AEP Experience with Harmonic Filter Bank Protection

Transcription

1 Initially Presented at 215 WPC AEP Experience with Harmonic Filter Bank Protection Zachary P. Campbell Matthew. Louderback American Electric Power, AEP Inc. Abstract AEP is home to a number of harmonic filter banks. ecently, AEP has installed new or replaced a significant portion of the fleet of the filter bank protection devices. Due to the nature of harmonic filter banks, conventional protection devices sometimes have trouble performing adequately when faced with the challenges present in harmonic filter bank applications. This paper chronicles the challenges that AEP engineers have faced attempting to apply conventional and non-conventional protection practices on recent harmonic filter bank applications. Additionally, at one facility AEP engineers have employed a novel unbalance protection method using a conventional relay for protection of a harmonic filter bank. The unbalance element is described in detail and TDS simulations are shown which highlight the element response to various common system challenges present at harmonic filter bank applications. Keywords harmonic filter bank, unbalance protection, TDS I. INTODUCTION Over the past three years AEP has either installed new or upgraded the protection associated with a large number of AC harmonic filter banks. Harmonic filter banks assist the power system with reduction of non-fundamental voltage distortion present on the bulk electric grid [1]. This is necessary to guarantee that power quality delivery is within tolerable limits. The banks operate such that they provide a low impedance path to ground a certain harmonic, or set of harmonics sources. On the AEP Transmission system these banks have generally been situated immediately adjacent to some source of voltage distortion which generates harmonic content in the power system. Locations where these sources are common include Flexible AC Transmission System (FACTS) devices and High- Voltage Direct Current (HVDC) locations. AEP owns and operates over 36 harmonic filter banks. Though AEP is home to numerous harmonic filter banks, each one has been installed with its own suite of protection methods and equipment. At some locations the protection is developed within the FACTS or HVDC control system, and at others this protection is developed within some conventional relaying platforms. This paper chronicles some challenges that AEP engineers have encountered in the development of protection on these devices, and establishes a new unbalance detection principle. A. AEP Facility One II. EXISTING AEP INSTALLATIONS At one facility the protection methods that are employed on a C type, shunt ungrounded harmonic filter bank are made up of overcurrent elements and a single overvoltage based unbalance detection element. erence [2] discusses this application in significant detail. The unbalance detection principle employed at this facility is a common method of unbalance protection which is used to sense individual capacitor element failures. This method is used to detect both element failures within the main section of the filter bank as well as failure of elements within the tuning section. In this instance the capacitor can type making up both the main and tuning section capacitances are shorting type. The protection had to be set up such that the capacitor can elements even in the tuning section would not be exposed to excessively high voltage. After proper evaluation, it came to be that the number of element failures within a single can which would generate unacceptably high voltage across the remaining can elements would be 3. This meant that the unbalance voltage that would be present for necessary trip conditions would be 75V (primary). Under balanced system conditions and without harmonic distortion, sensing this level of voltage would be difficult. This is because the relay was attempting to sense a less than 1 Volt deviation in secondary voltage caused by capacitor element failure. Unfortunately, the presence of the SVC at the facility in combination with an arc furnace made the application of an unbalance method dependent upon the detection of a fundamental frequency overvoltage element even more difficult. Fig. 1 shows secondary voltage levels of the harmonic and inter-harmonic content under normal operating conditions. The figure illustrates that under normal conditions, significant harmonic distortion is present across a broad range of inter-harmonics. Clearly the levels of nonfundamental voltage signal present exceed the unbalance element setting quantity. Fig.1 Harmonic Content with Furnace and SVC in Operation Several relays were tested for proper operation in this application. Successful or unsuccessful operation of the relays was initially based upon a combination of the signal processing task(s) being performed to both filter the fundamental voltage information and the minimum setting quantity which would have operated to detect the low signal level. The device which was finally selected for operation, and which has been in service for nearly two years, is a device which makes use of a sliding window FFT to obtain Pg. 1 of 9

