Partial discharge (PD) testing of in-situ power cable and accessories is an emerging technology. The development of many diagnostic systems — de-energized and energized — provide much data. However, the key for success is in better data interpreting and understanding skills. These skills will evolve as the technology progresses, the data analysis tools improve and the test bank data increases. The industry now has access to several commercially available diagnostic tools, yielding much experience for usage on paper-insulated lead cable (PILC) and solid dielectric polymeric insulated cables.

The Need

The age and size of most installed utility underground facilities is increasing. This directly corresponds to the overall failure trend growth unless these same utilities implement a directed maintenance program.

Utilities also are more commonly embracing underground. For example, Colorado Springs Utilities (CSU, Colorado Springs, Colorado, U.S.) has 62% of its distribution network underground, with the first installations occurring 33 years ago. The average age of installed cable at failure is 21.6 years.

Utilities are searching for a diagnostic tool that will locate cables and accessories in weakened condition and likely to fail. This leads to the use of proactive maintenance measures to replace or repair equipment prior to failure.

Today, de-energized and energized diagnostics are available commercially and used to determine the condition of cable and cable accessory insulation. These systems must locate types of insulation degradation that occur in cable systems.

Summary Table of Diagnostic Techniques
Destructive Non-Destructive, De-Energized Non-Destructive, Energized
Dissection and microscopic examination Electrical tests Methods to detect singular faults Integral measurement methods Integral & measurement singular methods
Contaminants, voids and water trees 60 Hz step tests, acbd tests, radial “power factor” PD at 60 Hz Kinectrics factor at 0.1 Hz Dissipation BAUR, Kinectrics Leakage current Sumitomo
Chemical evaluation FTIR, DSC, DMA, OIT, DP, PIXE DC hipot Resonant circuit 0.1 Hz sine Oscill. wave Impulse PD at 0.1 Hz KEMA Dielectric Spectroscopy ABB DC component Fujikura
Mechanical evaluation tensile, elongation, burst test DIACS PD location system (<2U0) (IMCORP) LIpATEST Powertech Harmonic Current Sumitomo & NRC
PD location system (3U0) (IMCORP) CDA and OWTS, PD Lemke & Univ. Delft Isothermal relaxation current SINTEF PD, Power Diagnostix, KEMA, DTE, Sumitomo Eaton
Return voltage Hagenuk & Univ. of Siegen

The first is an average or overall condition caused by chemical aging or water treeing. The diagnostics for this type of aging include dissipation factor (loss angle), harmonic analysis, return voltage, isothermal relaxation current, dielectric response or dc leakage current.

The second type of degradation is localized condition assessment using dissipation factor measurements or PD level measurements. No matter which type of diagnostic is used, apply it in a non-destruction manner so the diagnostic test does not reduce cable or accessory life.

Evaluating results is more difficult because cable accessories behave differently than cables. For example, accessory designs are not always properly tested. Some accessories (splices) are man-made in the field, so workmanship is a concern, and accessories are not always properly tested after installation. Typically, accessories are made of materials that are resistant to PD activity, and can withstand PD and treeing activity better than adjacent cable insulation. However, PD detection is particularly valuable in evaluating cable accessories, because there are likely to be more defects in a cable accessory than in a cable.

De-Energized Diagnostic Techniques

De-energized diagnostic techniques require the de-energization of the cable under test and, in some cases, completely removed from the underground distribution system. Before any diagnostic test begins, remove any residual space charge remaining in the cable under test, as this may influence the test results. Accomplish this by either direct grounding of the cable conductor, or for longer cable lengths, by using a graduated resistive grounding stick. The grounding stick also limits internal switching surges that could further damage cable insulation.

At the conclusion of the diagnostic test program, remove any charge left on the cable by the test procedure. This will prevent damage to the cable insulation from polarity reversals upon re-energization.

De-energized testing has the advantage of close control of the test voltage, and if necessary, of raising the voltages above the normal operating voltage. However, exercise extreme caution if using overvoltages.

