The population of aged cables is increasing in many power systems at a time when there is an increasing demand for energy and improved continuity of supply. Insulation failure of a cable due to partial discharge (PD) can have severe social and economic consequences. Constraints on preventive maintenance work imposed by revenue budgets and a lack of trained technicians, coupled with the need to maintain or improve supply reliability, means that PD measurements for insulation assessment are a cost-effective diagnostic tool.
On-line PD measurements provide an efficient method for detecting insulation defects, and assessing and monitoring the insulation of cables to prevent in-service failures. Noninvasive, efficient and widely applicable portable on-line PD measurement techniques are now the preferred choice of the electricity industry.
The portable on-line PD measurement system is comprised of proprietary clamp-on PD sensors and a filter with a specifically selected portable digital storage oscilloscope (DSO). The sensors can be either temporarily clamped or permanently installed noninvasively in several locations at the termination of live high-voltage equipment. These flexible, noninvasive methods of installation avoid the expense of system outages and plant disconnections for testing purposes. The sophisticated functions of the DSO allow real-time display of phase-resolved PD signals and statistical PD signatures (such as PD levels, phase positions and occurring frequencies) that are distinct from noise.
Detecting PD locations becomes simpler and more accurate with clear identification of an incident PD pulse and its reflection measured using a 2.5-billion-samples-per-sec DSO. PD location is critical when large, dangerous PD is detected.
Major failures of Auckland, New Zealand's extra-high-voltage (EHV) power cables in 1998 highlighted the importance of reliable cables. This event challenged electricity network operators to ensure the reliable operation of numerous widely located distribution cables, especially as many cable systems have now been in operation for more than 30 years. The replacement of cables after a specified time period is not practical from an environmental, manpower or economical perspective. Therefore, cable defects need to be identified, located and repaired in a cost-effective way before failing while in service.
For installed cables, the installation of PD sensors in cable joints or terminations on in-service cables cannot be justified. The wide variance of cables, terminations and joints also makes it difficult to noninvasively position on-line PD sensors. Figure 1 shows several practical, noninvasive options for attaching clamp-on PD sensors to in-service cables.
With three successive cable-joint failures within a week, the New Zealand utility, Electricity Ashburton, invited Industrial Research Ltd., a New Zealand-based research and consulting organization that has developed successful noninvasive portable on-line PD techniques to undertake PD tests, to assess the insulation condition of the utility's cable circuit.
For the on-line PD measurements, the PD sensor was placed at the cable termination earth end or applied at the cable termination box screen earth on the ring main switch (Fig. 2). Corona discharges at cable terminations were detected and proved by the PD location test and an ultrasonic detector, which can pinpoint the location of corona.
Since these cables could be isolated via the ring main switches, off-line PD measurements were also carried out for more informative results. The off-line PD tests (Fig. 3) found that a joint had PD at 7.5 kV, which was 1.2 times the normal operating voltage (Uo). The PD location in Fig. 4 indicates that the PD was in the joint position at 25.6% of the total cable length (L) from the test end. The joint was replaced, and Electricity Ashburton installed a permanent PD sensor on a crucial 33-kV cable for more frequent PD monitoring.
This particular example of PD measurements proved that both on-line and off-line PD tests provide more informative results and that the higher test voltages can reveal defects that may not show at the circuit operating voltage. For many short-cable circuits (less than 500 m [1640 ft]) that can be isolated by ring main switches, on-line and off-line PD tests can be achieved with the portable on-line PD measuring system.
PD tests on a 2.1-km (1.3-mile), 3-phase 33-kV cable were conducted on a Northpower Ltd. (Whangarei, New Zealand) circuit as a commissioning test to determine whether there was PD due to suspected mechanical damage during cable installation. A 0.1-Hz very-low-frequency (VLF) generator was used to energize the cable for off-line PD tests. An on-line PD test also was performed to compare the test results for more informative insulation assessment.