2 fundamental frequency information. Other devices which were tested, but were unsuccessful when attempting to operate in the environment had been based on a correlation type FI filter, often known as the cosine filter, and another known as a sine filter. Since both filters have significant signal leakage in their inter-harmonics regions, both types were susceptible to the inter-harmonics noise present normally at the facility. Fig. 2 illustrates the mentioned cosine and sine filter frequency response characteristics. GAIN Fig. 2 Cosine and Sine Digital Filter Frequency esponses B. AEP Facility Two At the previously mentioned facility, the unbalance function which was developed was performing a protection function which acted to protect both the main section capacitance of the harmonic filter bank and the tuning section capacitance. At another facility, an unbalance function was developed solely for the tuning section capacitance. A singleline diagram of the filter bank of interest at this facility is shown in Fig. 3, where the tuning section capacitance is shown as element C2. C1 CT-H Nominal Frequency frequency, Hz C2 Upper Envelope Lower Envelope L1 CT- CT-B Fig.3 Harmonic Filter Bank Single Line Diagram At this facility the main section capacitance, C1, is protected using an H-Bridge based configuration with an overcurrent element based on the H-Bridge mid-section current (measured by CT-H), a common configuration for capacitor banks [3]. The tuning section protection employed something that many conventional relays cannot accomplish. The tuning section unbalance method measures the bank current and the resistor section current. Based on the current in the bank (measured by CT-B), the current that should flow thru the resistor (measured by CT-) is estimated. This estimated quantity and the measured quantity then make up a difference calculation. If there is some difference it is assumed that this is caused by a current diversion in the tuning section caused by element failures within the tuning section capacitance. This difference drives alarming and tripping functions. The way that the method estimates the current in the resistor branch is by using a digital filter to emulate the behavior of the current sharing relationship between the bank current and the resistor current. It can be shown that the relationship between bank current, I B, and resistor current, I, is that of Eq. 1. ()= Eq. 1 This relationship is estimated in protection using a second order Chebychev filter. Fig. 4 illustrates the frequency response of Eq.1 and of the estimated relationship established using the Chebyshev filter. Gain Chebyshev Filter Current Sharing elationship X: 6 Y: 4.129e Frequency (Hertz) Fig. 4 Frequency esponse of Sharing elationship between Bank Current and esistor Current in Filter Bank and Digital Filter Estimated Behavior As Fig. 4 illustrates, the resistor branch of the filter bank suppresses nominal frequency current. This means that nominal frequency bank current does not flow thru the resistor, and if the resistor CT detects nominal frequency current than a problem likely exists. In addition, as Fig. 4 illustrates, the digital filter emulates the behavior of the current sharing relationship between the bank current and the resistor current with a large degree of success. It should also be noted that most conventional relaying platforms cannot perform this task, making this protection method inextensible to most modern conventional protective relaying platforms. This is because it is generally not possible for the user to configure the signal processing tasks to affect the way that raw data samples are processed. C. AEP Facility Three At a third facility, only the protection and control devices associated with four doubly-tuned filter banks were being Pg. 2 of 9

3 replaced. The existing filter bank configurations, showing measuring core locations, are shown in the Fig. 5. CT- CT-P C1 L1 C2 L2 CT-C21 CT-C22 CT-N Fig. 5 Facility Three Existing Doubly Tuned Filter Bank Single Line The initial installation was placed in service in the middle of the 199 s, when some of the modern practices associated with filter bank protection had not yet been established [3]. This being the case, the bank was designed such that the main section capacitance, which is made up of externally fused cans, was not set up as an H-Bridge. The existing unbalance protection method measured phase voltage and phase current. The phase voltage was then used to calculate the theoretical phase current. If there was some difference between the calculated phase current and the measured phase current, alarming or tripping was initiated. Fig. 6 illustrates the behavior of