The industry has always sought better tools to assess the general condition of high voltage equipment and identify potential problems. Today’s microprocessor technology provides a sophisticated means to capture information. While much of this technology has already been integrated into off-line techniques for over a decade, it also adapts well to on-line measurement. One of the primary challenges to the on-line approach is the management of a tremendous amount of data. Fundamentally, the desired result of the diagnostic test is to determine the status of the apparatus, and if an anomaly is detected, how critical is it, and how soon will it require attention? This level of information analysis cannot be achieved with a simple monitoring system; it requires an expert system, which can digest all of the data being generated. It is this expert system, which differentiates an on-line monitoring system from an on-line diagnostic approach. The intent of this article is to outline how on-line diagnostics can be applied to evaluating bushings and also to share some applications experiences.
Off-line diagnostics have served the industry well and they continue to play an important role in the reliability of the power system. The off-line application has some inherent flaws - the obvious one being the need to remove the apparatus from service -and the other flaw being the inability to identify serious changes taking place between test intervals. If the gestation period (time from normal condition to failure) is shorter than the test interval, then the test result will never exhibit the problem. The frequency of the on-line measurements provides the ability to determine the rate of change, knowing if the change was gradual, over a long period of time or sudden, as well as when the change took place. This is all information that can be used to determine the appropriate course of action. An apparatus owner may decide to tolerate an abnormality that has stabilized and let the expert system determine when the situation becomes more critical or plan corrective action at a more optimal time. To achieve this level of confidence, one must place a great deal of trust in the expert system. The expert system must distinguish between noise and actual change, identify a wide array of problems at the incipient state, and provide evidence why an alarm is being issued, as well as some type of traceability. An expert system should not rely solely on thresholds set from conventional offline methods, which traditionally have been conservative due to the limitations of the approach. The diagnostics should be capable of learning the specific characteristics of the individual apparatus and not rely solely on user limits or average values from other similar apparatus. The expert system should be capable of learning normal behavior of the apparatus being evaluated and apply this to the analysis.
Why apply on-line diagnostics to bushings? Bushings are certainly not the most expensive piece of equipment in a substation, so the financial loss of a bushing failure is not the driving force. However, the damage that a failed bushing can inflict on its affiliated apparatus could indeed be catastrophic. Specifically, bushing failures leading to a damaged power transformer have been well documented. The role of the bushing subjects them to high dielectric, thermal, and mechanical stresses, which tends to make bushings one of the most vulnerable components of major apparatus. Most bushing failures can be attributed to internal deterioration or contamination and being able to detect these irregularities is essential to maintaining a stable system. Many of the same dynamics affect the performance of current transformers. For stand-alone current transformers the insulation system is very similar to bushings making it feasible to apply the same diagnostic tools.
The industry has accepted the conventional off-line power factor/capacitance test as the most reliable tool for identifying problem bushings and current transformers. It is the success of the power factor/ capacitance measurement in bushing diagnostics and the awareness of the catastrophic results of bushing failure that have lead industry experts and insurance providers to recommend more frequent testing of bushings. These guidelines are in conflict with the current philosophy of the industry, which is to minimize down time. The concept of bushing/current transformer online diagnostics combines the advantage of the power factor/capacitance test with the ability to perform the measurements on-line, without interruption of service.
One approach to determining in-service condition of bushings is to calculate the imbalance current measured at their tap for a three-phase set, Figure 1A. The sum current method is based on the principal that in a symmetrical three phase system, the sum of the voltage and current vectors is zero, Figure 1B. This allows the condition of bushings to be determined by vectorially adding the currents measured at the bushing taps. If the bushings are identical and system voltages are perfectly balanced, then the sum current will equal zero. In this situation, the expert system would only need to rely on the most recent recordings to determine the condition of the bushings. Since bushings/current transformers are never identical and system voltages are never perfectly balanced, the sum current is a non-zero value, which is unique to the bushing/current transformer set. As a result, the sum current is a vector unique to that bushing set, Figure 1C. The expert system establishes a benchmark sum current during an initial learning period, which is then compared to the configuration data, which consists of the last offline measurement and/or nameplate data. This configuration is also used in the analysis to determine the present power factor and capacitance of the problem bushing/current transformer.
