Electrical Behavior of Contaminated Distribution Insulators Exposed to Natural Wetting

Transcription

1 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 18, NO. 2, APRIL Electrical Behavior of Contaminated Distribution Insulators Exposed to Natural Wetting Chris S. Richards, Carl L. Benner, Member, IEEE, Karen L. Butler-Purry, Senior Member, IEEE, and B. Don Russell, Fellow, IEEE Abstract Working insulators begin failing as airborne contaminants and moisture from natural wetting combine on insulator surfaces to cause a drop in surface resistivity. This enables current to conduct across the insulators, thereby changing the electrical activity exhibited by the insulators when clean. If the drop in surface resistivity is severe enough, then the leakage current may escalate into a service interrupting flashover that degrades power quality. To help improve power quality, Texas A&M University developed an experimental methodology to investigate the electrical activity of contaminated insulators exposed to natural wetting. Leakage current and weather data obtained during experimentation showed that humidity and rain cause a deviation in the electrical activity of contaminated insulators from that of clean insulators. Analysis of leakage current data showed that this electrical activity was characterized by transient arcing behavior. Further, this nonsteady state activity is small, intermittent, and broad band in nature. Index Terms Arcing, contaminated insulators, leakage current. I. INTRODUCTION CONTAMINATED insulator failure has long plagued transmission and distribution system power quality. The failure process begins when airborne contaminants combine with moisture from fog, rain, or dew to form a conductive pollution region on the insulator surface. This results in a drop in surface resistivity, which enables current to conduct across the insulators. Unless natural cleaning or corrective maintenance occurs, this electrical activity may eventually escalate into an overcurrent fault in the form of a flashover (insulator failure). Flashovers occurring on transmission line insulators adversely affect a much larger number of customers than similar events on distribution line insulators. Consequently, several utilities have invested resources in developing contamination monitors that provide information on the condition of working transmission line insulators, so that corrective maintenance actions can prevent potential flashovers [1] [3]. Since customers historically had no choice in their power provider and since there are fewer customers over which to spread the cost of monitoring equipment on a distribution line relative to a transmission line, similar contamination monitoring on distri- Manuscript received August 1, 2000; revised March 5, This work was supported in part by the Electric Power Research Institute under Grant W C. S. Richards is with ExxonMobil Baytown Oil Refinery, Baytown, TX USA ( christopher.s.richards@exxonmobil.com). C. L. Benner, K. L. Butler-Purry, and B. D. Russell are with the Department of Electrical Engineering, Texas A&M University, College Station, TX USA ( carl.benner@ieee.org; klbutler@ee.tamu.edu; bdrussell@tamu.edu). Digital Object Identifier /TPWRD bution lines has not been a priority. However, in the emerging deregulated environment, individual customers will have the ability to choose utilities that can provide high power quality. Therefore, utilities must now prevent service interruptions resulting from the failure of contaminated distribution line insulators. To help achieve this, the Power System Automation Laboratory (PSAL) at Texas A&M University is investigating contaminated distribution insulators. Previous work has characterized the electrical activity of contaminated insulators wetted in a controlled fog chamber [4]. The next step was to characterize the electrical activity of contaminated insulators exposed to natural wetting. This paper presents the experimental methodology developed to investigate contaminated insulators exposed to natural wetting and results from a characterization of their electrical activity. II. EXPERIMENTAL SETUP The methodology used to investigate the electrical activity of contaminated insulators exposed to natural wetting was to connect 45 contaminated and 45 clean insulators between one phase of a 12.5-kV distribution feeder and the system neutral, and monitor their leakage current behavior over a long period of time (i.e., months). The purpose for the clean insulators was to provide a controlled set of insulators, under the same atmospheric conditions, in which to compare the electrical activity of the contaminated insulators. Four design constraints influenced the development of the experimental setup. The first constraint involved safely mounting the insulators out of human reach. The second constraint involved monitoring the weather conditions affecting the leakage current