the mentioned, existing unbalance element. performance of the element. The following section documents the improved method. III. CAPACITO UNBALANCE DIFFEENTIAL ELEMENT A. Introduction to Unbalance Differential In an attempt to improve upon the unbalance method concept illustrated at AEP facility three, in the previous section, the following protection concept has been developed for use at that facility for capacitor bank and filter bank unbalance detection. Considering the generic wye-grounded, externally fused capacitor bank of Fig. 7, the relationships established in Eq.2 thru Eq. 7 will hold. C can V A p s V B p s V C I A I B I C CT p s elay PT C CB ωc X - Counter Alarm/ Trip Fig. 6 Facility Three Main Capacitance Unbalance Element Description As can be seen in Fig. 6, the unbalance element attempts to correct for changes in the capacitance of the cans due to temperature. It does this using a moving average calculation of the difference between the calculated phase current and the measured phase current which is fed back into the initial calculation to attempt to remove the effects of small errors in differential magnitude. If the difference in this calculation results in a number outside of a small hysteresis window, a counter will increment. If the total count exceeds some preset limits alarms or trips will be generated. Operating experience with this protection element has shown that the moving average calculation has allowed for can failures to not be detected by this element. Though unproven, this is largely thought to be due to the fact that the moving average calculation can be fooled into thinking partial can failures are acceptable. In addition, operating experience has shown that the presence of harmonic content in the phase current of the filter bank leads to phasor measurement inaccuracies which would require the protection element to be desensitized to avoid mal-operation. To improve upon this behavior a different method is being employed to improve upon the = Fig. 7 Wye-Grounded Fused Capacitor Bank = =3 Eq. 2 Eq. 3 =3= Eq. 4 =3 Eq. 5 Eq. 6 can be rewritten as Eq. 7. =3 Eq. 6 3 = Eq. 7 Based on the relationship shown in Eq. 7 an unbalance detection principle can be established which calculates the difference between the measured residual current of the capacitor bank and a calculated estimate, using bank impedance as a constant value. However, we must include an error term in the difference calculation. This is a requirement so that phase impedance differences, caused by capacitor can manufacturing differences, do not result in non-zero difference calculations. Therefore, a final unbalance detection principle is settled into, illustrated in Eq. 8, where ϵ is an error term which accounts for natural phase impedance imbalance. 3 Eq. 8 Pg. 3 of 9

4 The principle shown, which we shall call the operating quantity, in Eq. 8 can be used to detect can failures within a capacitor bank or harmonic filter bank because, at the fundamental frequency, a harmonic filter bank impedance is made up, primarily, of the main section capacitance. Additionally, establishing an unbalance method wherein the operating quantity is compensated using system voltage improves upon many common capacitor bank unbalance protection elements whose methods do not employ some form of compensation [4][5][6]. It is necessary when employing this method to consider the location of the voltage instrument transformers. This is because when the capacitor bank is offline, there will be no current flowing into the capacitor bank. If this is the case, there will be no 3I ELAY quantity for the unbalance differential element to use for restraining. If the voltages are connected on the bus side instead of being located on the capacitor bank side of an interrupting device, the differential quantity will likely result in a non-zero quantity. If the voltage source is, however, located on the capacitor bank side of the interrupting device the differential quantity will likely result in a near-zero result, depending on the error term magnitude. Therefore it is necessary when developing a protection element to consider the source of the voltages used in the protection element. B. Element Investigation and Development Using the relationship established in Eq. 8, and with ϵ set to zero, if one fuse within phase A operates to remove a failed can from the capacitor bank of Fig. 7, it can be shown that the phasor current which would develop from the principle established in Eq. 8 is that illustrated in Eq. 9. ( ) Eq. 9 Note that the VA term, which is a phasor quantity, will help to identify the phase of unbalance. If another phase was the phase housing the failed can its term would have emerged. Eq. 9 also provides us with the amplitude of operating principle when one can fails in the phases. If one divides this quantity by nominal bank line to ground voltage multiplied by can capacitance the sensitivity characteristic