This benchmark value of sum current is compared to subsequent measurements. Subtracting the benchmark from the latest measured sum current provides a third phasor, which is referred to as the ‘change in sum current’, Figure 1D. The angle of this third vector with respect to the reference bushing is used to identify which bushing is causing the change. Once the deteriorated bushing is known, the magnitude and phase of the change in sum current vector is used to calculate the change in capacitance and power/dissipation factor of this bushing. The quadrature component of the change in sum current is used to calculate the change in capacitance while the in-phase element can be attributed to a change power/dissipation, Figure 1E.
This exercise calculates the absolute values of the power factor and capacitance of the dominant bushing (the bushing experiencing the greatest degree of degradation). This essentially replicates the off-line tests, provided the test conditions are the same. The advantage of the on-line approach is the frequency of data points, which provides the means to determine the rate of change. The expert system arrives at this information by performing a least-squared fitting on a subset of consecutive power factor and capacitance values accumulated over a specified period of time. This exercise will produce a quadratic polynomial equation that, if plotted, would generate a curve that provides a “best fit” through a series of points, in this case power factor and capacitance. By representing a series of power factor and capacitance values as a polynomial, the analysis can use applied mathematical tools to determine the stability of the situation. From this information, the expert system can reach a more informed conclusion on the criticality of the incident.
This technique can also be applied to standalone current transformers with and without taps. If the current transformers are equipped with taps the imbalance current is calculated using the tap current, similar to bushings. To apply the sum current approach to current transformers without taps typically requires electrically isolating the current transformer from the ground grid with the exception of one grounding point. The current measured at the grounding point is used in place of the tap current to calculate the imbalance current.
In the last discussion, it was suggested that the absolute power factor measurement using the online sum current approach duplicates the off-line measurements, provided the test condition were similar. The key point in this statement is the test condition. The experience from off-line diagnostics indicates that bushing problems effecting power factor measurements are accentuated at elevated voltage and temperature. The influences of voltage and temperature on good and deteriorated insulation are illustrated in Figure 2A and 2B. The plots track the power factor using the conventional off-line C1 (center conductor to tapped layer) insulation of two bushings as test conditions (voltage and temperature) are varied. The two specimens are both115KV bushings of the same manufacturer, type, and vintage - one is considered to have good, low power factor while the other has deteriorated high power factor. Figure 2A shows no change in power factor for the good bushing as the voltage is ramped up to rated voltage, while the deteriorated bushing exhibited an increase. A similar behavior is observed, Figure 2B, when the power factor of the two bushings was measured at increasing temperatures. This phenomenon was realized in an on-line situation when two sets of three bushings, one set containing a degraded bushing, were correlated to the top oil temperature of the transformer they were installed on. Figure 3 shows seven days of top oil temperature plotted with sum current data from two sets of Type U bushings in the same transformer. The middle plot is for the high-side (230-kV) bushings demonstrated "good": 0.3%, 0.32%, and 0.29%. The top plot is for the low-side (138-kV) bushings with elevated power factor values: 0.73%, 0.9%, and 1.08% (phase A, B and C respectively). The influence of temperature (lower plot) on the sum current is clearly evident, providing on-line confirmation that deteriorated bushings experience greater fluctuations caused by change in the operating environment.
Case Story
The high power factor associated with the 138-KV bushings was realized during the commissioning of the on-line diagnostic system. The bushings were General Electric, Type U, manufactured in 1972, installed on a 224 MVA transformer. Bushings that exhibited this level of power factor would typically be removed from service, especially the phase C bushing, which registered a power factor greater than three times the nameplate. In this situation, spare bushings were not available; thus, it was decided to rely on the expert system to monitor the situation. This would provide an opportunity to prove that this technology was not only capable of detecting a problem, but also would permit the user to continue operating with a known problem. As long as the situation remained stable, a high power factor bushing could remain in service. In order to be successful, the expert system would have to overcome the thermal fluctuation to be able to track the rate of change. The on-line diagnostic system was installed in April ’97, and the system continued to monitor the bushing for more than a year without issuing any alerts. In May the following year, the transformer began running hotter than usually; this triggering the expert system to issue an alert based on a significant change in sum current. The expert system identified the power factor on the phase C bushing to be the cause; it also noted there was no significant change in the bushing capacitance. This suggested that the problem with the phase C bushing was contamination. With spare bushings available, the utility elected to replace the bushing. The plan is for removed phase C bushing to be energized again in a protected environment with an on-line diagnostic system installed and run to failure, in order to learn more about the behavior of bushings just prior to failing.
