Partial discharge (PD) is a localized dielectric breakdown that only partially bridges the insulation between conductors. PD in transformers can be categorized into two types:
- PD occurring in gas bubbles within oil
- Sparking-type PD that happens in the oil or paper insulation, with the latter being significantly more concerning [1]
The continuous presence of PD activity in a transformer’s solid insulation during operation can lead to the degradation of surrounding materials and potentially result in a complete failure of the electrical insulation system over time. Consequently, PD measurements are routinely conducted at factories to identify quality issues. It can also be performed during commissioning to assess the insulation condition on-site following transportation and installation or for diagnostic purposes.
Dissolved gas analysis (DGA) of the transformer insulation liquid is a well-established method for evaluating transformer insulation and can also indicate signs of PD among other failure modes. Traditionally, DGA has been the primary method for detecting PD in liquid-filled transformers on-site, with PD measurement and acoustic localization methods employed as follow-up diagnostic methods to identify the location and/or severity of the issue. Additionally, visual inspections or significant repairs may prompt on-site PD testing.
Although DGA has numerous advantages, it may miss other potential issues, such as those related to bushing insulation. There have also been instances where no increase in combustible gases was detected, even when a PD source was identified through electrical methods and confirmed by disassembling the transformer. Thus, PD measurements provide a more immediate, direct, and accurate assessment.[1] With lead times now often extending to up to four years for large transformers and between 6–12 months for bushings, the added reliability from PD testing is crucial for determining whether a transformer requires further corrective measures.
Due to high costs and complicated logistics, on-site PD measurements for liquid-filled transformers are often regarded as specialized tests, conducted only on critical units. However, a variety of alternative PD testing methods are available. Selecting the most effective one can be difficult, as each comes with its own set of limitations and benefits. This article introduces some of these PD testing methods, demonstrating their effectiveness through case studies.
OVERVIEW OF ON-SITE PD TESTING METHODS FOR POWER TRANSFORMERS
On-site PD measurements are typically classified as off-line or on-line. Off-line measurements require disconnection from the grid and an external voltage source to energize the transformer being tested; on-line measurements are performed with the transformer connected to the grid. On-line PD measurements can be categorized as intrusive or nonintrusive, based on whether an outage is necessary for sensor installation. In a factory setting, PD measurements are typically conducted using the bushing taps. Alternatively, various types of PD sensors can be utilized on-site.
A PD event produces a current pulse that contains a wide range of frequencies. High-frequency components emit radially as electromagnetic signals, while lower-frequency signals are transmitted through the conductor. A PD event generates sound, light, and heat, and creates chemical reactions with surrounding materials. Due to the diverse phenomena associated with a PD event, multiple methods for measuring and detecting PD can be employed, including DGA, optical, and acoustic methods for detection, electrical measurements, and electromagnetic measurements in the ultra-high frequency (UHF) range. Some of these methods will be discussed in the following sections.
Off-Line PD Measurements
For off-line measurements, there are two methods for energizing the transformer: the induced voltage test and the applied voltage test.
In the induced voltage method, a voltage is applied across the low-voltage terminals or the tertiary terminals of the transformer (whichever has the lowest voltage rating) that induces a voltage in the other windings. This is typically done using a motor-generator set or a frequency converter. This setup stresses the phase-to-phase insulation as well as the phase-to-ground insulation, typically mimicking the electrical field distribution during operation. This makes it the preferred method for critical transformers. For on-site testing, this voltage source is usually mounted on a commercial trailer due to its large size. Figure 1 illustrates an induced trailer used for off-line partial discharge measurements on a 330MVA, 525kV / 21kV transformer.
In the applied voltage test, the terminals of the winding under test are set to the same potential, which prevents any magnetizing current from flowing in the core and prevents voltage from being induced in the other windings. This results in a lower power requirement. A standard 60 Hz voltage source is suitable for this test, but a resonant test system (RTS) can also be used. There is no phase-to-phase or turn-to-turn stress during the test since the voltages at each terminal of the winding under test are nearly identical. Figure 2 illustrates an applied voltage test using an RTS during off-line partial discharge measurements on a 90 MVA, 53 kV / 1 kV transformer.
The primary benefits of conducting off-line PD measurements on-site are reduced interference levels compared to on-line measurements, resulting in increased sensitivity to defects. Additionally, the voltage can be controlled, providing greater flexibility that can aid in the differentiation of various issues. The main drawbacks include the need for a power outage and the logistical challenges and costs associated with performing these tests.