Energized by a 60-kV (peak) low-frequency (0.1-Hz) generator, the PD test setup is shown in Fig. 5. The PD and noise identification were based on checking each PD using the pulse-by-pulse method, which compares every captured PD pulse from the two PD sensors and then decides if it is a PD or a noise pulse. Fault location by PD incident and reflection further confirms whether it is PD or noise. The PD levels are shown in Table 1.
The large PD shown in Table 1 was corona discharge outside the cable termination on the test end due to the 0.1-Hz power supply, which changed the capacitive and resistive electric field distribution of the cable termination. These external corona discharges have a much less damaging effect than the PD from cable insulation. The main concern is PD from the cable insulation and joints, which often causes the cable insulation to fail.
To obtain high-sensitivity PD measurements, on-line measuring was carried out at both cable terminations and at two cross-bond joint screen earths. At the substation end, the clamp-on PD sensor was placed on the earth lead of the cable termination while the system was energized. At the cross-bond of the cable joint, the PD sensor was clamped on the temporary direct short link at each cross-bond of cable joint (Fig. 6). The PD level measured at each position and phase is listed in Table 2.
The yellow phase at the substation end had the highest PD level, which was more than five to seven times higher than that of the red and blue phases, respectively. The PD from the substation cable termination is often detected from 33-kV cable terminations shortly after installation.
The test results showed that no PD arose from cable insulation as a result of suspected mechanical stress during the installation. The condition of the cable terminations and joints were also assessed from the PD test results and a future monitoring program was suggested. Comparison of the on-line and 0.1-Hz PD test results showed no correlation in PD level or in PD inception voltage.
The PD level measured on-line was the highest for the yellow phase, but it was the lowest in the 0.1-Hz PD measurement. The PD inception voltages from 0.1-Hz measurements indicated there would be no PD at operating voltage, but this was not the case. The difference in PD results at 50 Hz and 0.1 Hz is most likely due to the differences in the electric field distributions in the cable termination at the two very different frequencies as illustrated by the simple capacitor/resistor circuit in Fig. 7.
This comparison test indicated that the 0.1-Hz generator could provide an efficient high-voltage power supply for on-site tests on long cables, but the PD detected at 0.1 Hz may not occur at 50 Hz.
The objectives of on-line PD measurement and location on a Powerco Ltd. (New Plymouth, New Zealand) 5.3-km (3.3-mile) circuit comprising 33-kV single-core cables were to assess the condition of the insulation and to establish the PD signatures. The PD tests were to be carried out before and after a civil construction project in close proximity to this cable to compare the test results and determine if there was any PD due to possible damage from the construction project.
The three single-core cables had 17 joints and two sets of cross-bond earths. The cable screens were:
Directly earthed at the substation end
Cross-bonded at joint 3 (cross-bond 1) and joint 6 (cross-bond 2), and directly earthed at joint 8; then, cross-bonded at joint 11 (cross-bond 4) and joint 15 (cross-bond 5), and directly earthed at the far-end overhead termination.
The PD signal attenuation and cross talk between phases for the long cable with cross-bonds presented a challenge for the on-line PD measurements. Measurements were taken at the cable termination in the substation switchboard and at every cross-bond earth to achieve high sensitivity for every section of the cable between the cross-bonds. After checking the currents and voltages of the cross-bond earth while the cable was in normal operation, the PD sensor was clamped on the cross-bond earth (Fig. 8).
Both phase-resolved PD measurements for PD levels in each phase and PD location for each discharge pulse were recorded to determine the phase in which the PD occurred and where the discharge was located. The PD position, in terms of distance to the cross-bond where the PD was measured, was determined by the PD locations at the two cross-bonds of the cable section to provide two alternative measurements of the PD position. Figure 9 shows a typical phase-resolved PD measurement (16 acquisitions) for the red phase at cross-bond. This measurement detects if and during which phase the PD occurs. The ac reference was the current in the screen earth, which was not in phase with the phase voltage. Table 3 includes the PD levels measured for each phase at each cross-bond.