This example provides an opportunity to point out the importance of the rate of change. If the expert system relied solely on a limit, it is very likely that it would have triggered alarms soon after installation, requiring the asset owner to make a decision based solely on absolute values. In order to arrive at an intelligent conclusion about the bushing's condition, the power factor and capacitance values along with rate of change needed to be available.
The intent of this discussion is not to lobby for the exclusion of the traditional method of apparatus testing, but to show that in certain applications on-line diagnostics can be a better alternative. The conventional off-line power factor test for bushings and current transformers continues to be an excellent tool to evaluate their condition, however there are situations where its inherent constraints cannot be tolerated. Bushings which are tied to critical system apparatus or which cannot be readily removed from service are candidates for on-line diagnostics. The on-line diagnostics should offer the possibility of duplicating the traditional tests, as well as taking advantage of the large sample size to perform trending and projection of potential problems.
Reference:
Mark F. Lachman, Stephen Skinner and Wolf Walter, “Experience with On-Line Diagnostics and Life Management of High Voltage Bushings” Proceeding of the Sixty-Six Annual International Conference of Doble Clients, 1999, Sec 3-4
Mark F. Lachman, Wolf Walter, and Philip A.Van Guggenberg, “Experience with Application of Sum Current Methods to On- Line Diagnostics of High Voltage Bushings and Current Transformers,” Proceeding of the Sixty- Fifth Annual International Conference of Doble Clients, 1998, Sec 3-5
Donald T. Angell, Reneé B. Kringel, Stephen Skinner, “Report of On-Line Diagnostics at Idaho Power,” Proceeding of the Sixty-Fifth Annual International Conference of Doble Clients, 1998, Sec 3-6
Paul J. Griffin, Dennis J. Kopaczynski and Mark H. Rivers, “Continuous Diagnosis: The Ultimate Defense Against Failures” Proceeding of the Sixty-Fourth Annual International Conference of Doble Clients, 1997, Sec 1-2
About the Author
Robert Brusetti, P.E., received his BS degree from the University of Vermont in 1984 and a Masters in Business Administration from Boston College in 1988. He has been employed at Doble Engineering Company for the past twelve years and is currently Product Manager. Prior to his present position he worked as a field engineer and assisted in the development of the expert system for the Insite online diagnostic system. Mr. Brusetti is a licensed Professional Engineer in the state of Massachusetts.
Reprinted with permission of the International Electrical Testing Association (NETA) Copyright Fall 2002.
Off-line diagnostics have served the industry well and they continue to play an important role in the reliability of the power system. The off-line application has some inherent flaws - the obvious one being the need to remove the apparatus from service -and the other flaw being the inability to identify serious changes taking place between test intervals. If the gestation period (time from normal condition to failure) is shorter than the test interval, then the test result will never exhibit the problem. The frequency of the on-line measurements provides the ability to determine the rate of change, knowing if the change was gradual, over a long period of time or sudden, as well as when the change took place. This is all information that can be used to determine the appropriate course of action. An apparatus owner may decide to tolerate an abnormality that has stabilized and let the expert system determine when the situation becomes more critical or plan corrective action at a more optimal time. To achieve this level of confidence, one must place a great deal of trust in the expert system. The expert system must distinguish between noise and actual change, identify a wide array of problems at the incipient state, and provide evidence why an alarm is being issued, as well as some type of traceability. An expert system should not rely solely on thresholds set from conventional offline methods, which traditionally have been conservative due to the limitations of the approach. The diagnostics should be capable of learning the specific characteristics of the individual apparatus and not rely solely on user limits or average values from other similar apparatus. The expert system should be capable of learning normal behavior of the apparatus being evaluated and apply this to the analysis.
Why apply on-line diagnostics to bushings? Bushings are certainly not the most expensive piece of equipment in a substation, so the financial loss of a bushing failure is not the driving force. However, the damage that a failed bushing can inflict on its affiliated apparatus could indeed be catastrophic. Specifically, bushing failures leading to a damaged power transformer have been well documented. The role of the bushing subjects them to high dielectric, thermal, and mechanical stresses, which tends to make bushings one of the most vulnerable components of major apparatus. Most bushing failures can be attributed to internal deterioration or contamination and being able to detect these irregularities is essential to maintaining a stable system. Many of the same dynamics affect the performance of current transformers. For stand-alone current transformers the insulation system is very similar to bushings making it feasible to apply the same diagnostic tools.