behavior of the insulators. The third constraint involved minimizing any potential adverse effects to other customers served by the same feeder that supplied the experiments. The final constraint was to protect the experimental equipment from damage. A. Personnel Safety To remain energized while exposed to the atmosphere over long periods of time, the insulators had to be mounted safely out of human reach. The approach used to accomplish this task was to mount the insulators at pole top height. Fig. 1 shows the physical orientation of the insulators under investigation at Texas A&M University s Riverside Campus Downed Conductor Test Facility. The figure shows 90 ANSI class 55-4 porcelain pin-type insulators mounted on nine crossarms that span two existing power poles. The insulators were mounted using standard construction techniques /03$ IEEE

2 552 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 18, NO. 2, APRIL 2003 Fig. 1. Experimental setup. B. Weather Readings To monitor the weather conditions affecting the 90 insulators, a Davis Instruments model # 7440 Weather Monitor II System was mounted adjacent to the insulators on one of the supporting power poles. Fig. 1 shows the weather station and its proximity to the insulators. The weather station provided information such as temperature, humidity, wind speed, and wind direction information. C. Minimizing Effect on Other Customers With the experimental setup being connected to a lateral from a local distribution feeder, it was necessary that the long-term experimentation at the Riverside Campus Test Facility did not affect local customers. The approach used to accomplish this task was to insert a 500- current limiting resistor in the leakage current return path. Therefore, if the leakage current from either set of insulators escalated into a flashover, the flashover current would be limited to 14.4 A. At this current level, no upstream protective devices would operate and cause service interruptions to local customers. In addition, this level of flashover current would have a negligible effect on the line voltage. Although incorporating this resistor into the experimental setup prevented service interruptions to local customers, its presence must not affect the leakage current behavior of the insulators being monitored. Previous experiments using a controlled fog chamber have shown that depending upon the levels of contamination and moisture present on insulators, the level of leakage current could range between a few milliamps to several hundred milliamps. At these minimal levels, the current limiting resistor has a negligible effect on the leakage current being monitored. For example, a 100-mA leakage current would result in a 50-volt voltage drop across the 500- resistor, which accounts for less than one percent of the supply voltage. This small voltage drop will negligibly affect the voltage driving the electrical activity, and therefore, would not distort the leakage current being monitored. D. Equipment Protection In the event that during experimentation a flashover occurred on either set of insulators, 14.4 A would flow through the current limiting resistor. This current level would produce 100 kw of I R heat. With the resistors only rated to withstand this level of heat generation for approximately 1 s, an approach was needed to interrupt this current before permanent damage occurred. The approach adopted was to insert a fused cutout, loaded with a type 2T-time delay fuse, in the leakage current return path. Therefore, during a flashover on the insulators, the type 2T fuse would Fig. 2. Fig. 3. Circuit model of experimental setup and signal sensing. Monitoring approach. cause the fused cutout to open, and thus, interrupt the current in less than 0.5 s. Each of the design constraints influenced the development of the experimental setup. The circuit model for this setup and the sensing equipment used in the monitoring of the leakage current behavior of the two sets of insulators are shown in Fig. 2. The figure shows the connection of both sets of insulators to a lateral from a local 7.2-kV (line-to-neutral) distribution feeder through a fused cutout. To create a leakage current return path, the pins from both sets of insulators were tied together and then connected through two separate but equally rated 500- resistors to the system neutral. The approach used to monitor the leakage current behavior of both sets of insulators is presented in the following section. III. MONITORING APPROACH The approach used to monitor the leakage current behavior of the 45 clean and 45 contaminated insulators consisted of the four distinct steps shown in Fig. 3. The first step included the conversion of the power system quantities (i.e., leakage current and insulator voltage) into proportional electronic level signals. The second step included filtering the electronic level signals and amplifying them