If bushing taps are unavailable, coupling capacitors can be temporarily connected to the winding terminals. Alternatively, if a drain valve is accessible, a UHF antenna can be inserted into the transformer tank, as illustrated in Figure 3. The key advantage of using a UHF antenna is that the transformer tank acts as a Faraday cage, significantly reducing interference levels. Therefore, if PD is detected by the UHF antenna, it is highly probable that it is occurring inside the tank. However, the pathway between the PD site and the antenna must be free of metallic obstructions, as conductors can block the radiated electromagnetic waves, preventing them from reaching the antenna. Additionally, not all valves are compatible with a UHF sensor.
In certain situations where temporarily installing a coupling capacitor at the terminals of transformers can be difficult, alternative sensors like high-frequency current transformers (HFCT) can sometimes be employed. A typical example of this is with distribution transformers that utilize separable insulated connector systems. The HFCTs can be placed over the semiconductive layer of the cables, as illustrated in Figure 4, or wrapped around the shield of the insulated cables.
On-Line PD Measurements
When PD activity is indicated within a transformer, deciding to take the transformer off-line solely for electrical testing can be challenging and often does not receive approval from asset management outside of scheduled maintenance periods, which can be as long as 7 years.[2] Consequently, on-line measurements serve as a valuable alternative or follow-up action when PD is suspected in a transformer. During the test, the transformer is energized by the electrical grid. Measurements can be intrusive, requiring an outage to temporarily install sensors, or nonintrusive, when the measurement is performed seamlessly.
The primary difficulty with on-line PD measurement arises from the high levels of interference encountered during testing, which can reduce the sensitivity to defects. Various software tools[3, 4] are available to assist test operators in distinguishing between internal and external PD, as well as separating PD from interferences. However, successful testing typically requires an experienced user, particularly when using alternative sensors in nonintrusive scenarios.
- Intrusive. Intrusive on-line measurements require the transformer to be taken off-line temporarily to install and remove PD sensors. The sensors used are similar to those previously described for off-line measurements, with the primary difference being that they are typically more robust. Figure 5 illustrates a bolted bushing tap adapter used during on-line PD measurements. Figure 6 shows three HFCTs installed inside the terminal box on the low-voltage side of a transformer, where an outage was required.
- Nonintrusive. A PD measurement is nonintrusive if it can be conducted while the transformer is still in operation without disruption to service. This is typically feasible only if permanent PD sensors are already in place, which is uncommon for power transformers in North America. Alternatively, HFCT and TEV sensors can be installed while the transformer is in operation. The HFCT can be positioned around the ground connections of the tank due to its split-core design (Figure 7), allowing installation with the asset on-line.
If the transformer is connected to shielded power cables that are safely accessible, the sensors can also be placed on the cable shields (Figure 8).
Until recently, TEV sensors were primarily used to detect PD in metal-clad and metal-enclosed switchgear. However, TEV sensors can also be effectively utilized for nonintrusive PD detection on-site for liquid-filled transformers.[5] Figure 9 shows a TEV sensor installed on a transformer tank.
Monitoring
Like intrusive on-line measurements, installing sensors for monitoring purposes requires an initial outage. The installation is meant to be permanent, enabling nonintrusive continuous measurements to be performed. By triggering alerts when elevated PD levels are detected, monitoring can help prevent in-service failures by detecting issues immediately. Typical sensors for transformer PD monitoring are UHF sensors (Figure 3) and/or bushing tap sensors (Figure 5).
Acoustic PD Localization
The specific types of PD activities and their precise locations within the transformer are crucial pieces of information that can lead to significantly different actions. There are various methods to gather information about the PD’s location. When measurements are taken from the bushing taps, the polarity of the detected PD impulses can indicate whether or not they originate from the bushing. For PD within the transformer, the location can be identified using piezoelectric sensors placed on the outer surface of the transformer tank, a method often referred to as acoustic PD localization. These sensors detect the vibrations from acoustic waves that reach the tank walls. If the PD is not occurring within the oil space of the transformer (for example, in a bushing or gas head space), acoustic PD localization may not be very effective.
One significant challenge is that a transformer tank contains not just oil, but also various components, such as windings, a magnetic core, accessories, and solid insulation, that interact with the acoustic wave. Mechanical interferences at the site pose another challenge. To enhance sensitivity in the presence of these interferences, post-processing techniques like averaging and digital filters can be employed.
CASE STUDIES
This section presents six case studies of PD measurements performed off-line and on-line. For off-line measurements, two voltage sources are considered: induced and applied voltage. For on-line measurements, intrusive and nonintrusive measurements are used, in addition to permanent PD monitoring.
