Neutral corrosion on underground cable is less of a problem than engineers imagined, indicates a study presented this spring at the Insulated Conductor Committee (ICC) meeting. In fact, the study shows that the problem is responsible for only about one-tenth of the overall failure rate and is often an isolated episode.

Experience at Pennsylvania Power & Light Co. (PP&L), Allentown, Pennsylvania, U.S., seems to concur with the study's conclusion, even though it may represent a worst-case scenario. Since neutral corrosion is a complex phenomenon, it is difficult to predict where it will occur. Some areas are harder hit by the problem than others. One such area is Pennsylvania, which has wide variations in native soils and an abundance of rocky terrain. Of PP&L's 18 million ft (5.5 Mm) of underground distribution cable, 8 million ft (2.4 Mm) consist of constructions having full, bare concentric neutrals. This type of cable was installed between 1963 and 1991 before switching to jacketed cable.

Looking for Corrosion Based on past experience, PP&L thought that extensive neutral corrosion would be found in those problem areas slated for cable replacement. If the findings reported in the ICC paper were true, however, PP&L would be in for a pleasant surprise. The paper, by Bertini of UTILX Corp., Kent, Washington, U.S., concluded that the neutral corrosion problem was largely overstated and was, in fact, one order of magnitude below the problem of premature dielectric failure. Bertini drew this conclusion from the study of national survey data from the Association of Edison Illuminating Companies (AEIC) Cable Report and the Institute of Electrical and Electronic Engineers (IEEE) ICC Task Force 6-21 Cable Neutral Corrosion Report (Table 1). He found that corrosion tended to be localized in discreet geographical neighborhoods and even on specific afflicted cables. Field data, collected by Utilx, while providing neutral detection service for various utilities, confirmed the findings.

Unaware of the ICC paper and its conclusions, PP&L had already decided to use a new technology to monitor the status of its neutrals before automatically replacing all cable in problem areas. The utility selected a new radar-testing process that identified badly corroded runs for replacement. Under contract with PP&L, UTILX found that neutral corrosion was less of an issue than anticipated. Under the circumstances, PP&L cut its recabling costs by more than half.

At one subdivision where problems were experienced with the underground cable, PP&L tested 6184 ft (1885 m) of cable, installed in a low-lying area of Butler Township. Test results showed that only three runs would have to be replaced, for a total of 2104 ft (641 m). Neutral corrosion was especially bad in two sections of a 400-ft (122 m) run in the loop-fed system, where it ran between the first and second transformers. Although the corrosion was isolated, the neutral was compromised in the entire subdivision. The other problem sections were contained in two phases also coming off of the same riser pole. Instead of replacing all 6184 ft in the problem area, testing enabled us to realize substantial savings by replacing only 34% of the footage. What had been discovered in Butler Township was representative of results with radar testing elsewhere during the program. Only 36% of runs, in the troublesome 105,000 ft (32,004 m) selected for testing, had sufficient corrosion to merit replacement.

In the past, the practice had been to replace 100% of the cable, since there had been only sketchy information on the extent of corrosion. Visual examination in conjunction with routine maintenance or repairs helped to shape the assumption of extensive neutral problems. Many runs that had failed for reasons other than dig-ins contained evidence of significant corrosion. Repeated repairs in the same area led PP&L staff to conclude that where there was some corrosion, there was likely to be more. With the 100%-replacement program, it appears that at times good cable was replaced.

A New Approach to Replacement Several years ago, PP&L set up a committee to identify and set priorities for cable replacement. The committee used a number of variables to establish these priorities, which included the importance of the customer base, cable age, previous failures and the state of the neutral during earlier repairs. Although there was a plan to use some kind of testing, it was not clear how crucial its role would be. Vendors offering testing were evaluated and the committee selected UTILX for the work. Although the testing would have to be done on de-energized cable with the high-resolution time domain reflectometer (TDR), costs were only 20% of other methods.

The cost to test, including utility staff costs, totaled about 60 cents/ft (US$1.97/m) and enabled PP&L to test more footage than had been originally planned. The accuracy of the radar-testing device in pinpointing corrosion gave PP&L the confidence to use a spot-replacement program instead of 100% replacement. The high-resolution TDR showed precisely which runs displayed more than 25% neutral corrosion, the criterion used to identify cable that had to be replaced. In line with the ICC study, the corrosion found through testing tended to be localized, isolated in 2- to 5-ft (0.6-1.5 m) segments of a given section. The testing device pinpointed corrosion within inches and was accurate for cables with up to 75% corroded neutrals. If this pattern proved to be typical of the system, then PP&L had been spending 64% more for recabling than was necessary.

