According to reports from the field, bushing failure usually accounts for from 15 to 25% of all transformer failures. To improve reliability and safety of their transformers, utilities therefore look to online monitoring solutions to manage and prevent such failures. Online condition assessment is an effective tool to identify early signs of degradation and prevent catastrophic failures of bushings.

C57.143 – IEEE’s Guide for Application of Monitoring Equipment to Liquid Immersed Transformers and Components, for example, provides a comprehensive list of techniques. Perhaps most notable among these are the Sum of Currents (SoC) and Voltage Reference (VR) Methods.

This edited contribution to INMR comes from Mark Tostrud at Dynamic Ratings. He explains the pros and cons of each method and compares their sensitivity, accuracy, complexity of implementation and susceptibility to system disturbances. Case studies and simulation results are also presented to highlight the strengths as well as limitations of each method. Findings provide practical insight for utility engineers to further enhance transformer condition monitoring strategies.

Sum of Currents (SoC)

The Sum of Currents (SoC) method relies on Kirchhoff’s Current Law to detect insulation degradation. The sum of currents is equal to the sum of the bushing leakage current vectors, as illustrated in Fig. 1. In a perfectly balanced system, where C1 power factor and capacitance values of bushings are identical, the sum of leakage currents will equal zero.

If a system is not perfectly balanced, however, or if the C1 Power Factor and Capacitance values of bushings are not identical, the equation for the sum of currents must account for some level of imbalance current (as in Fig. 2). At time of commissioning, this initial imbalance current is measured by the system and filed.

As shown in Equation 1, a change in C1 Power Factor or in Capacitance will drive a change in the sum of the imbalance current, which can then be used to indicate that there is some change in bushing condition.

In utility systems before the advent of renewables generation, a transmission system was typically balanced and rarely deviated since the magnitude and vector angle of system voltages were tightly regulated by generator controls. Minor variations in system voltages normally seen on such a traditional utility grid would not be sufficient to drive significant changes in imbalance current. As such, the sum of currents method performed well.

Systems with large amounts of generation from renewables or with excessive harmonics can have system voltages that are not as balanced, i.e. there can be frequent deviations in system voltage magnitude and vector angles. If these variations occur at the same time each day or only during certain system loading conditions, the system can be configured to accommodate these variations to prevent false alarms. However, if deviations are random and affect one phase more than another, they can trigger false alarms from systems using the SoC method or result in errors in calculating relative Power Factor and Capacitance values.

The Sum of Current Method requires a signal from each of the bushings being monitored to be wired to a central location. With banks of single-phase transformers, this could prove a challenge. Should the bank of single-phase transformers also includes a spare, the system must be re-balanced if the configuration of a single-phase transformer changes.

Voltage Reference Method

The Voltage Reference Method more closely matches offline Power Factor and Capacitance testing. However, instead of energizing bushings at 10 kV using a voltage source in the test instrument, the method requires inputs from voltage transformers, potential transformers or CCVTs installed on the same bus as the bushings.

This method measures leakage current of each bushing relative to the voltage at which the bushing is energized. Since both voltage and current are being measured, as well as phase angle between voltage and current, the system can directly calculate actual Power Factor and Capacitance of each bushing.

Any voltage deviations are accounted for in system measurements, resulting in an accurate measurement of the resistive and reactive components of bushing leakage current. Once these values are known, actual bushing Power Factor and Capacitance can be directly calculated. Since system voltage deviations are being measured, any changes in the resistive or reactive components of bushing leakage currents can be attributed to changes in Power Factor or Capacitance of the bushings.

Although Power Factor and Capacitance values are more accurate than the relative Power Factor and Capacitance values derived from the Sum of Current Method, values may still not match those from offline tests. This is because voltage and temperature dependencies may exist with bushing Power Factor measurements. As a result, when monitoring bushings continuously online, intermittent changes in Power Factor and Capacitance can occur that offline testing is unable to detect.

When using the Voltage Reference Method, the system can also monitor health of CCVTs. Since the design of a CCVT is similar to that of bushings, these are prone to the same types of failure modes. Monitoring health of CCVTs, in addition to bushings, allows users to identify which component of the system may be degrading. The Voltage Reference Method is also not impacted by system anomalies, harmonic distortion or by changes in power flow direction. That makes it ideal for today’s power grids.

To realize most benefit from the Voltage Reference Method, a reference should be provided for each bushing being monitored. The reference should come from the same bus from which the bushing is being energized. Phase-to-ground voltage references are preferred.

While many voltage reference systems have the ability to phase angle compensate the reference, this should be avoided since it will result in measurement errors. On autotransformers without an OLTC, using the same reference for both high and low side bushings is acceptable. Any phase shift that exists from the HV to LV side of the transformer should be stable and will have minimal impact on accuracy of readings.

One of the best applications for the Voltage Reference Method is single phase transformers. Providing a voltage reference for the bushings on single phase transformers may be easier than wiring the bushing currents from each phase to a common location. This is especially true when the bank of single-phase transformers includes a spare since the Voltage Reference Method eliminates need to re-balance when changes are made to configuration of the transformer bank
.