to achieve high resolution during the analog-to-digital conversion process. The next step included converting the analog electronic level signals into a digital format. The last step included the analysis of the digitized signals and data logging. A. Conversion of Power System Quantities Conversion of the leakage current into an electronic level signal was accomplished by inserting a Pearson Electronics model #411 current transducer in the leakage current return path, as shown in Fig. 2. This was an air core current transducer with a 0.1 V/amp gain and a passband extending from 1 Hz to 20 MHz. As leakage current flowed across the insulators to the system neutral, the current transducer (CT) created a proportional electronic level signal. With the electrical activity

3 RICHARDS et al.: BEHAVIOR OF CONTAMINATED DISTRIBUTION INSULATORS EXPOSED TO NATURAL WETTING 553 PCI-MIO-16-E-1 12-b A/D card. Both the and voltage signals were sampled at a rate of 1,920 samples/s while the signal was sampled at a rate of samples/s. The remaining signal was sampled by a PCI-6032E 16-b A/D at a rate of 1,920 samples/s. Fig. 4. Signal conditioning approach. being driven by the line voltage from the distribution feeder, a decision was made to also monitor the insulator voltage. This was accomplished by placing a standard 60:1 potential transformer across the insulators between the supplying phase of the distribution feeder and the system neutral. Since the voltage was only needed as a reference for the leakage current, high precision sensing equipment was not deemed necessary. B. Filtering and Amplification of Signals Previous experiments using a controlled fog chamber showed that the electrical activity of wet contaminated insulators was characterized by arcing. This transient behavior of the leakage current resulted in a broadband increase in energy, extending from a dominant 60 Hz into much higher frequencies. Frequency domain analysis of data from these experiments showed that this energy was largely concentrated in the low frequencies below 780 Hz. However, although possessing a much smaller amount of energy, significant activity was also seen in the higher frequencies between 2 to 10 khz. Based upon these results, the signal conditioning approach shown in Fig. 4 was adopted to investigate these same frequency components. The figure shows three separate outputs resulting from the filtering of the electronic level signal from the current transducer. Investigation of the low-frequency leakage current component activity below 780 Hz was accomplished using the first two outputs. The first output was obtained by anti-aliasing the CT input at 780 Hz. The second output allowed these lower frequency leakage current components to be investigated without the dominant 60 Hz present. This was accomplished by notch filtering the CT input at 60 Hz and then anti-aliasing it at 780 Hz. In addition to the low-frequency components, high-frequency leakage current components were also investigated. This was accomplished by filtering the notch filtered CT input with a 2-kHz highpass filter and subsequently with a 10-kHz lowpass filter. Each of these three filtered outputs was then amplified to achieve high resolution when digitized during the analog-to-digital conversion process. C. Analog-to-Digital Conversion Once the electronic level signals were filtered and amplified, they were then converted into a digital format. This was accomplished using two National Instruments data acquisition cards (A/D) installed in a Pentium II 450-MHz PC and specially developed real-time data acquisition software. The and signals along with the insulator voltage were sampled by a D. Analysis and Data Logging Once digitized, the outputs from the signal conditioning stage were analyzed using specially developed real-time analysis software. This software performed a specific operation on each of the three outputs. The first operation consisted of performing a RMS calculation on the signal. The second operation consisted of performing a fast Fourier transform (FFT) on the signal. The final operation consisted of performing an energy calculation on the signal. These operations were performed on each contiguous two-cycle interval of the digitized outputs from the signal conditioning stage. Data from this analysis were used to help determine the atmospheric conditions that influence the electrical activity of the contaminated insulators and also to help investigate this activity. Determining these conditions required tracking the atmospheric data from the weather station and the corresponding behavior of these two cycle leakage current quantities over a duration of months. To accomplish this task without exceeding data storage limitations, statistical values (i.e., mean, minimum, maximum, and standard deviation) were calculated and logged for both the atmospheric data and two cycle leakage current quantities