The ICC Website is changing. As this transition is made, our new format can be found at www.pesicc.org/ICCWP. Please send any comments or questions to mvh@voncorp.com.
John Tanaka, Chairman, University of Connecticut, reported that the Insulations Subcommittee met at 11:00 am on Wednesday, November 5, 1997 with 142 members and guests in attendance.
Chairman Tanaka opened the subcommittee 5 meeting with the usual announcement of the time and place of the next ICC meeting. He then asked whether there were additions or corrections to the minutes of the previous meeting as published in the ICC minutes. Hearing no corrections, he proceeded with the agenda of the meeting which was on an overhead.
Mark Walton made the first presentation on the subject of "Dielectric Breakdown Strength Correlation with Time-to-Failure during Wet Aging of XLPE-Insulated Cables." The paper was co-authored by Mark Walton, Bruce Bernstein, William Thue and John T. Smith, III.
Walton started by explaining that the GMTF, the geometric mean time-to-failure, is obtained by statistical analysis (lognormal) of the failure times of a group of cable samples which have been energized at a constant voltage stress. The GMTF value corresponds to the time at which 50% of the samples fail. The GMBD, the geometric mean breakdown value, is obtained by statistical analysis (lognormal) of the breakdown voltages of a group of cable samples which have been subjected to a step or ramp voltage stress is continued until breakdown. This latter test was started at 3.94 kV/mm (100 V/mil) followed by 5 minute steps of 1.58 kV/mm (40V/mil) increments. All breakdown testing was conducted at ambient temperature. The voltage at which 50% of the samples fail is the GMBD value.
The test protocol used is shown in Appendix 5-N-1. Breakdown tests were done on cables after aging periods which were related to the GMTF. In this way, a correlation was obtained between GMTF and GMBD. Three aging periods were used at each aging condition in the aging matrix shown in Appendix 5-N-2, one which provided significant aging, one which provided moderate aging, and one which resulted in lightly aged cables. Three different runs of XLPE were tested. Run 1 outperformed Run 2 in time-to-failure accelerated cable life testing at each aging condition in the test matrix. Both of these runs utilized a conventional conductor shield material. Run 3, the run which utilized a super-smooth conductor shield, performed much better than either Run 1 or Run 2. For the three aging periods (significant, moderate, lightly aged), it was also found that as the aging stress goes down from four times rated voltage, to three times rated voltage, and then to two times rated voltage, the GMBD also goes down. The ambient temperature breakdown values did not seem to be very dependent on aging temperature. The Weibull plots of the ambient temperature breakdown data for each aging period showed very little overlap in the 90% confidence bands for cables aged at different stress levels (4Vg, 3Vg, 2Vg). The data for 1 GMTF cables are shown in Appendix 5-N-2, and those for 1/2 GMTF and 1/4 GMTF cables are shown in Appendix 5-N-3. It should be noted that for 1/2 GMTF cables, the GMBD values come up in comparison to the 1 GMTF cables. The 1/4 GMTF cables had higher GMBD values than the 1/2 GMTF cables.
In conclusion, the GMBD when plotted against ACLT aging stress shows linearity for each of the aging conditions. The GMBD values vs. aging time also show a correlation for each set of aging stresses. These are shown on Appendix 5-N-4. Also as shown in Appendix 5-N-4, the conclusions reached were that for equivalently aged cables, the ambient temperature ac breakdown strength values (GMBD) show a strong dependence on the voltage stress during aging, but not on the load cycle temperature during aging. The load cycle temperature, however, strongly influences aging time required to achieve equivalent degrees of degradation. For equivalently aged cables, ambient temperature ac breakdown strength (GMBD) decreases with the aging stress at all aging temperatures. The very interesting result is that this test methodology is useful for establishing a correlation between GMTF and ambient temperature ac breakdown strength (GMBD) over a wide range of aging stresses and load cycle temperatures.
Jean Marie Braun of Ontario Hydro asked Walton whether he was comfortable with an extrapolation to zero. Al Mendelsohn of Union Carbide asked whether the aging times at lower stresses were carried out for a longer time. At this time, Walton showed the final slides he felt would clarify some of these questions.
Stan Heyer of PECO Energy asked whether the samples were equivalently aged. Walton said that the state of aging was equivalent, not the time of aging.