Case Study 1

A U.S. utility had a 138/69kV autotransformer with oil-impregnated paper bushings that suddenly went into alarm due to high imbalance current and high imbalance rate of change. Based on rate of change shown in online data, it seemed unlikely this problem would have been identified before the bushing and transformer had failed.

In this situation, the monitoring system had been installed for about a year when, suddenly, imbalance percentage increased dramatically over a few days. Fig. 4 shows that the imbalance current alarm was due to a change in the X2 bushing leakage current.

Data was sent for assistance in analyzing the situation. Detailed analysis identified that the change in imbalance current was due to a change in the X2 bushing Capacitance and Power Factor (shown in Fig. 5). The 13% change in bushing C1 Capacitance was of particular concern since the most likely cause for a decrease of such magnitude is loss of oil in the bushing. Based on the large drop in Capacitance, an immediate inspection of bushings was recommended to look for signs of low oil or oil leak.

During inspection, oil was found on the side of the transformer and the X2 bushing was also found to be low on oil (see Fig 6). The rapid oil loss in the bushing coincided with a through fault on the transformer during a thunderstorm.

An outage was requested and the transformer was removed from service to perform offline testing. While the offline testing showed a problem with the bushing, the offline tests did not reveal the urgency of the problem. The C1 Power Factor had increased to 1.11% since the previous test and the C1 Capacitance was slightly lower at approximately 98% of nameplate.

This example highlights the sensitivity of the online bushing monitor vs. offline testing. When testing a bushing under stress, it is more likely to show anomalies. C2 measurements could not be performed on the bushing since the test set would trip out once the tap was energized.

As shown in Equation 1, Sum of Currents can detect both changes in tan delta (Power Factor) or Capacitance which provides early warning of changes in bushing health. Since the Sum of Current Method does not have a reference voltage for the bushings, the Power Factor and Capacitance of the bushings cannot be directly measured. Rather, the method relies on an analytic in the monitor to calculate the relative tan delta (Power Factor) and Capacitance of the bushings. The analytic uses the imbalance magnitude, imbalance vector as well as the individual bushing leakage currents to identify changes in the tan delta and Capacitance of the bushings.

Relying on only the imbalance magnitude and imbalance vector could result in errors since an increase or decrease in bushing leakage currents will drive changes in Sum of Currents. Fig. 8 shows the polar diagram of the Sum of Currents for the above case study.

The polar diagram leads to believing that the X3 bushing is driving alarms from the monitor. The polar diagram works well to identify which bushing is degrading when the Power Factor or Capacitance of the bushings is increasing. However, when they are decreasing, the polar diagram can be misleading.

Having a clear understanding of what is driving the imbalance current helps interpretation. When considering bushing leakage currents in addition to the Sum of Currents, it becomes clear that the X2 bushing is degrading and driving the alarms. The analytic for the relative tan delta and Capacitance correctly identified that the X2 bushing had been degrading and that the failure mode was a decrease in the C1 Capacitance and an increase in the C1 Power Factor.

Since the relative values are calculated using a moving average, sufficient data must be collected to accurately calculate the values. Figs. 4 and 5 illustrate that while the Sum of Currents (imbalance) and bushing leakage currents show the changes immediately, changes in relative tan delta and Capacitance values are delayed.

Comparison of Sum of Currents & Voltage Reference Methods

To analyze how the Voltage Reference Method would compare for this failure, a failure simulation was carried out. With the monitor configured to perform both sum of current and voltage reference measurements, a simulator was used to inject signals into the monitor.

A voltage reference equivalent to the nominal output from the secondary of the 69 kV CCVTs was wired to the monitor. The simulator was then configured to inject currents equal to the bushing leakage currents recorded during the failure.

Figs. 9 & 10 compare results. The simulation showed that the voltage reference method will begin to show the changes in C1 Power Factor and capacitance of the bushings immediately and the changes correlate with the imbalance current from the sum of currents.

The voltage reference method also clearly showed the full magnitude of the change as well as the trend in changes. By contrast, changes in relative tan delta and capacitance for the Sum of Currents Method are delayed. Even after the relative measurements begin to show the change, the full magnitude of the change in tan delta and Capacitance may not be reflected in the values.

While the Sum of Currents Method provided alerts early in the failure process, it required additional analysis to identify the root cause due to the delay in the reporting of the tan delta and Capacitance values. For failures that occur gradually over time, this delay is unlikely to cause problems. For failures that occur quickly, waiting to react to changes in relative Power Factor and Capacitance could result in catastrophic bushing and transformer failure.

When reviewing Sum of Currents data following an alarm, the leakage currents (magnitude and angle) will better correlate to changes in the sum of currents than the relative tan delta and capacitance measurements. With the Voltage Reference Method, both the leakage currents and the Power Factor and Capacitance measurements will have direct correlation with any changes in imbalance percentage.