Case Study 1: Off-line Induced Voltage PD Measurement + Acoustic Localization
Metallic shavings were found floating in the oil during installation of a newly manufactured 330 MVA, 300 kV/20 kV single-phase generator step-up (GSU) transformer in a nuclear power plant. This triggered PD testing conducted after the oil was filtered on-site. Due to the high criticality of the unit, conventional PD measurements via the bushing taps according to IEC 60270 were conducted using an induced trailer. Although the GSU is manufactured and tested as a single-phase unit, it operates in a three-phase system with two other GSUs. Therefore, the transformer was energized using a phase-to-phase source between X1 and X2 to replicate operating conditions.
PD was detected from Terminal X2. The polarity of the pulses indicated internal PD inside the X2 bushing. The measured PRPD pattern is available in Figure 10. Acoustic localization was attempted but unsuccessful, further indicating that it was not occurring in the oil. To ensure the PD tests resulted in no degradation, DGA was performed before and after the test and showed no increase in gas concentration. After the findings, the X2 bushing was replaced, and subsequent PD measurements showed no PD activity.
A bushing replacement performed on-site solved the issue. Being able to localize the PD activity enabled the owner to take quick corrective actions without the need to disassemble the transformer and ship it back to a factory.
Case Study 2: Off-Line Applied Voltage PD Measurement + Acoustic Localization
Two low-voltage solid dielectric bushing failures occurred months apart on a transformer at an industrial facility. After the first failure and bushing replacements, no additional spare solid insulated bushings were available. Because the lead time for new solid insulated bushings was several months, the owner replaced the failed bushing with a spare oil-impregnated paper (OIP) bushing. Due to the bushing replacement and prior issues, a PD test was conducted using the taps of each LV bushing. An applied voltage test was performed for logistical and economic reasons.
Since the main focus was on the bushing, the applied voltage test resulted in the same electric field distribution at the bushing as the induced voltage test. PD activities were detected from all terminals, but the PD detected from X3 was significantly higher than the other bushings, and the polarity of the PD also pointed to X3. The measured PRPD pattern is available in Figure 11.
After PD was detected, acoustic localization was attempted but unsuccessful, indicating that the PD was not located inside the tank. The way the bushing was stored was identified as a potential cause of the PD; the OIP bushing had been stored horizontally for over 30 years against manufacturer recommendations.
Case Study 3: Off-Line Induced Voltage PD Measurement + Acoustic Localization
DGA analysis revealed an increasing trend of acetylene (C2H2) in a 525 kV, 374 MVA single-phase autotransformer. The values rose from 1 ppm to 6 ppm over a six-month period, indicating a developing fault condition, such as PD of sparking type or arcing. An off-line PD test was requested using an induced trailer due to the high criticality of this unit.
After PD was detected from Terminal H1, PD localization was applied. The localization software indicated issues in the tank beneath the H1 bushing. Figure 12 shows the transformer and the simulated model in the partial discharge localization (PDL) software.
Visual inspection of the turrets on which the bushing was mounted was recommended, as they are excluded from the simplified model in the software but located in the same general area. The visual inspection showed damage inside the turret. As illustrated in Figure 13, black spots on the corona ring, the pressboard shield of the bushing, and the inner tank of the turret indicated arcing.
The PD measurement supported indications of a failure mode by the DGA, but PD acoustic localization provided additional insights that led to the identification of the failure. After inspection, it was concluded that on-site repair was possible, and shipment of the large transformer to a repair facility was avoided.
Case Study 4: Intrusive On-Line PD Detection + Acoustic Localization
After several failures of identical 7 MVA, 34.5 kV / 0.72 kV transformers installed in different wind farms, the operator requested PD detection and localization on two identical transformers. The DGA analysis of those transformers showed high levels of hydrogen (H2) and acetylene (C2H2).
The main reason for the measurements was to decide whether an on-site repair would be possible, as transformer lead times were increasing. The transformers were in the nacelle of the wind turbines, which made off-line measurements logistically difficult. It was decided to perform on-line measurements as a first attempt and re-assess based on the results. The measurements were done using HFCT and TEV sensors, and acoustic localization was performed. PD was measured on both units, and the acoustic localization revealed a very similar location at the bottom of one winding. The result of the PDL instrument for one transformer is available in Figure 14. Based on this, the operator decided that on-site repairs would not be possible.
Case Study 5: Nonintrusive On-line PD Detection + Acoustic Localization
After a 10 MVA, 72 kV / 13.8 kV transformer was moved from one location to another in a mining facility, DGA analysis revealed an increasing trend of hydrogen (H2) and methane (CH4). An increase in hydrogen concentration along with some methane typically indicates low energy discharges such as PD in gas cavities. This transformer was critical for the operation of the mine; therefore, taking the transformer off-line for an extended period for testing was not feasible. Instead, nonintrusive on-line PD detection was performed using TEV sensors installed on the tank of the transformer.