Nigel Hampton of BICC observed that in the last slide, the lines appeared to be straight. He wondered about the error bar, were the values significantly different? Walton said that there was some overlap at the 90% confidence level. Hampton reiterated the question. He wondered if the failed data had been included to determine the most accurate GMBD value for the partially aged cables when 1/2 and 1/4 life testing was performed. Walton said no. Hampton then asked if there was any censoring of the breakdown data to determine the initial or unaged GMBD value. Walton said that there was a much lower GMBD value reported because a lot of censoring was required to handle termination failures.
Simon Terry of Lantor, Inc. suggested an improved test procedure. Walton said that the test protocols were decided early in the test program and were strictly adhered to.
John Holmes of GPU Energy asked whether all cables had been manufactured at the same time. Walton said that Run 1 and Run 2 cables were produced 14 months apart on the same line attempting to duplicate the same materials and conditions. Run 3 cables were produced three years later once again using the same line.
Al Mendelsohn of Union Carbide noted that the cables had been made in the 1980s. He wondered if testing had been done on cables made in 1997. Walton said that he would like to do more testing, but there is the matter of time and costs.
Larry Kelly consultant asked whether physical testing had been done after the testing was completed. For example, was the conductivity of the semicon the same as at the start? Walton said that these tests were not done.
Carlos Katz CTL wondered whether the power law was applicable to the experimental data. Walton said this has not been done. He expected people to massage the data in a number of ways now that it is published.
John Furno of Union Carbide said he had a problem with the GMTF data in that some points were being left out. These should be part of the average. He also had a problem in that the cables in this study were aged for different periods of time. Walton said that they were trying to get to the same level of aging for the Run 1, Run 2, and Run 3 cables. These cables did not age at the same rate (as determined by the different GMTF values). Because of the different aging rates, different aging periods were required.
Carlos Katz CTL asked about the nature of the elevated temperature. Were not the cables load cycled? How can the temperature effect be determined? Walton said that this was indeed so. Some cables failed in the heat cycle and some at cooler portions of the cycle.
J. M. Braun of Ontario Hydro asked whether there were any failures for the 1/4 and 1/2 or slightly aged and moderately aged samples. Walton said that for all the cable groups aged for 1/4 GMTF aging period, there had only been two or three failures total. For the cable groups aged for 1/2 GMTF aging period, one failure occurred for most groups. For some groups there were two failures.
Haridoss Sarma of AT Plastics asked whether a time temperature superposition treatment had been tried. Walton said that they would need help from someone like Haridoss for studies of this sort.
Nigel Hampton of BICC pointed out that without controlling the fraction of the aging time, the data could not be deconvoluted. As it is, he complimented the authors for a very good piece of work. Walton commented that the test protocol and been decided upon years ago with the help of a number of experts.
Haridoss Sarma of AT Plastics observed that the title of the paper is correlation with life. He felt that the level of breakdowns is not related to lifetimes. Walton said he was told at the beginning of the study that a BD voltage of 7.87 kV/mm (200 V/mil) was an indication of a significantly aged cable. This seemed to apply to field aged cables.
Bruce Bernstein of EPRI observed that the cables tested were not preconditioned.
Steve Szaniszlo of Union Carbide pointed out that in a big production, the quality varies somewhat within the production run.
Eric Marsden of Nova-Borealis reported on a survey of high voltage extruded dielectric cables from around the world. These were 100 to 250 kV cables with one 30 km of 420 kV cable. The majority of the cables were of European origin. The compounding for these cables is also being done in Japan. The high voltage cables were classed into groups of 60-90 kV, 200 kV, 200-300 kV and 300-500 kV. In a slide showing where cables were made and where they are used, he observed that more high voltage cable is installed in Asia than produced in Asia. The important part of producing high voltage cables is to have clean materials. Currently the contaminant counts are being done with an Intec 5000 laser system. This detects impurities from 50( to 2 mil and up. It is not able to detect pale ambers. For the production of extra high voltage cables, a better system is needed. Marsden showed some specs for impurities. Superclean compounds may be used at 275 kV. For 400 kV cables, the Extra High Voltage (EHV) grade should be used, using the second generation contamination detector.
Carlos Katz, CTL, asked whether these cables were all XLPE. The answer was yes.
Nigel Hampton, BICC, noted that Marsden made a point of needing a better quality magnifying glass. He felt that the same magnifying glass could accomplish the same purpose by increasing the sampling. Marsden answered that a better magnifying glass could see things not seen by less powerful tools. He pointed out that no total amount of sample would do if the bare eyeballs were being used. The better quality magnifying glass would see things that the other magnifying glass could not see. Cleaner materials are needed. There are techniques such as ultrafiltration.