Case Study 2

A bushing monitoring system using both the sum of currents and voltage reference method was installed to monitor the 220 kV bushings on a 220/69/10 kV transformer. Prior to the installation, the bushings were power factor tested and found to be in good condition. Shortly after the monitor was installed, the system began to alarm due to changes in the sum of currents and high Power Factor values. Data was sent out for review and analysis.

The polar diagram in Fig. 11 shows that the sum of currents is highly erratic. When this occurs, analysis of the Sum of Currents data can be difficult since one cannot easily identify whether the erratic imbalance current is due to multiple bushings degrading at the same time or if system anomalies are driving the erratic readings. Having voltage reference information available makes this analysis easier. While it is unusual to have three bushings failing on a transformer at the same time, the bushings in this case study were resin-bonded paper bushings. High Power Factors, changes in Capacitance as well as partial discharges are relatively common with this style of bushing.

Fig. 12 graphs the sum of currents and the bushing voltage references. The spikes in imbalance percentage do not show any correlation with changes in the voltage references. Since system anomalies can be ruled out, spikes in imbalance percentage can be attributed to changes in bushing power factor and/or capacitance.

Without the voltage reference signals, it would have been difficult to conclude this from the sum of current data alone.

Fig. 13 shows the Power Factor values from the Sum of Currents and Voltage Reference Methods. From these graphs it is easy to observe that changes in imbalance current are due to changes in the Power Factor of all three bushings. As in the above case study, while the Sum of Currents Method indicates changes in Power Factor in relative tan delta measurements, Voltage Reference Power Factor measurements show stronger correlation.

Fig. 14 indicates that, in addition to changes in Power Factor, changes in C1 Capacitance of the bushings are also driving spikes in imbalance current.

Conclusions

Bushing failure remains a leading cause of transformer failure. While certain styles of bushings are more susceptible to fail, failure can occur with any design.

Most bushing failures can be detected through continuous online monitoring. When deploying online monitoring, it is important to develop an understanding of the failure modes of the equipment being monitored as well as how the monitoring equipment will respond to these.

The Sum of Currents Method has been used to successfully identify degrading bushings for more than 30 years. There are thousands of installations worldwide that utilize this method and hundreds of bushings (and transformers) have been saved through early detection of degraded bushings. All signals required are available on the transformer, meaning installation is more cost-effective than other methods, especially on three phase transformers. The method is also sensitive to changes in both power factor and capacitance.

While a polar chart view of the Sum of Currents is an excellent tool for determining which bushing is experiencing problems, a polar chart only works if the Power Factor and/or Capacitance of the bushings are increasing. If the Power Factor or Capacitance of bushing(s) are decreasing, the polar chart can be misleading. Additional analysis techniques such as trending leakage currents (magnitude and phase angle) along with the imbalance percentage is the best means to confirm what is being displayed on a polar chart.

The Voltage Reference Method offers the capability to rely solely on Power Factor and Capacitance measurements of bushings for analysis and has been used to identify defective equipment for almost 100 years. It is a proven and well understood tool to identify bushings that are starting to degrade. While this method offers more accurate Power Factor and Capacitance measurements, values from the online monitor may not always match offline tests. Power Factor and Capacitance in degraded bushings are often voltage as well as temperature dependent. Defects are more likely to show when bushings are under stress. An online monitor will identify defects much sooner than an offline test.

Systems with large amounts of renewables generation or systems with excessive harmonics can have system voltages that are not as balanced or have frequent deviations in system voltage magnitudes and vector angles. Since system voltage anomalies could result in false alarms from the Sum of Currents Method, the Voltage Reference Method should be considered for these applications. This method should also be considered for single-phase transformer applications since it eliminates need to wire the signals from each single-phase tank to one location.

REFERENCES
[1] Sokolov, V. and Vanin, B., “Evaluation and Identification of Typical Defects and Failure Modes of 110-750kV Bushings”, Proceedings of the Sixty-Fourth Annual International Doble Conference, 1997.

[2] Sokolov, V. and Vanin, B., “Evaluation and Identification of Typical Defects and Failure Modes of 110-750kV Bushings”, Proceedings of the Sixty-Fourth Annual International Doble Conference, 1997.

[3] IEEE C57.143:2024, IEEE Guide for Application of Monitoring Equipment to Liquid-Immersed Transformers and Components.

[4] Tostrud, M. “Failure Characteristics of OIP vs. RIP Bushings”, Proceedings of the Eighty-Fourth Annual International Doble Conference, 2017.

[5] Tostrud, M. “Why Offline Bushing Tests May Not Confirm Alarms from Online Bushing Monitoring Systems”, TechCon 2017.

[6] Tostrud, M. and Schneider C. “Lessons Learned from Online Bushing Health Monitoring”, TechCon 2020.

[7] Tostrud, M. “Unusual Bushing Failure Modes Detected with Continuous Online Monitoring, INMR 2023.