for the minimum number of can failures can be evaluated by analyzing deviations of bank size parameters s and p. Fig. 8 illustrates the effects of bank size on minimum operating quantity. As Fig. 8 shows that as the multiplier of capacitor can capacitance and voltage applied to bank increases, so does operating quantity. However, as both bank size parameters increase, operating quantity decreases. To illustrate how this figure can be used consider the facility where this method is being deployed where s=12, p = 1, and the voltage level is 345kV with can capacitance being micro-farad. Therefore, expected operating quantity magnitude, in primary Amps, will be that shown in Eq. 1. This filter bank is considered large; with ratings of 1MVar and 345kV. Using this protection method at this facility stretches the applicability of the principle due to sensitivity concerns, but, as the reader will see, the operating principle still adequately protects the bank. magnitude of operating quantity (1/(VCcan)) ( )(2.882)=1.535[] Eq Parallel Units / Cans (p) Series Groups parallel cans (s) (p) series groups (s) Fig. 8 Sensitivity Characteristic of Operating Quantity Magnitude Based on Deviations in Capacitor Bank Size Parameters At AEP, capacitor banks have traditionally been designed to operate near their steady state voltage capability. This means that a well-developed AEP philosophy exists for externally fused capacitor banks which require that an alarm be provided to personnel when one can is out of service and a trip is issued for the capacitor bank or harmonic filter bank when two cans have failed. Though we have defined the phasor quantity which will be developed from the operating principle in Eq. 8 for a single can failure, we want to also be able to define a relationship for when a second can is out of service to abide by established AEP philosophy. There are three possible scenarios that can occur for the second can failure: A.) 1 more can fails on the same parallel group of cans B.) 1 more can fails of same phase, but not within same group C.) Can failure occurs in a different phase of the bank For each of the scenarios listed above, the following operating quantity phasor relationships, Eq. 11- Eq. 13, will develop from the principle of Eq. 8, with ϵ set to zero. It should be noted that initial can failure in these scenarios are associated with phase A. In scenario C, the second can failure occurs in phase B. A.) B.) ( ( C.) ( ) X: 12 Y: 1 Z: ) Eq. 11 ) Eq. 12 ( ) Eq. 13 Of these three scenarios, we will need to be able to sense the smallest amplitude increase in the unbalance principle so as to guarantee proper sensitivity for the second can failure. This means that we need to establish the smallest operating quantity that evolves from the three previously mentioned scenarios. By inspection, it can be seen that scenario C will 5 Pg. 4 of 9

5 generate no increase in operating quantity magnitude beyond the operating quantity magnitude that develops for the initial can failure. Only a shift in the phase angle of the operate quantity will occur for this scenario. Both scenario A and B will generate an increase in operating quantity magnitude with the same phase angle as the original operate quantity. It can be shown that the smaller condition from the two is condition B. This means that in order to develop the protection element based on the principle of Eq. 8, both magnitude and phase angle must be considered. Table 1 illustrates all possible can failure scenarios and the phase angle which would develop as a result of either the failure of one can or two cans (if cross phase failure occurs). Table 1 Phase Angle of Operating Quantity Developed From Can Failure Phase(s) of Can Failure(s) Angle of Operating Quantity A 9º B -3º C -15º A&C 15º A&B 3º B&C -9º Based on the relationships established so far, an unbalance element can be established. To illustrate how this element would work, consider the illustration of the unbalance differential element shown in Fig. 9. When there is little magnitude, with any phase angle, of the differential, the element will neither detect a single can failure nor will it detect two can failures. Highset Trip Threshold 2 Can Threshold 2 Cans A 1 Can A No Operate 2 Cans B&C Highset Trip 1 Can Threshold Fig. 9 Unbalance Differential Element Characteristic Based on previous analysis and as Fig. 9 shows, the operating characteristic relies on the magnitude and phase angle of the operating quantity. When the operating quantity calculation yields a result which lies within the green zones, shown as the 1 Can regions, the relay will identify one can having failed, and can generate an alarm for the condition. When the operating quantity calculation yields a result which lies within the yellow zones, shown as 2 Cans, the relay will identify two cans having failed and will trip will the bank with some time delay. If the magnitude of the operating quantity