John Densley of Ontario Hydro Technologies gave a paper authored by Densley, J. M. Braun, and Z. Nadolny. The paper deals with the partial discharge characteristics of interfaces in cables. In dry systems, the weakest area are the interfaces. There are protrusions, contaminants, etc. Most cables are now manufactured so that they are clean and fault free. However, the joints and terminations are done in the field and thus more prone to contamination. The study involved looking at PD in joints. In the first slide, a model of a prefabricated joint was shown with the interfaces in the prefabricated joints. A second slide (Appendix 5-O-1) shows the blow up of the interfaces, and indicates how silicone grease can affect the partial discharge. The real life situation was modeled by using needles in epoxy and EPR to within 1 mm of the interface. High voltage gave PD along two interfaces and electrical trees within the epoxy and EPR. The measurement of the PD was done with a wide band detection system which retains the wave shape of the discharge current pulse. By using an ultra wide band system, a lot more information can be obtained than by use of conventional PD detection systems. The system was capable of measuring rise and fall times and pulse widths to better than 1 ns amplitude and frequency to characterize discharge pulses. Mechanical pressure to the interface could be varied. At a low pressure, the inception voltage was 7 kV. At higher pressures the inception voltage rose to about 14 kV. Some of the pulses at 100 hours were about two nanoseconds in duration. After a period of stability, the pulses become irregular. It was possible to distinguish between interfacial PD and discharge in the sample itself. The shape of the discharge pulses in epoxy of about 2 nanoseconds was found to be similar to that in EPR. In conclusion, it was noted that test methods had been developed to give information on the shapes of the partial discharge pulses. This can be used to tell where the discharge is occurring. The technique is a useful tool. (See slides in Appendices 5-O-1 to 5-O-3.)
Bogdan Fryszczyn of Cable Technologies Laboratory asked about effects of pulses in different types of insulators. Densley said that this would be the next step.
Carlos Katz, CTL, asked whether there was the same level of inception in EPR and epoxy. What is the ratio of the treeing susceptibility and what is the discharge resistance? Densley said they have not measured these quantities. They can detect when the discharge moves from the epoxy to the interface. They don't know when it goes from the interface into the EPR as the PD at the interface swamp those occurring in the EPR or epoxy. Trees grow more slowly in epoxy than in polyethylene.
Harry Orton, consultant, asked whether another step could be taken and pattern recognition be applied? Densley said it is something which can be done.
Torben Aabo asked how the electrical stress compared to that seen in a cable system. Densley said that the electrical stresses were similar to those in actual cable systems.
Nigel Hampton, BICC, asked about the frequency when the discharges got into the interfaces. Does it occur on every cycle or does it switch on and off. Densley said that the frequency was not monitored constantly. There appeared to be no quiescent period.
Harry Orton, consultant, announced that a survey to replace the AEIC survey which was discontinued in 1991 would be undertaken. The goal is to reestablish the communication between the users and the manufacturers. Orton observed that the NELPA survey is still in place. Benefits of past surveys have included in-service data on the comparison between jacket vs. non-jacket, direct buried vs. duct, and the effect of low electrical stress vs. higher stress. The goals would be to answer questions of acceptable cable designs, new materials, and the reason for failures at the end of summer. One would hope to obtain data on the performance of reduced insulation wall systems. Questions on elevated temperature operations at 140o C. would be assessed. It is hoped to gather data on underground cable systems with good performances as well as those systems with poor performance to find out why one system has a better performance than another. All information will be treated confidentially. Other personnel involved will be Al Kong, Push Patel, and George Austin. Some 83 utilities have already been contacted.
Paul McTigue of Con Ed asked what voltage level would be surveyed? Orton said that the survey would include all distribution voltages.
Lauri Hiivala of Alcatel Canada Wire asked who is going to pay for the survey. Orton said that the utilities would be asked to share in the cost.
John T. Smith, III of BICC asked how the data would become available. Orton said that the data would be presented at the ICC.
Evangeline Cometa of AT Plastics asked whether there would be a replacement recommendation. The answer was that the utility would have to use the data to make its own decision.
Nigel Hampton, BICC, pointed out that it is important to include the cause of failure. The response was that this is not always known.
John Pegram, Union Carbide asked about dig in failures. The answer was that these failures would be eliminated.
Paul McTigue, Con Ed pointed out that it is important to know the loading as well as the failures.
The subcommittee 5 meeting was adjourned at 12:25 pm.
Return to the SC-5 Homepage
