Greetings, thank you for taking this important step to discovering more about partial discharge (PD) measurements and analysis! Among a broad range of insulation tests, PD measurements are used to detect localized defects and anomalies within electrical insulation systems. PD measurements were introduced to the power industry several decades ago; the measurements were mainly performed in the factories and during the R&D processes. In the last two decades, significant technological advances were made, allowing PD measurements to be performed consistently both in the factory and in the field, at different stages in the service life of many medium-voltage (MV) and high-voltage (HV) apparatus.
The goal of this PD Bulletin is to increase awareness toward PD measurements and provide a platform of unbiased technical references on the topic of partial discharge. The content of the PD bulletin will be constantly updated and will cover many aspects of PD measurements for a wide range of MV and HV assets.
We encourage you to give us your feedback on our new PD Bulletin and make suggestions you would like to see. The PD bulletin team whole heartily welcomes you!
Join our mailing list to stay up-to-date when new information is released. As this content is focused on applications within Canada and the US, we ask only North American customers to join the newsletter. We welcome anyone to view the page.
Tuned Medium Band UHF PD Measurements for GIS
Over the last 50 years, gas-insulated switchgear (GIS) has become increasingly present in the high-voltage (HV) substation landscape. Their small footprint, in comparison with traditional air-insulated substations, has contributed to their popularity by decreasing the overall costs of a substation, especially in densely populated areas. Their use has also recently been extended to distribution networks at the medium voltage level and in industrial facilities.
Thus, with this growing installed base of GIS, there is also a growing need to keep those assets in operation to avoid unplanned outages. Partial discharge (PD) measurement has established itself as the main dielectric test for installed GIS. This is especially important as a survey from CIGRE reported that the percentage of failures related to a dielectric breakdown in GIS has continuously increased from 20% in the early 1980s to nearly 80% for the years 2004-2007.
Onsite PD measurements in GIS are usually performed in the ultra-high frequency (UHF) range. In most PD instruments, this measuring range is a non-configurable parameter, which means that the users can’t modify it to adapt the PD measurements to the specific site conditions, which are governed on a case-by-case basis. This bulletin-featured article showcases how selecting the right measuring frequencies can improve the sensitivity of onsite PD measurements on GIS.
Other Articles of Interest:
- Online PD Testing on Metering Instrument Transformers
- Onsite Measurement, Localization, and Monitoring of Partial Discharges on a Power Transformer
- A Guide for PD Measurements - Noise mitigation
- A Study of the Pulse Propagation Behavior in a Large Turbo Generator
- Assessment of the Condition of Dry-Type Transformers by Means of Partial Discharge Diagnosis Using a Mobile Voltage Source
- Experiences with the Acoustic Localization of Partial Discharge in Liquid-Immersed Power and Distribution Transformer with Help of UHF Measurement Technology
- On-site Testing of High-voltage Cable Systems
- Partial Discharge Measurements on Rotating Machines – Experience and Innovation
- A Guide for PD Measurements - Introduction
- A Guide for PD Measurements - Measurements according to IEC60270
Are there any documents that suggest PD test voltage levels?
Test voltage levels or even voltage profiles vary depending on the purpose of the test. For many apparatus, factory acceptance tests are carried following either IEC or IEEE standards.
For maintenance and diagnostic purposes, the test voltage can vary depending on the history of the asset, the age of the equipment and other factors.
Does it mean my specimen is PD-free if I don’t detect anything above the noise floor?
The short answer is no. It is possible that the noise floor has a higher magnitude and masks some PD activities.
As a first step, checking your connection setup and grounding could improve the sensitivity of the measurement. If this does not improve the situation, additional steps can be taken as an attempt to lower that noise floor or to separate the PD signals. More can be found on our article above (Part 3 - noise mitigation) Therefore, it is not possible to calculate the apparent charge (pC/nC) value from those results.
What is the divider ratio or calibration factor?
The divider ratio (for MPD600 users) or the calibration factor (for MPD800 users) is a multiplying factor that is applied to the calculated apparent charge value. It is used during the calibration process. When a calibration pulse is injected across a test object, part of the energy will circulate through the stray capacitance of the system and therefore, will not reach the sensor. The divider/calibration factor corresponds to the ratio between the measured charge at the sensor and the charge of the injected calibration pulse.
What is the measured center frequency (fcenter) and measured frequency bandwidth (Δf)?
When using the frequency domain integration, a bandpass filter is applied to every recorded PD pulse. The parameters fcenter & Δf define the cut-off frequencies of that filter (fcenter ± Δf/2). The filtered signal is afterward integrated to obtain the apparent charge.
Will partial discharge measurements detect every type of defects in my test object?
No. Partial discharge measurements will only detect anomalies that generate partial discharges. Depending on the asset, some defects creating PD could also go undetected for different reasons (e.g. PD attenuation, user errors and high level of interference).
Can partial discharges be measured under DC voltage?
Yes. However, data interpretation will be different as no phase resolved partial discharge (PRPD) diagrams will be created. Both the MPD600 & MPD800 can measure PD under DC voltage. You will also need a HV resistor in the circuit as voltage divider as the coupling capacitor can no longer function as voltage divider under DC Voltage.
How do I select a coupling capacitor?
In general, a higher capacitance value will provide a higher sensitivity. This will result in more charges stored in the coupling capacitor and therefore, more charge available to reload to PD site. IEC60270:2015 says that the highest sensitivity is realized when Ck >> Ca, which are respectively the coupling capacitor and the test object. This is rarely feasible in practice and therefore, it is recommended to use a value of 1nF or higher.
We also recommend using a coupling capacitor that is PD free (<1 pc at rated voltage).
Is there are predefined limit on PD level for the power apparatus?
It is not available for every apparatus. For example, power transformers, bushings and cables have published limits for Factory Acceptance Test (FAT). However, these limits are usually not applicable for on-site testing.
Are PD levels (high/low) sufficient to determine whether there are severe PD activities in my insulation system?
For specific apparatus and for specific standards, PD levels can be used as a pass or fail criteria for factory testing. However, it is usually not enough to assess the severity of a defect. PD are created by anomalies and defects within the insulation system.. Some defects are less severe and can create high PD level such as Corona discharges (discharges to the air). However, void discharges are more severe in comparison and can deteriorate the insulation materials much quicker.
It is therefore important to also use other criteria such as the Phase Resolved Partial Discharge (PRPD) pattern, repetition rate, relationship with voltage in order to conduct a comprehensive study on the PD phenomenon.
Can I covert mV/ dbm reading to apparent charge?
Usually the equipment that used to measure mV/ dbm signal are using different technology (frequency range, data processing method). Therefore, it is not possible to calculate the apparent charge (pC/nC) value from those results.