PD was detected with the TEV, and acoustic PD localization successfully located the PD site inside the transformer tank. The transformer and the results of the PDL instrument are shown in Figure 15. The customer decided to open the transformer to perform a partial inspection. Unfortunately, the PD location was not accessible for inspection with the active part still in the transformer. The owner decided that the transformer would need to be shipped to a repair facility if corrective actions were to be taken.
Case Study 6: On-Line Monitoring via Bushing Tap and UHF
A monitoring system had been installed on a 300 MVA, 426 kV / 15.75 kV power transformer in 2015. The monitoring system continuously monitored the dissipation factor and capacitance of the HV bushings, DGA, and PD charge values of the transformer via the bushing taps and UHF method.
In June 2023, an alarm triggered due to PD charge values on one of the HV bushings. The recorded trend is available in Figure 16. The PD alarm was triggered by measurements from the bushing tap, not from the DGA analysis or the UHF PD sensor.
By analyzing the PRPD patterns, it was confirmed that it was internal PD. Because the alarms pointed to the HV bushing corresponding to the yellow phase, a DGA was performed on the bushing itself and a significant level of H2 was found. Visual inspection further confirmed the findings.
After the bushing was replaced, the PD charge values returned to normal. This case study illustrates the benefits of PD measurements on transformers since alternative testing methods are not sensitive to these types of defects.
CONCLUSION
These case studies highlight the benefits of having various onsite PD testing options to complement other insulation assessment techniques, such as DGA. A combined strategy for onsite PD testing is suggested, as it offers a more thorough evaluation of the transformer’s condition. When it comes to off-line PD measurements, the choice of voltage injection method and PD sensors should be carefully considered based on the specific asset being tested, as well as the logistical and financial implications.
In addition to PD measurement and detection techniques, PD localization methods can provide crucial information and are critical in determining whether repairs can be conducted on-site or if the transformer needs to be sent back to a factory. On-line measurements can serve as a practical alternative when off-line testing cannot be conducted promptly or at all. Moreover, nonintrusive methods can be employed for PD detection and identification when service interruptions are not feasible. Finally, although continuous monitoring may be more expensive, it is beneficial for the early identification of potential failure modes.
REFERENCES
CIGRE. CIGRE Technical Brochure 676, Partial Discharges in Transformers, February 2017. Accessed at https://www.e-cigre.org/publications/detail/676-partial-discharges-in-transformers.html.
IEEE. IEEE Guide for Installation and Maintenance of Liquid-Immersed Power Transformers in IEEE Std C57.93-2019 (Revision of IEEE Std C57.93–2007), pp. 1–61, 29 May 2019, doi: 10.1109/IEEESTD.2019.8713998.
Kraetge, S. Hoek, K. Rethmeier, M. Krüger, and P. Winter. “Advanced noise suppression during PD measurements by real-time pulse-waveform analysis of PD pulses and pulse-shaped disturbances,” 2010 IEEE International Symposium on Electrical Insulation, San Diego, CA, USA, 2010, pp. 1–6, doi: 10.1109/ELINSL.2010.5549723.
Kraetge, K. Rethmeier, M. Krüger and P. Winter, “Synchronous multi-channel PD measurements and the benefits for PD analyses,” IEEE PES T&D 2010, New Orleans, LA, USA, 2010, pp. 1–6, doi: 10.1109/TDC.2010.5484343.
M. Lachance and M. Al-Gunaid, “The use of TEV sensors for on-line PD detection in oil-filled transformers,” 2023 IEEE Electrical Insulation Conference (EIC), Quebec City, QC, Canada, 2023, pp. 1–6, doi: 10.1109/EIC55835.2023.10177345.
Mathieu Lachance joined OMICRON electronics Canada Corp. in 2019 and presently holds the position of Regional Application Specialist for partial discharges, for North America. Lachance previously worked as a test engineer in the fields of partial discharges and high voltage. He actively participates in many IEEE working groups and task forces related to partial discharge measurements on different equipment. Lachance received a BS in electrical engineering from Université Laval and an MS in Applied Science in electrical engineering from École de technologie supérieure.
Fabiana Cirino joined OMICRON electronics Corp in 2018 as an application engineer. She supports the North American Region through customer support, demonstrations, and training in applying OMICRON products. Her current role focuses primarily on partial discharge measurements and their applications. Cirino received a BSc in electrical engineering from the University of Houston and has completed extensive studies in electrical power engineering at RWTH Aachen University in Germany.