Q&A: Partial discharge testing

Author: Charles Nybeck, Ph.D., Substation Applications Engineer

Partial discharge testing is a powerful technique for determining the health of high voltage assets and diagnosing faults in them. For many engineers and technicians, however, this technique is relatively new and unfamiliar, so it unsurprising that it generates a lot of questions for our support team. Here are the answers to some of the most frequently asked of those questions.

Q: How can I determine if there is partial discharge activity in a high voltage asset?

A: There are several methods for doing this and they can be broken down into two groups: conventional testing and non-conventional testing. The conventional method of testing is defined by the IEC 60270 standard, in which the measurement circuit consists of a coupling capacitor to decouple the PD activity, an input impedance and filter for the measurement system, and the measurement instrument. This method requires calibration once the measurement system is in place to ensure that the apparent charge at the test terminals, measured in pC, is representative of the actual discharges taking place in the asset as seen at the test terminals.

When using this test method, one of the most effective ways to investigate PD-related issues is to look not only at the magnitude and number of discharges, but also at the phase resolved partial discharge (PRPD) pattern. The PRPD pattern allows for the amplitude in pC and number of discharges to be overlaid with the phase angle of the applied voltage, offering insight into the type and severity of the defect.

Several PD testing methods fall into the non-conventional group, including optical, electromagnetic, chemical, and acoustic detection methods. The electromagnetic and acoustic methods are the most common. Partial discharges give off high frequency transient signals that can be captured by means of inductive and capacitive sensors, and by specially designed field probes.

Measurements taken using this method are in dBμV or dBm rather than pC, because this method cannot be calibrated. Partial discharges also produce pressure waves that can be decoupled and measured using special sensors such as piezoelectric transducers. This method is typically used in conjunction with an electrical trigger signal, primarily for locating PD defects in high voltage assets such as transformers or GIS equipment.

Q: Are on-line measurements more critical than off-line measurements for determining machine health, or is it the other way around?

A: Both on-line and off-line measurements are critical for determining the health and condition of the insulation system. On-line measurements made with permanently installed couplers are the only way to achieve PD monitoring of assets. On-line monitoring allows data to be collected continuously, which means that alarm thresholds can be set and arrangements made for the user to be notified automatically if the thresholds are exceeded. On-line measurement are very similar to on-line monitoring in that they both use permanently installed couplers, but the measurements are made periodically rather than continuously.

For both on-line measurements and on-line monitoring, the operating conditions must be considered. The device under test is subject to operating loading, humidity, temperature, and voltage. Off-line measurements allow more freedom when testing because the applied voltage can be varied as an aid to identifying the inception (PDIV) and extinction (PDEV) voltages. The couplers can also be moved to perform a variety of tests, and the ambient conditions, such as temperature and humidity, can often be varied.

Q: Power frequency and Very Low Frequency (VLF) PD test methods differ, but do the differences apply specifically to sinusoidal VLF testing or do they extend to Damped AC (DAC) and cosine rectangular testing?

A: During power frequency measurements, the applied voltage cycles are 50 or 60 times a second, depending on the supply frequency. However, VLF testing is typically performed at 0.1 Hz, which is 500/600 times slower than line frequency. Partial discharge activity is strongly dependent on the dV/dt, or change in voltage with respect to time, so testing with sinusoidal VLF can lead to longer test times and may be limited in the amount of PD activity it can excite. While using DAC or cosine rectangular testing techniques, dV/dt is much greater than with sinusoidal VLF testing and is more likely to induce partial discharge activity.

In cosine rectangular testing, the applied voltage still has a fundamental frequency of 0.1 Hz, but during polarity reversal, dV/dt is close to that of a power frequency sine wave. This allows VLF withstand and PD tests to be performed at the same time. Test times may be longer utilising the sinusoidal VLF method, but as the dV/ dt of the polarity reversal is close to that of the power frequency waveform, defects inducing PD activity should be easier to identify.

With the DAC methodology, the voltage is applied in excitations that are short in duration, and the frequency is dependent on the inductance and capacitance of the system. This is ideal for diagnostics on cables where you do not want to apply a constant stress to identify PD activity, while still maintaining results comparable to power frequency. Results from each method should be compared with results that were obtained using the same method to avoid any misleading analysis.

Q: What is the difference in procedure for an applied voltage PD test and an induced voltage PD test for transformers?

A: In an applied voltage PD test, all terminals of the windings under test are shorted together and connected to a power frequency supply. All other winding terminals are shorted together and connected to ground. During the test, all parts of the winding and leads are at the same voltage with respect to ground and the other windings. When carrying out the test, the voltage should be started at one quarter or less of the maximum voltage and increased gradually up to the maximum, in no more than 15 s. After being held for 1 minute, the voltage should be reduced gradually, in no more than 5 s, to one quarter of the maximum voltage or less, and the circuit opened.

For an induced voltage PD test, a three-phase voltage at greater than or equal to twice the rated frequency is applied to the low voltage terminals. Unlike the applied voltage test, all the other line terminals are left open while the neutral and tank are grounded. The applied voltage depends on the voltage class of the transformer. For class II transformers, the test voltage is raised slowly to 150 % and held for few minutes before being raised to an enhancement level of approximately 173 % for 7200 cycles. It is then reduced to 150 % and maintained for 1 hour. During this test, PD measurements are recorded every 5 minutes. For both tests, the pass condition is no collapse in voltage and no audible internal sound. More information can be found in IEEE standards C57.12.01, C57.12.91, and C57.113.

Q: What is the difference in method when performing a PD test on a motor or generator with an accessible neutral and one that does not have an accessible neutral?

A: On a motor or generator with an accessible neutral, it is possible to energise the complete winding to ground and also the individual phases to ground. It is recommended that the measurement is performed at the line terminals of the individual phases (A1, B1, C1) while grounding the terminals that are not being tested. Voltage should be injected from the neutral to introduce additional damping or filtering for more in-depth measurements. For instance, when testing A phase, voltage would be applied to the neutral side (A2), while the coupling capacitor and measurement circuit are connected to the line side (A1) and the additional phases (B1 and C1) are grounded.

When performing a PD test on a motor or generator without an accessible neutral, it might seem that the test could only be performed by shorting the line side windings, applying a voltage, and taking a measurement. This would result in an averaged measurement between all windings and could potentially mask a problem with one winding. Therefore, rather than performing one measurement, it is recommended that the measurement is performed at each individual winding, for instance, by applying voltage to A1 and measuring at B1, or by applying voltage to B1 and measuring C1 and so on. This allows for an analysis of each winding, even if the neutral is inaccessible, and can help to prevent the masking of potential issues.