over a 15-min interval. Although this allowed the long term electrical activity to be investigated, it did not provide information on the short-term activity. Investigating this short-term electrical activity was accomplished by implementing simple threshold-based data captures. These data captures were triggered when one of the two cycle quantities exceeded a software programmable threshold. The thresholds used were based upon analysis of the electrical activity of contaminated insulators in a controlled fog chamber. Once triggered, the three digitized outputs from the signal conditioning stage in addition to each of the two-cycle leakage current quantities were continuously logged over a 5-s period of time. IV. CHARACTERIZATION OF ELECTRICAL ACTIVITY For purposes of analyzing results presented in this section, it is important to discuss the preparation of the insulators for monitoring. Insulator preparation began with a thorough cleaning of all 90 insulators. While 45 of the insulators remained clean, the remaining 45 insulators were contaminated in accordance with the Clean Fog Test of the IEEE Standard Techniques for High Voltage Testing [5]. This standard requires the insulators to be contaminated from a slurry composed of 40 g of kaolin, 1000 g of water, and a desired amount of NaCl. Once prepared, a hand held sprayer was used to evenly apply the slurry to both the top and bottom of the insulators. Note that the level of insulator contamination was quantified in terms of equivalent salt deposit density (ESDD), which is directly related the NaCl component of the slurry. In order to quantify an ESDD level for the 45 contaminated insulators, an additional insulator was added to the contamination set for analysis. The analysis involved rinsing

4 554 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 18, NO. 2, APRIL 2003 Fig. 6. Electrical activity of the contaminated and clean insulators at 48% humidity. Fig. 5. Behavior of the contaminated and clean insulators while exposed to natural wetting. the contaminated insulator with deionized water and measuring the conductivity of the rinse water. An ESDD level was then calculated from the resulting conductivity [6]. Once the surface contamination was dry, the insulators were energized and data logging commenced. Data logging continued uninterrupted for several weeks, during which time natural atmospheric conditions were the sole influence on the insulators. After several weeks of logging data, the insulators were de-energized and recontaminated, at a different ESDD level, for additional monitoring. Note that the ESDD levels that are quantified during the insulator preparation only reflect the contamination level at the onset of the data logging. The change in ESDD over the monitoring period was not monitored. Currently, the experimental setup has been logging data for more than nine months, during which time the 45 contaminated insulators were subjected to various levels of contamination represented by equivalent salt deposit density levels. To date, analysis of the logged data has revealed two atmospheric conditions that influence the electrical activity of the contaminated insulators: humidity and rain. This influence is clearly shown by the behavior of the RMS current in Fig. 5. Note that the contaminated insulators referenced in Fig. 5 were contaminated on February 2, 2000 with an ESDD level mg/cm. Four important observations can be made regarding the influence of humidity and rain on the RMS current. First, at low humidity (i.e., 50%) the contaminated insulators are dry. Consequently, there is no moisture available to combine with contamination to lower surface resistivity so that leakage current could flow across the insulator surfaces. As a result, there is no observable difference between the electrical activity of the contaminated and the clean insulators. Second, above approximately 50% humidity, the contaminated insulators exhibited a marked increase in RMS current while the clean insulators remained practically unchanged. This increase in leakage current was attributed to dew, which was enhanced by the presence of hygroscopic NaCl in the surface contamination and the cooling of the insulators by long wave radiation [1], [7]. Third, above approximately 50% humidity, the RMS current strongly correlated with humidity. Fourth, the sudden increase in surface moisture that occurs at the onset of a rain causes a dramatic increase in RMS current on both the contaminated and clean insulators. The behavior of the RMS current during these conditions represents a preliminary characterization of the electrical activity of contaminated insulators exposed to natural wetting. More detailed characterization during each of these conditions will be subsequently presented: low humidity, high humidity, and rain. A. Low Humidity As shown in Fig. 5, the electrical activity of the contaminated insulators