The TDR The high-resolution TDR, model CC502b, designed to locate and measure neutral discontinuities, has a microprocessor and a graphical display that indicates the exact location of the corroded area (Figs. 1 & 2). The device is connected to the de-energized cable, and a 2-nsec square pulse with 20-V amplitude is sent through the cable. Changes in impedance, which could be caused by corroded neutrals, reflect energy to the TDR and appear as peaks on the display. Several other key features make it useful for field analysis. The all-digital unit allows signal averaging techniques to improve the signal-to-noise ratio and is capable of storing up to 64 waveforms that can be uploaded to a PC for additional analysis.

Testing Cable was tested at 10 service areas throughout the Poconos area in PP&L service territory. The distance along the cable path was first measured with the use of a measuring wheel in order to set the TDR's propagation velocity to the proper distance, matching the length of the cable. The crew was also instructed in techniques for reading the TDR waveforms. Changes in impedance appear as positives or negatives across the screen. Placing the cursors on any two points of the cable being tested provides a precise indication of the distance to a given anomoly. For example, a splice appears as a distinct, positive peak, followed by a negative, wavelike reflection. Damaged neutrals reflect as positive peaks that increase with the extent of corrosion. Corrosion of less than 25% is barely detectable, showing up as very small peaks. Corrosion between 25% and 50% reflects as a positive peak, whose amplitude is slightly less than for a splice (Fig. 3). More extensive corrosion, from 50- 75%, displays a reflection whose amplitude is greater than a splice but less than the peak denoting the cable end. Finally, 75-100% corrosion creates a peak whose amplitude is even greater than the peak denoting the cable end.

The testing done on the cable under investigation showed that neutrals with 25% corrosion or less occurred about 64% of the time; 25-50%, 9% of the time; 50-75%, 9% of the time; and corrosion greater than 75% occurred 18% of the time. As PP&L discovered, about 40% of the runs in problem areas showed sufficient neutral corrosion to merit replacement under company guidelines. Most cable tested showed less than 25% corrosion.

In keeping with the findings of the study presented at the ICC meeting, testing disclosed that corrosion was localized within specific areas. Where it was localized, it was found to be further focused in specific problem runs. Where it was focused in specific runs, it was often isolated in sections measuring 2-5 ft (0.6-1.5 m). By replacing only the necessary 36% of its problem runs, PP&L saved more than US$2.5 million based on its average annual cable replacement cost of US$3.5 million. The testing program cost of 60 cents/ft (US$1.97/m) compared favorably with the cost of replacing an entire subdivision's cable, which averages about US$30/ft (US$98/m).

Conclusion PP&L expects to continue testing as part of its cable-replacement program and will continue to monitor its results. As for the 64% of the runs that had less than 25% corrosion, these cables had already provided good service for about 20 years, and PP&L decided they would continue to do so. PP&L is still considering treating these cables with an injection of silicone fluid to restore dielectric strength to like-new conditions.

Key to the success of the program was the utility's willingness to reconsider its assumptions that neutral corrosion was a significant problem in the underground system. Testing results showed that the problem was, in fact, isolated. Testing gave PP&L management the confidence to target areas for replacement rather than recable entire subdivisions. As a result, PP&L's recabling budget could be stretched to cover more problem areas than anticipated and bring them up to full reliability. TDW

Bob Gurniak received an ASEET degree from Penn State University and a BSEE degree from Lafayette College in Easton, Pennsylvania, U.S. In 1967 he joined PP&L's Distribution Design Engineering Group and later its Concept Development and Test Center, where he was responsible for connector testing and various other developmental studies. He later joined the Distribution Operations & Maintenance Group, where he became involved with underground distribution recabling and testing criteria. During a 1995 PP&L reorganization, he moved to the T&D Component Engineering Section, where his prior expertise led to the task of guiding PP&L's recabling projects. He is chairman of the American Natural Standards Institute (ANSI) C119.4 subcommittee on overhead bare connectors and chairman of ANSI's C119 main committee for all connectors used in the industry.