The industry has accepted the conventional off-line power factor/capacitance test as the most reliable tool for identifying problem bushings and current transformers. It is the success of the power factor/ capacitance measurement in bushing diagnostics and the awareness of the catastrophic results of bushing failure that have lead industry experts and insurance providers to recommend more frequent testing of bushings. These guidelines are in conflict with the current philosophy of the industry, which is to minimize down time. The concept of bushing/current transformer online diagnostics combines the advantage of the power factor/capacitance test with the ability to perform the measurements on-line, without interruption of service.
One approach to determining in-service condition of bushings is to calculate the imbalance current measured at their tap for a three-phase set, Figure 1A. The sum current method is based on the principal that in a symmetrical three phase system, the sum of the voltage and current vectors is zero, Figure 1B. This allows the condition of bushings to be determined by vectorially adding the currents measured at the bushing taps. If the bushings are identical and system voltages are perfectly balanced, then the sum current will equal zero. In this situation, the expert system would only need to rely on the most recent recordings to determine the condition of the bushings. Since bushings/current transformers are never identical and system voltages are never perfectly balanced, the sum current is a non-zero value, which is unique to the bushing/current transformer set. As a result, the sum current is a vector unique to that bushing set, Figure 1C. The expert system establishes a benchmark sum current during an initial learning period, which is then compared to the configuration data, which consists of the last offline measurement and/or nameplate data. This configuration is also used in the analysis to determine the present power factor and capacitance of the problem bushing/current transformer.
This benchmark value of sum current is compared to subsequent measurements. Subtracting the benchmark from the latest measured sum current provides a third phasor, which is referred to as the ‘change in sum current’, Figure 1D. The angle of this third vector with respect to the reference bushing is used to identify which bushing is causing the change. Once the deteriorated bushing is known, the magnitude and phase of the change in sum current vector is used to calculate the change in capacitance and power/dissipation factor of this bushing. The quadrature component of the change in sum current is used to calculate the change in capacitance while the in-phase element can be attributed to a change power/dissipation, Figure 1E.
Figure 1
This exercise calculates the absolute values of the power factor and capacitance of the dominant bushing (the bushing experiencing the greatest degree of degradation). This essentially replicates the off-line tests, provided the test conditions are the same. The advantage of the on-line approach is the frequency of data points, which provides the means to determine the rate of change. The expert system arrives at this information by performing a least-squared fitting on a subset of consecutive power factor and capacitance values accumulated over a specified period of time. This exercise will produce a quadratic polynomial equation that, if plotted, would generate a curve that provides a “best fit” through a series of points, in this case power factor and capacitance. By representing a series of power factor and capacitance values as a polynomial, the analysis can use applied mathematical tools to determine the stability of the situation. From this information, the expert system can reach a more informed conclusion on the criticality of the incident.
This technique can also be applied to standalone current transformers with and without taps. If the current transformers are equipped with taps the imbalance current is calculated using the tap current, similar to bushings. To apply the sum current approach to current transformers without taps typically requires electrically isolating the current transformer from the ground grid with the exception of one grounding point. The current measured at the grounding point is used in place of the tap current to calculate the imbalance current.
In the last discussion, it was suggested that the absolute power factor measurement using the online sum current approach duplicates the off-line measurements, provided the test condition were similar. The key point in this statement is the test condition. The experience from off-line diagnostics indicates that bushing problems effecting power factor measurements are accentuated at elevated voltage and temperature. The influences of voltage and temperature on good and deteriorated insulation are illustrated in Figure 2A and 2B. The plots track the power factor using the conventional off-line C1 (center conductor to tapped layer) insulation of two bushings as test conditions (voltage and temperature) are varied. The two specimens are both115KV bushings of the same manufacturer, type, and vintage - one is considered to have good, low power factor while the other has deteriorated high power factor. Figure 2A shows no change in power factor for the good bushing as the voltage is ramped up to rated voltage, while the deteriorated bushing exhibited an increase. A similar behavior is observed, Figure 2B, when the power factor of the two bushings was measured at increasing temperatures. This phenomenon was realized in an on-line situation when two sets of three bushings, one set containing a degraded bushing, were correlated to the top oil temperature of the transformer they were installed on. Figure 3 shows seven days of top oil temperature plotted with sum current data from two sets of Type U bushings in the same transformer. The middle plot is for the high-side (230-kV) bushings demonstrated "good": 0.3%, 0.32%, and 0.29%. The top plot is for the low-side (138-kV) bushings with elevated power factor values: 0.73%, 0.9%, and 1.08% (phase A, B and C respectively). The influence of temperature (lower plot) on the sum current is clearly evident, providing on-line confirmation that deteriorated bushings experience greater fluctuations caused by change in the operating environment.