during low humidity was characterized by a relatively constant 3-mA RMS current. Leakage current waveforms captured during low-humidity periods revealed that this current was capacitive. An example of these waveforms is shown in Fig. 6. As shown, the leakage current waveform is a well-behaved (i.e., no arcing) steady state waveform that leads the insulator voltage by approximately 90. This 90 phase difference indicates that the leakage current is composed almost entirely of capacitive dielectric charging current. The lack of any significant resistive current suggests that there is insufficient moisture available to combine with the contamination to lower surface resistivity. As a result, there is no measurable current conducting across the surface of the insulators, making the leakage current from the contaminated insulators electrically indistinguishable from that of the clean insulators. This, however, is not the case at high humidity. B. High Humidity As shown by the behavior of the RMS current in Fig. 5, humidity begins to influence the electrical activity of the contaminated insulators as it exceeds approximately 50%. This is when the hygroscopic NaCl contaminant starts absorbing moisture from the atmosphere, consequently causing a drop in surface resistivity that enables current to conduct across the surface of the insulators. As a result, the electrical activity begins to deviate from the steady state capacitive current that was seen at low humidity when the insulators were dry. With the rate of moisture absorption being dependent on the level of humidity, this deviation in electrical activity increases as humidity reaches higher levels. By contrast, the electrical activity of the clean insulators exhibits no significant deviation from its behavior when dry

5 RICHARDS et al.: BEHAVIOR OF CONTAMINATED DISTRIBUTION INSULATORS EXPOSED TO NATURAL WETTING 555 Fig. 8. Frequency spectrum of arc component. Fig. 7. Electrical activity of contaminated and clean insulators at 90% humidity. until reaching very high levels (i.e. 90%). Even at these high levels, without surface contamination present, the deviation in electrical activity is minimal. Examples of the electrical activity of both the contaminated and clean insulators at high humidity are shown in Fig. 7. As shown by the leakage waveform from the contaminated insulators, there is a considerable deviation in electrical activity from the 4-mA steady state 60-Hz capacitive current that was exhibited at low humidity when the insulators were dry. This deviation is the result of a drop in surface resistivity which caused the addition of a 25-mA steady state 60-Hz resistive component and small arc components to the 4-mA capacitive leakage current. The addition of these components has two noticeable effects on the characteristics of the leakage current. First, the addition of a resistive component (i.e., in phase with voltage) to the capacitive leakage current causes the 90 phase difference with the insulator voltage, seen at low humidity, to decrease significantly. As a result, the composite leakage current waveform appears almost in phase with the insulator voltage. Second, the intermittent addition of arc components causes sudden increases in leakage current near the peaks of the insulator voltage. As a result, a transient component is added to the steady state leakage current waveform. The effect of the deviation in electrical activity was not solely seen in the time domain. It is also manifested in the frequency domain as a broadband increase in leakage current energy, which occurs as a result of arcing. This spectral characteristic is shown by the frequency spectrum in Fig. 8. The figure shows a spectrum obtained by calculating an average for each frequency component over several seconds of arcing. The resulting spectrum was then normalized by the fundamental frequency. Two noticeable characteristics can be extracted from the figure. First, the leakage current components are largely odd harmonics. Second, the spectrum roughly decreases with frequency. Although this spectrum only shows the frequency composition of the leakage current up to 930 Hz, the arc spectrum actually extends into much higher frequencies, as illustrated by the behavior of the high-frequency current in Fig. 7. This figure illustrates one last important characteristic of the electrical activity of contaminated insulators at high humidity, which is that bursts of high-frequency energy only result from the arcing behavior. Therefore, no high-frequency activity was exhibited by the contaminated and clean insulators at low humidity and the clean insulators at high humidity, where the leakage current is composed of nonarcing steady state components (i.e., capacitive and/or resistive). By contrast, the clean insulators exhibited only a small deviation in electrical activity from the 4-mA capacitive current seen at low humidity. This