exceeds the yellow region magnitude and enters the red region shown as the Highset Trip region, the relay will trip with a short time delay of less than 1 cycles. The highset trip region is used in the detection of a rack-to-rack fault. Based on the information obtained from the Table 1 results, phase identification is possible using the phase angle which results from the operating quantity calculation. The reason that the phase angle windows are set up in the way they are is to capture all scenarios where the operating quantity angles are within a known window. These windows are 9º wide for the 1 can detection window which is wider than the 2 can detection windows because this condition creates only an alarm. The windows are 3º wide for the 2 can detection region because this scenario will generate a trip. It is desirable to be more selective when a trip is issued. This increases the security of the method. The windows being setup in this way establishes a secure but dependable method for can failure detection. In addition, these windows are fixed values and do not need configured to set up the protection element. The logic which is used to detect for the conditions mentioned are setup to detect for six different phase angle windows, each of which will indicate a specific phase or phases in which can failure has occurred. The logic also determines which conditions will generate a trip and those which will generate an alarm. As mentioned earlier, the protection element must consider the source of the voltage measurements. The logic that has been developed for this application uses an instantaneous overcurrent element set to.5 per unit of nominal phase current to supervise the operation of the unbalance differential element. Scheme logic for the element can be seen in Appendix A. C. Simulations To test the method that has been established, a eal Time Digital Simulator (TDS) was used to inject signals into a relay which was programmed with the aforementioned unbalance differential method. The capacitor bank which was developed for the test has a bank configuration equivalent to Fig. 6 with parameters p=1, s=12, C CAN = 2.674μF, with a system voltage of 345kV, CT ratio of 6/1, and PT ratio of 3/1. With these bank parameters it is now possible to calculate the necessary element parameters. The following analysis is provided for unbalance differential element settings development. 1 Can Threshold (From Eq. 9): =1.53[] It is usually desirable to set the pickup of an element for a value less than the maximum theoretical current magnitude. This is generally a preference of the user, so we shall choose 2 times pickup. Therefore, the setting of the 1 Can Threshold becomes.76 Amps. 2 Can Threshold (From Eq. 12): Pg. 5 of 9

6 =3.4[] As noted earlier, because it is usually desirable to set the pickup of an element below the maximum theoretical current magnitude we shall do so again with this setting. However, this threshold must be greater than the 1 Can Threshold with enough margin to ensure proper operation of the element. In this case we shall set this threshold to 75% of calculated amplitude. With this decision, the 2 Can Threshold setting pickup is now set to 2.3 Amps. Highset Trip Threshold: The highset trip threshold, as noted earlier, should be set to detect rack-to-rack faults. It can be shown that for a rack-torack fault on phase A, the phasor relationship of Eq. 14 will develop from the operating quantity. ( ) Eq. 14 Using the relationship established in Eq. 14 it is possible to establish a setting quantity for the highset trip threshold as the following, which ignores phase angle =15.2[] Once again, the pickup of this threshold shall be set to some user preferred sensitivity level, which at AEP is often 2 times pickup, making the pickup of the element 7.5 Amps. These settings are sufficiently large so as not to require additional coordination among the other settings of the unbalance differential. Now that all of the settings which define the unbalance differential element have been established, it is possible to illustrate element behavior. Fig. 1 illustrates the behavior of the element within the relay when a single phase A can fails. Detection of a single can failure is usually allowed a couple of minutes, but, in the case of the simulation the one can out logic has been programmed with a 3 cycle time delay. This allows the reader to note can failure inception. Fig. 1 shows that 3 cycles after the measured 3I current increases suddenly, the relay declares a single can failure has occurred on phase A. Fig. 1 elay Oscillography Capturing 1 Can Failure on Phase A As Fig. 1 shows, when the phase A can failed the calculated unbalance differential quantity magnitude rose to nearly 1.5 Amps, while the calculated unbalance differential phase angle approached nearly 92 degrees. This magnitude and phase angle results in activating the logic which makes up the green region of Fig. 