially when additional experience is gained. This, in turn, will further improve the cost-benefit ratio of nondestructive diagnostic testing.
Conclusion Using 0.1-Hz partial-discharge diagnostic testing, we successfully diagnosed 52 distribution cable circuits. The 25 worst discharge sites were prioritized for inspection and possible repair. A number of discharge sites of varying severity will not be repaired in order to collect information about remaining life.
The inspection and repair process will include a careful dissection of the splices that are discharging in order to assess the rate of deterioration in the splices. The goal of this effort will be to improve our ability to correlate discharge magnitude with damage to the splice and ultimately use partial discharge test data to predict remaining life of the discharging splices.
The cost-benefit analysis indicated a positive benefit. For every dollar spent for diagnostic testing, there is a 1.5-dollar benefit. However, several benefits that were difficult to quantify were not included, and there is still a significant potential for cost reduction, resulting in a further improvement of the cost/benefit ratio of diagnostic testing.
Stanley V. Heyer is an engineer with PECO Energy Co., Philadelphia, Pennsylvania, which he joined in 1969. He has the BSEE and MSEE degrees from Drexel Institute of Technology. His current responsibilities include distribution and transmission cable and accessories. He is a member of CIGRE, IEEE Power Engineering Society, IEEE Electrical Insulation Committee and AEIC Cable Engineering Section. Heyer is a registered professional engineer in Pennsylvania.
Alexander J. Jushchyshyn is an engineer with PECO Energy Co., which he joined in 1996. He has the BSEE degree from Drexel University. His current responsibilities include regional distribution reliability and preventive maintenance.
Willem Boone is manager-diagnostic services with KEMA-Powertest, Inc., Chalfont, Pennsylvania, which he joined in 1997. He has the MSEE degree from the Technical University Delft, The Netherlands. Currently, his responsibilities include the management and coordination of KEMA's cable diagnostic services in the United States. He is a member of CIGRE and IEC and is involved in the international activities of these organizations.
Torben Aabo is a principal engineer with Power Cable Consultants, Inc., Ballston Spa., New York, which he formed in 1995. He has the BSEE degree from Aarhus Teknikum, Denmark and has done graduate work in electrical engineering and industrial management at Fairleigh Dickinson University. His current responsibilities include engineering projects pertaining to underground transmission and distribution cable systems. He is also involved with failure investigations of transmission and distribution cables and their accessories. He is a member of IEEE and a voting member of the Insulated Conductors Committee. He is the chairman of working groups involved with developing transmission cables to 60 kV and above and reports of cable failures 69 kV and above.