deviation was also the result of a drop in surface resistivity, which led to the addition of a 5-mA resistive component. Similar to behavior of the contaminated insulators, the addition of this component also caused the phase difference with the insulator voltage to decrease. However, since its magnitude is significantly smaller, the composite leakage current waveform still retained a significant leading characteristic. As shown by the leakage current behavior just discussed, insulator wetting at high humidity affects the electrical activity of both the contaminated and clean insulators. The effect on the clean insulators was minimal, adding only a small steady state resistive component to the leakage current. The effect on the contaminated insulators, however, was much more severe, adding a much larger resistive component and transient arc components. It is this arcing characteristic which distinguishes the electrical activity of the contaminated insulators from the clean insulators at high humidity. C. Rain As seen by the behavior of the RMS current shown in Fig. 5, the 0.63 inches of rain that occurred on February 22, 20000

6 556 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 18, NO. 2, APRIL 2003 Fig. 10. Frequency spectrum of arc component. Fig. 9. Electrical activity of the contaminated and clean insulators during a 0.63 rain. significantly influenced the electrical activity of the contaminated insulators. This was the result of a drop in surface resistivity caused by the sudden increase in moisture from falling water droplets, which led to a significant increase in current conducting across the surface of the insulators. Similarly, the clean insulators also experienced a change in electrical activity as a result of the rain. However, without contamination available to combine with the colliding water droplets, the drop in surface resistivity, and consequently, the deviation in electrical activity were not as significant as that of the contaminated insulators. Examples of the electrical activity of both the contaminated and clean insulators during the 0.63 rain are shown in Fig. 9. As shown by the leakage current waveform from the contaminated insulators, there is considerable deviation in electrical activity from the 4-mA steady state 60-Hz capacitive current that was exhibited at low humidity, when the insulators were dry. This was the result of a significant drop in surface resistivity, caused by the collision of falling water droplets with the insulators, which enabled current to conduct across the surface of the insulators. As a result, both arc components and steady state resistive components were added to the steady state capacitive leakage current. Comparing the 165-mA leakage current in Fig. 9 with the 4-mA leakage current in Fig. 6, demonstrates that the addition of these components can be quite significant. Besides just the sheer increase in magnitude, the addition of these components also had four noticeable affects on the characteristics of the leakage current waveform. First, similar to the behavior at high humidity, the addition of a large resistive compo- nent to the capacitive leakage current caused the 90 phase difference with the insulator voltage to decrease significantly. With resistive components reaching magnitudes as high as 61 ma, the phase difference decreases to approximately 4, which causes the composite leakage current waveform to appear almost in phase with the insulator voltage. Second, unlike the steady state characteristic of the resistive component at high humidity, the resistive component during the rain exhibited a temporal variation by changing from half cycle to half cycle. This was attributed to the surface resistivity being kept in a constant state of change by the volatile and random collision of falling water droplets with the insulator surfaces. Third, the addition of the arc components, characterized by a sudden increase in leakage current occurring near the peak of the insulator voltage, added a transient characteristic to the steady state leakage current waveform. Fourth, multiple arcs frequently formed in the same half cycle, resulting in a second sudden increase in leakage current magnitude. It is believed that this behavior was the result of arcs forming on multiple insulators during the same half cycle. Similar to the behavior seen at high humidity, the effect of the deviation in electrical activity caused by insulator wetting during rain, was not solely seen in the time domain. It was also manifested in the frequency domain as a broadband increase in leakage current energy, which occurred as a result of arcing. This spectral characteristic of the electrical activity of contaminated insulators during rain is illustrated by the frequency spectrum in Fig. 10. Two important characteristics of the electrical activity can be extracted from the figure. First, the arc s energy is largely