9 noted as 1 Can A. This is because the operating quantity angle lies within the window established between 75 and 15 degrees, and the magnitude of this quantity lies between the one can threshold and the two can threshold. Suffice it to note to the reader that element behavior for the detection of phase B and C single can failures operates in a similar manner. For the second can failure, it has been established that we must be able to detect cross phase can failure as well as multiple can failure within the same phase. Fig. 11 illustrates the relay behavior for a phase C can failure after an initial phase B can failure. Fig. 11 elay Oscillography Capturing 2 Can Failures on Phases B & C The oscillography of Fig. 11 shows that at the 2 cycle mark, a single can within phase C fails. Prior to this a can within phase B had failed. As can be seen, the second failure does not generate an increase in the unbalance differential quantity magnitude. Instead this second can failure only changes the angle developed from the unbalance differential quantity. In fact, the angle of the calculated unbalance differential changes from nearly -3 degrees to -9 degrees. This change shifts the activated element logic from the green region noted in Fig. 9 as 1 Can B to the yellow region noted as 2 Cans B&C. This shift generates a trip condition and fault clearing occurs near the 37 cycle mark. This behavior is as expected and confirms earlier analysis. Assertion of the second can detection logic will trip the capacitor bank and annunciate for the appropriate phases of can failure. This behavior can be seen in the digital channels of Fig. 11. It should be noted that though second can failure will trip the capacitor bank, it usually does so after a longer time delay than that shown in Fig. 11. The standard time delay for operation was shortened to illustrate to the reader element behavior only. Once again, suffice it to note to the reader that element behavior for the failure of two cans across multiple other phases (phase A&C and phases A&B) will operate in a similar fashion. Now that element behavior for cross phase can failure has been illustrated, the same should be done for second can failure within the same phase. Fig. 12 illustrates this circumstance for the second can failure occurring on phase C. As the figure illustrates, at the 2 cycle mark the second can within phase C fails. This generates an increase in the Pg. 6 of 9

7 unbalance differential magnitude while the angle of the unbalance differential quantity stays constant until the interrupting device of the bank clears, at the 37 cycle mark. In fact, the amplitude changes from nearly 1.5 Amps to 3 Amps while the phase angle is maintained at nearly -15 degrees, all of which confirms earlier analysis. The change in magnitude increase of the unbalance differential causes the activated element logic to shift from the region noted in Fig. 9 as the green region noted as 1 Can C to the yellow region noted as 2 Cans C. Once again, this shift generates a trip condition and expected behavior from previous analysis. For the rack to rack fault shown in Fig. 13 it can be seen that at the 28 cycle mark the measures a large increase in measured 3I which is uncompensated in the unbalance differential by any reduction in any phase of the system voltages. This generates nearly 15A of unbalance differential magnitude with an angle of 15 degrees, confirming earlier analysis. This resultant calculation allows the highset trip region of Fig. 9 to be activated which results in a trip with short time delay, as can be seen in the figure. To illustrate element resiliency several tests have been formulated to ensure security of the element. The first to be illustrated is element behavior when faced with a close-in external fault. Fig. 14 illustrates the behavior C Fig. 12 elay Oscillography Capturing 2 Can Failures on Phase C It should be noted that though second can failure will trip the capacitor bank, it usually does so after a longer time delay than that shown in Fig. 12. The standard time delay for this type of operation was shortened to illustrate to the reader element behavior only. At this point the reader should recognize that element behavior for the failure of two cans on the same phase of the other phases (phase B and phase C) will operate in a similar fashion. The last function of the unbalance differential element that has not yet been illustrated is the element behavior for rack-torack faults. Fig. 13 illustrates the relay and element behavior for this circumstance. Fig. 13 elay Oscillography Capturing ack to ack Fault on Phase B Fig. 14 elay Oscillography Capturing External Fault Event As can be seen in Fig. 14, when the external phase A fault occurs, at the 35 cycle mark, system voltage on the effected phase becomes depressed which generates an increase in 3V. This voltage depression generates current unbalance in the measured residual current