  • The advantage of 60 Hz de-energized diagnostic testing is that the obtained test results relate to the cable operating conditions, as the frequency is exactly the same. Using a resonant test set reduces the size of the test set size substantially. Furthermore, control of the test voltage and frequency along with overvoltage testing leads to location of problem areas that normally are not seen at the rated voltage. It is imperative, however, that the user exercise extreme caution during any overvoltage test because of the potential of damaging the insulation in other weak locations.

    PD patterns at 60 Hz provide a great deal of information on the type, severity and location of the degradation area but may change with load and temperature. It is not easy to assess PD severity, but PD analysis finds a good proportion of serious cable insulation defects.

  • De-energized very low frequency (VLF) testing in the range of 0.1 Hz successfully measures dissipation factor and PD levels in cables and accessories. VLF damages the insulation less than dc testing and locates potential failure sites more reliably than a dc high-potential (hipot) test. Portability is extremely important in field testing. VLF has the advantage of low energy requirements, 1/600 for 60 Hz voltages and 1/500 for 50 Hz voltages, which results in much smaller test sets than those required for normal operating frequency measurements. VLF PD detection can find major defects in cable accessories.

  • De-energized complex discharge analysis (CDA) combines the meaningfulness of an ac test with the small power demand of a VLF test. The testing method charges the cable slowly for 10 seconds and upon reaching the recommended peak voltage, the cable discharges with a periodic oscillation of about 10 ms, that is, about 50 to 60 Hz. PD activity is recorded during the discharge cycle with the PD pulse repetition rate and magnitude determined during the discharge period to locate any defects. For statistical significance the charge/discharge cycle repeats 10 times.

  • The International Electrotechnical Commission (IEC) standardized oscillating wave test system (OWTS) is a non-destructive, after-laying test. Further developments resulted in a diagnostic test suitable for de-energized applications in noisy environments. The OWTS produces ac voltages in the range of 20 to 1000 Hz depending upon the cable length and can be used for PD pattern analysis provided there are no high disturbances present. The method has been used on XLPE cables with a maximum length of 7348 ft (2243 m). It is suggested that the time for each test is the same as a dc hipot test. The method is also used for dissipation factor measurements.

Energized Diagnostic Techniques

Energized diagnostics have the advantage that the cable is not switched out of service, leaving it energized as it is for normal operation. This removes the potential of damage due to inappropriate switching and yields no system contingency problems. The disadvantage is the lack of test voltage control and the removal of overvoltage testing.

  • Ultrasonic detection of PD in cable accessories can pinpoint a suspected problem if the cable accessory is not direct buried, or is at least physically accessible. Ultrasonic detectors are lightweight and portable. PD within a cable accessory produces a broad range of sound that can be detected with ultrasonic translators. The high-frequency ultrasonic components are extremely short wave in nature, fairly directional and easy to isolate from background noise. Parabolic reflectors or concentrators can detect emissions at a distance.

  • Direct, capacitive or inductive couplers (sensors) install on or near the cable accessory to locate and measure PD in cable splices and terminations while the cable is energized. Some diagnostics require installation of the sensors prior to energization of the cable system while others allow retrofitting. Energized diagnostics allow for data trending and analysis such as Pulse Phase Analysis. Evaluate trends to take into account the influence of load, voltage, temperature and humidity.

One PD detection method detects PD components up to 300 MHz near the cable accessory. Pattern recognition combined with Neural Networks improves PD phenomena and cable accessory diagnostics. The system recognizes each detected pulse as PD or noise and dramatically improve data analysis.

Utility Experience

CSU removed many splices with an indicated high PD and dissected each splice to determine the cause. Workmanship was the cause of most of the problems. In addition, some samples were sent to a laboratory for further investigation. However, the high PD could not be repeated, nor could any related defects be found. CSU also found that PD attenuation occurred in locations with a missing concentric neutral. CSU realized considerable cost savings with one diagnostic test.

After a series of in-situ PD tests, Oklahoma Gas and Electric (OGE, Oklahoma City, Oklahoma, U.S.) found it difficult to take correct action because of the limit of the applied voltage accuracy, the difficulty in assessing the PD types and the needed improvement in the accuracy of the PD location. While repeatability was good, the effectiveness was dependent upon the test personnel.