concentrated in the leakage current s low-frequency odd harmonics. Second, the arc s energy contribution to the leakage current roughly decreases with increase in frequency. Although the figure only shows this spectral contribution up to 930 Hz, there is also significant activity in the higher frequencies, as shown by the behavior of the high-frequency current in Fig. 9. The figure shows that this high-frequency activity corresponds to the arcing behavior, which is characteristic of the electrical activity of contaminated insulators during rain. By contrast, the deviation in electrical activity of the clean insulators from the capacitive leakage current seen at low humidity when the insulators were dry, was much less severe, exhibiting only a 100% increase in leakage current magnitude as compared to the 4025% exhibited by the contaminated insulators. This deviation was also attributed to a decrease in surface resistivity, which enabled nonarcing resistive current to conduct

7 RICHARDS et al.: BEHAVIOR OF CONTAMINATED DISTRIBUTION INSULATORS EXPOSED TO NATURAL WETTING 557 across the insulator surfaces. Like the behavior of the contaminated insulators, this resistive component also exhibited a temporal variance by varying from half cycle to half cycle. However, without contamination present, this variation was much less severe. It also similarly caused the 90 phase difference with the insulator voltage to decrease noticeably. However, since its approximately 7-mA magnitude was significantly smaller than the 60 ma of the contaminated insulators, the phase difference decreased only to approximately 30 as compared to 4. Therefore, unlike that of the contaminated insulators, the composite leakage current waveform from the clean insulators retained some of the leading characteristic that was exhibited by both the contaminated and clean insulators when dry. The leakage current behavior just discussed, demonstrates that insulator wetting from rain also effects the electrical activity of both the contaminated and clean insulators. The effect on the clean insulators was minimal, resulting in the addition of a small varying resistive component to the leakage current. The effect on the contaminated insulators, however, was much more severe, adding a much larger varying resistive component and large transient arc components. It appears that, like the behavior seen at high humidity, this arcing behavior is the one characteristic, which distinguishes the electrical activity of the contaminated insulators from that of the clean insulators during rain. V. CONCLUSION For purposes of improving distribution system power quality by preventing insulator failure, the Power System Automation Laboratory at Texas A&M University is investigating the electrical activity of contaminated distribution insulators. Previous experiments have characterized the electrical activity of contaminated insulators wetted in a controlled fog chamber. This paper presents the next step in the investigation, which was the development of an experimental methodology to characterize electrical activity when exposed to natural wetting. To date, the developed setup has been logging leakage current data at Texas A&M University s Riverside Campus Test Facility for nine months, over which time the insulators were exposed to various levels of contamination as well as atmospheric conditions. Analysis of this data has revealed the following characteristics of the electrical activity of contaminated insulators during natural wetting from humidity and rain. First, at low humidity, when the insulators are dry, the electrical activity of the contaminated insulators is indistinguishable from the clean insulators and is characterized by a small well-behaved capacitive leakage current. Second, as the insulators become wet from exposure to high humidity, the contaminated insulators exhibit a significant deviation in electrical activity from the small well behaved leakage current seen when the insulators are dry. This deviation is characterized by the addition of both a steady state resistive component and transient arc components to the leakage current. It was shown that the transient arcing component was the one characteristic that most distinguished this electrical activity from that of the clean insulators during high humidity. Third, as the insulators become wet from exposure to rain, the contaminated insulators also exhibit a deviation from their behavior when dry. This deviation was characterized by the addition of a varying resistive component and transient arc components. Once again, the arc components distinguished the electrical activity of the contaminated insulators from that of the clean insulators. The results of this characterization confirm the hypothesis that contaminated insulators exhibit measurable electrical activity when exposed to natural wetting. This activity was shown to be small and transient