Figure 2
Figure 3
Case Story
The high power factor associated with the 138-KV bushings was realized during the commissioning of the on-line diagnostic system. The bushings were General Electric, Type U, manufactured in 1972, installed on a 224 MVA transformer. Bushings that exhibited this level of power factor would typically be removed from service, especially the phase C bushing, which registered a power factor greater than three times the nameplate. In this situation, spare bushings were not available; thus, it was decided to rely on the expert system to monitor the situation. This would provide an opportunity to prove that this technology was not only capable of detecting a problem, but also would permit the user to continue operating with a known problem. As long as the situation remained stable, a high power factor bushing could remain in service. In order to be successful, the expert system would have to overcome the thermal fluctuation to be able to track the rate of change. The on-line diagnostic system was installed in April ’97, and the system continued to monitor the bushing for more than a year without issuing any alerts. In May the following year, the transformer began running hotter than usually; this triggering the expert system to issue an alert based on a significant change in sum current. The expert system identified the power factor on the phase C bushing to be the cause; it also noted there was no significant change in the bushing capacitance. This suggested that the problem with the phase C bushing was contamination. With spare bushings available, the utility elected to replace the bushing. The plan is for removed phase C bushing to be energized again in a protected environment with an on-line diagnostic system installed and run to failure, in order to learn more about the behavior of bushings just prior to failing.
This example provides an opportunity to point out the importance of the rate of change. If the expert system relied solely on a limit, it is very likely that it would have triggered alarms soon after installation, requiring the asset owner to make a decision based solely on absolute values. In order to arrive at an intelligent conclusion about the bushing's condition, the power factor and capacitance values along with rate of change needed to be available.
The intent of this discussion is not to lobby for the exclusion of the traditional method of apparatus testing, but to show that in certain applications on-line diagnostics can be a better alternative. The conventional off-line power factor test for bushings and current transformers continues to be an excellent tool to evaluate their condition, however there are situations where its inherent constraints cannot be tolerated. Bushings which are tied to critical system apparatus or which cannot be readily removed from service are candidates for on-line diagnostics. The on-line diagnostics should offer the possibility of duplicating the traditional tests, as well as taking advantage of the large sample size to perform trending and projection of potential problems.
Reference:
Mark F. Lachman, Stephen Skinner and Wolf Walter, “Experience with On-Line Diagnostics and Life Management of High Voltage Bushings” Proceeding of the Sixty-Six Annual International Conference of Doble Clients, 1999, Sec 3-4
Mark F. Lachman, Wolf Walter, and Philip A.Van Guggenberg, “Experience with Application of Sum Current Methods to On- Line Diagnostics of High Voltage Bushings and Current Transformers,” Proceeding of the Sixty- Fifth Annual International Conference of Doble Clients, 1998, Sec 3-5
Donald T. Angell, Reneé B. Kringel, Stephen Skinner, “Report of On-Line Diagnostics at Idaho Power,” Proceeding of the Sixty-Fifth Annual International Conference of Doble Clients, 1998, Sec 3-6
Paul J. Griffin, Dennis J. Kopaczynski and Mark H. Rivers, “Continuous Diagnosis: The Ultimate Defense Against Failures” Proceeding of the Sixty-Fourth Annual International Conference of Doble Clients, 1997, Sec 1-2
About the Author
Robert Brusetti, P.E., received his BS degree from the University of Vermont in 1984 and a Masters in Business Administration from Boston College in 1988. He has been employed at Doble Engineering Company for the past twelve years and is currently Product Manager. Prior to his present position he worked as a field engineer and assisted in the development of the expert system for the Insite online diagnostic system. Mr. Brusetti is a licensed Professional Engineer in the state of Massachusetts.
Reprinted with permission of the International Electrical Testing Association (NETA) Copyright Fall 2002.