PECO (Philadelphia, Pennsylvania, U.S.) found there was an inability to distinguish discharge sites that lead to failure from those that do not lead to failure. Dissection of three high-discharge splices did not locate any evidence of an incipient failure.

Xcel Energy (Minneapolis, Minnesota, U.S.) linemen expressed concern about splices removed from service when the splices appeared to be in good condition. The linemen kept asking, “Was replacement really necessary?” PD testing of the splices gave only limited information about the condition of splices, so this utility is proceeding with caution.

To maximize information gained, TXU Electric and Gas (Dallas, Texas, U.S.) went to overvoltage testing at three times the rated voltage (3Uo). It is interesting to note that one termination failed after testing that indicated a low PD level. This utility suggested the successful use of 60 Hz testing to determine the condition of cable and accessories, but that the cable is more susceptible to failure than the accessories.

Northeast Utilities (Hartford, Connecticut, U.S.) has concerns about further damage to the cable caused by the diagnostic, the effect cable length has on the diagnostic and the advantages/disadvantages of de-energized versus energized diagnostics. Northeast Utilities continues to use the ultrasonic detector.

Into the Future

PD detection and location in cable and cable accessories is still undergoing development. So far, interpretation has proven to be difficult because of the number of variables. Future diagnostics should look at PD activity over time or trend analysis that includes operational parameters, such as voltage, load, temperature and humidity.

Diagnostics with the intelligence to distinguish between PD types, especially between harmful and harmless PD, would be invaluable. The knowledge and training to improve the interpretation also is strongly needed.

As the technology evolves and additional experience is gained, the use of both energized and de-energized methods will continue to gain in importance along with the continuance in the advances in PD in ease of use and analysis.

Harry Orton is the president of Orton Consulting Engineers International (North Vancouver, British Columbia, Canada), which focuses on the design, installation and testing of underground transmission and distribution systems. Orton graduated from the University of New South Wales with the BSEE degree in 1966 and a Masters of Applied Science from the University of British Columbia (Vancouver) in 1969. After graduation, he worked at BC Hydro, the major electrical power utility in British Columbia, as an electrical research engineer and later as department manger where he helped build one of the largest utility-based research centers in North America.

Author's Note: This article is based in part on ongoing dialogs at industry meetings. Much of the background material is available in the minutes of the IEEE Insulated Conductor Committee Meetings, which can be accessed www.ewh.ieee.org/soc/pes/icc/

Partial Discharges

Partial discharges (PD) occur at voids (or cavities) in electrical insulation or at interfaces between materials such as in cable accessories. The microsparks that occur in these voids emit broadband radiation from 50 kHz to greater than 500 MHz, with pulse rise times of 1.0 ns.

The discharge localization method remains the same, but the magnitude and number of pulses can vary considerably with time, voltage, temperature, load and humidity, making detection and location difficult.

In addition, attenuation along the cable length, particularly at the higher frequencies, and background noise in the field can compound the methodology. Multiple discharge sites, cable branches, contamination and the splice material type can add to the complexity of interpretation.

In spite of these concerns, PD detection on installed power cables is an emerging technology showing considerable promise. The energy inherent in a PD leads to damage of the material surface surrounding the void or cavity. Surface erosion occurs and electrical trees initiate and grow in the body of the insulating material. The process becomes self-perpetuating until the electrical tree bridges the insulation and complete breakdown occurs. PD accompanies the whole process.

Material erosion rates because of PD activity are very material dependent. Some rubber materials can tolerate PD activity with minimum erosion, while other polymeric materials have minimum tolerance of PD activity making degradation assessments unreliable. Also, the PD magnitude is dependent upon many variables, including the shape of the offending void, a parameter that can change with time.

Unfortunately, the time-to-failure predictions for a cable system based upon PD magnitude is not possible. Studies show that growth of electrical trees can be quite rapid during low PD levels, where large PD levels can prove to be quite harmless. Cable-life estimation will depend upon the ability to distinguish between harmful and harmless PD, which is not an easy feat.

The tester can detect and locate PD activity in underground power cables and accessories “off-line” or “on-line.” A portable alternative-voltage supply is necessary for off-line detection.