in nature. Further, it is this transient behavior that distinguishes contaminated insulators from clean insulators when wet. REFERENCES [1] C. N. Richards and J. D. Renowden, Development of a remote insulator contamination monitoring system, IEEE Trans. Power Delivery, vol. 12, pp , Jan [2] N. Sugawara, K. Hokari, M. Hijikata, A. Saito, and K. Yamanouchi et al., Leakage resistance data acquisition system and several data for maintenance of transmission line porcelain insulators along coasts in wet snow, in Proc. 5th Int. Conf. Properties Applicat. Dielectric Materials, Seoul, Korea, 1997, pp [3] K. Iwai, Y. Hase, E. Nakamura, and H. Katsukawa, Development of a new apparatus for contamination measurement of overhead transmission line insulators, IEEE Trans. Power Delivery, vol. 13, pp , Oct [4] C. S. Richards, C. L. Benner, K. L. Butler, and B. D. Russell, Leakage current characteristics caused by contaminated distribution insulators, in Proc. 31st Annu. North Amer. Power Symp., San Luis Obispo, CA, U.S., Oct. 1999, pp [5] Standard Techniques for High-Voltage Testing, IEEE Std [6] W. A. Chisholm, P. G. Buchan, and T. Jary, Accurate measurement of low insulator contamination levels, IEEE Trans. Power Delivery, vol. 9, pp , July [7] C. S. Richards, Development of a Methodology to Discriminate Incipient Insulator Faults From Distribution System Load, M.S. Thesis, Texas A&M University, College Station, TX, May [8] J. G. Proakis and D. G. Manolakis, Digital Signal Processing Principles, Algorithms, and Applications, 3rd ed. Englewood Cliffs, NJ: Prentice- Hall, 1996, p [9] B. D. Russell, Detection of Arcing Faults on Distribution Feeders,, EPRI Final Rep. #EL-2757, Dec [10] B. D. Russell and C. L. Benner, Arcing fault detection for distribution feeders: security assessment in long term field trials, IEEE Trans. Power Delivery, vol. 10, pp , Apr Chris S. Richards received the B.Sc. and M.Sc. degrees in electrical engineering from Texas A&M University, College Station, in 1997 and 2000, respectively. Currently, he is with ExxonMobil s Baytown Oil Refinery, Baytown, TX. From 1995 to 1996, he worked three cooperative work study terms with Arco Chemical, Channelview, TX. From 1997 to 2000, he worked for the Power System Automation Laboratory at Texas A&M University, College Station. Mr. Richards is a member of the Gamma Mu chapter of Eta Kappa Nu. Carl L. Benner (M 88) received the B.Sc. and M.Sc. degrees in electrical engineering from Texas A&M University, College Station, in 1986 and 1988, respectively. Currently, he is a Research Engineer with Texas A&M s Department of Electrical Engineering, College Station, where he has managed the activities of the Power System Automation Laboratory for 14 years. His research interests include the application of advanced technologies to the solution of difficult power system monitoring, protection, and control problems. Much of his work has centered on practical techniques for the sensitive detection of arcing faults on power distribution systems. He holds four patents in this area. Mr. Benner is a member of the IEEE s Power Engineering and Industry Applications Societies.

8 558 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 18, NO. 2, APRIL 2003 Karen L. Butler-Purry (SM 01) is an Associate Professor in the department of electrical engineering at Texas A&M University, College Station. She received the B.Sc. degree in electrical engineering from Southern University, Baton Rouge, LA, in 1985, the M.S. degree in electrical engineering from the University of Texas at Austin in 1987, and the Ph.D. degree in electrical engineering from Howard University, Washington, D.C., in In , she was a Member of Technical Staff at Hughes Aircraft Co., Culver City, CA. Her research focuses on the areas of computer and intelligent systems applications in power, power distribution automation, and modeling and simulation of vehicles and power systems. Dr. Butler is a member of IEEE Power Engineering Society (PES), and the Louisiana Engineering Society. B. Don Russell (F 92) received the B.Sc. and M.E. degrees in electrical engineering from Texas A&M University, College Station, and the Ph.D. degree in power system engineering from the University of Oklahoma, Norman. Dr. Russell is Associate Vice Chancellor and Associate Dean of Engineering, Professor of Electrical Engineering, and Director of the Power System Automation Laboratory of the Electric Power Institute at Texas A&M University, College Station. His research interests include the use of advanced technologies to solve problems in power system control, protection, and monitoring. Dr. Russell is Division VII Director of the IEEE and Past President of the IEEE Power Engineering Society. He chairs the annual TAMU Conference for Protective Relay Engineers and the Substation Automation Conference. He holds several awards and patents for advanced digital technology applications.