Though various PD levels were measured at each cross-bond joint, PD fault location showed that relatively large PD (a few hundred pico-coulombs) was in the red phase of section X3-X4. The PD location was at 82% of X3-X4 from X3 end or 18% of X3-X4 from X4 end. This was confirmed by PD locations from both X3 and X4 joints.
With the length of cable between X3 and X4 taken from the cable records being 1138 m (3721 ft), the calculated PD location to X4 is 205 m (670 ft). The given distance between joint 10 and joint 11 (X4) is 206 m (673 ft). The ratio of 206 m/1138 m is 18.1%, indicating that the PD location is most likely at joint 10.
The cable joint consisted of three layers of heat-shrink sleeves. The large PD was likely to have occurred between two layers not bonded tightly as shown in Fig. 10. As the PD level is relatively high, a kit for this joint can be prepared in case it fails in the near future. With the material kit prepared and the PD location known, the recovery time following a failure would be much easier and quicker. Further confirmation of the PD location by excavating the joint, testing using an ultrasonic probe at the suspected joint and replacing it before it fails in operation, and a monitoring program were also suggested.
This particular test for a relatively long cable section with cross-bonds showed that high-sensitivity PD measurement can be achieved at cross-bond earth positions. The PD measurements at all cross-bond earths and PD fault locations at both sides of each section between the cross-bonds provided confirmative results.
The on-site testing undertaken for Powerco and other distribution utilities demonstrates the following advantages of the portable on-line PD measurement system:
Noninvasive, fast and cost-effective insulation assessment and monitoring
Does not degrade insulation
Suitable for a broad range of cables
Minimum or no power interruption
Various PD measurements and displays give informative and reliable insulation assessment
Easy to use and facilitate the understanding of the PD characteristics
Simple and accurate PD location.
The PD measurements for insulation assessment provide a sound scientific basis for preventive maintenance and replacement planning. This enables asset managers to make informed decisions about whether to replace aged assets or to extend their service lives. Identifying and quantifying PD performances can enhance the reliability of power systems through appropriate remedial actions.
Peter Chappell joined the former New Zealand Electricity Department as an engineering cadet, where he was involved with high-voltage asset construction and maintenance. He is completing courses with the Queensland University of Technology. Chappell is currently employed by Powerco Ltd. of New Zealand and is presently engaged in setting asset maintenance regimes, condition monitoring, and preparing design and construction standards. Peter.Chappell@powerco.co.nz
Yafei Zhou obtained a Ph.D. from the University of Queensland, Australia, in 1994. He spent two years with Queensland Electricity Commission (now Powerlink) developing infrared and ultrasonic techniques for faulty insulator detection. In 1994, he joined Industrial Research Ltd. (IRL) of New Zealand, where he initiated research and development on new on-line partial-discharge measuring techniques for insulation assessment and monitoring. In 2006, Zhou began to provide technical services through AP EnerTec Ltd. when IRL discontinued services. firstname.lastname@example.org
Y. Qin joined Industrial Research Ltd. of New Zealand (now AP EnerTec Ltd.) in 1996 and was appointed a contract research engineer in 1998. Her work includes providing technical services with the new on-line PD measuring techniques. Qin is involved in high-voltage and heavy current testing in the laboratories and has experience in on-site testing for HV equipment.
|Phase on Test||Red||Yellow||Blue|
|Inception voltage (kV, RMS)||31.8||21.2||31.8|
|PD max (mV)||4300||184||2300|
|PD max (pC)||5059||212||2706|
|Test voltage V (kV, RMS)||35.4||35.4||42.4|
|PD max (mV)||4740||2370||2440|
|PD max (pC)||5576||2788||2871|
|PD at substation||(mV)||4.5||26.4||3.4|
|PD at X-bond I||(mV)||9.1||11.0||9.4|
|PD at X-bond II||(mV)||6.0||7.1||6.4|
|Location of PD measurement||PD level (Max mV/Min mV) Red phase||PD level (Max mV/Min mV) Yellow phase||PD level (Max mV/Min mV) Blue phase|