shown as the channel designated Measured 3I. The increases in 3I and 3V occurring simultaneously, and with proper magnitude and phase, allow the unbalance differential calculation to only result in a minor increase in magnitude. Because the resultant magnitude lies below the highset trip threshold, the user can rely on the trip time delay of the remaining unbalance differential element, the two can out detection region, to restrain a trip issuance. In this way, security of the element is established for external faults. The second illustration of element resiliency can be established by analyzing element behavior when immersed in signal distortion caused by the presence of harmonic content. Fig. 15 illustrates this circumstance. Because the presence of signal distortion alone does not necessarily guarantee any unanticipated behavior, a carefully crafted signal was generated. It is well known that the signal processing tasks which occur on many conventional protective relaying platforms rely on digital filters to remove extraneous signal information. These filters generally perform well across a variety of applications because they have been built so as to reduce purely harmonic content present on the signal to zero. Pg. 7 of 9

8 However, as can be seen in Fig. 2, the inter-harmonic regions of these filters allow for nominal frequency signal misrepresentation [7] Therefore, if inter-harmonic voltage distortion was injected into the capacitor bank under test, some of the signal distortion caused by this should be present on the current channels and the voltage channels which are reading the raw data. Because the effects of this distortion on the current channels and voltage channels are not equal, the unbalance element may not behave as expected. Therefore, as Fig. 15 shows, 4% voltage distortion at 15 Hertz (2.5 times nominal frequency) does generate some amount of calculated unbalance differential magnitude. As can be seen this amount is nearly.5 Amps. Though this quantity is below the lowest pickup of the unbalance differential element, it does illustrate that depending on the amount of signal distortion and the frequency range of the distortion, there can be some impact on the element. At the facility where this method is being employed, THD is being limited to 4%. Even if all of the voltage distortion was lumped into the inter-harmonics region allowing for maximum signal leakage, it has been shown that this will not generate a relay mal-operation. Any improvement over this hypothetical phenomenon will generate an improvement in relay performance. It is more likely that there will not be 4% THD at the facility, lumped within the one inter-harmonic region. behavior are provided. The protection element exhibits adequate security, sensitivity, dependability, and resiliency in the face of common capacitor bank and harmonic filter bank challenges. EFEENCES [1] J. C. Das, Power System Harmonics and Passive Filter Designs, 1 st Ed., New Jersey: Wiley, pp [2] T. Ernst, M. Louderback, U. Khan, and P. Parikh, Protection Challenges of a Second Harmonic Capacitor Bank, unpublished, Presented at the 41 st Annual Western Protective elaying Conference, Spokane, WA, October 214. [3] IEEE Standard, C , IEEE Guide for the Protection of Shunt Capacitor Banks. [4] S. Samineni, C. Labushagne, J. Pope, Principles of Shunt Capacitor Bank Application and Protection, unpublished, Presented at the 64 th Annual Georgia Tech Protective elaying Conference, Atlanta, GA, May 21. [5] G. Brunello, B. Kasztenny, C. Wester, Shunt Capacitor Bank Fundamentals and Protection, unpublished, Presented at the Texas A&M 23 Conference for Protective elaying Engineers, Colege Station, Tx, April 23. [6] B. Kasztenny, J. Schaefer, E. Clark, Fundamentals of Adaptive Protection of Large Capacitor Banks, unpublished, available online: capacitor_banks.pdf [7] I. Voloh, D. Finney, Impact of Frequency Deviations on Protection Functions, Preseneted at the Texas A&M Conference for Protective elay Engineers, Austin, TX, April 29. BIOGAPHIES Zachary P. Campbell received his B.S.E.E. degree from the University of Akron, in Akron, Ohio, in 28, and his M.Sc. degree from The Ohio State University, in Columbus, Ohio, in 212. He has been an engineer at American Electric Power (AEP) since 28, working in various capacities within protective relaying departments including field services and engineering. Zak is a member of IEEE, CIGE and is a registered professional engineer in the state of Ohio. zpcampbell@aep.com Matthew. Louderback eceived B.S.E.E. from the University of Dayton and is a registered professional engineer in Ohio. From 28 to 212, Mr. Louderback worked for Dayton Power and Light as a relay engineer. He is currently working for American Electric Power (AEP) as a Protection and Control Standards Engineer. mrlouderback@aep.com Fig. 15 elay Oscillography During 4% Voltage Distortion at 15Hz IV. CONCLUSIONS Several challenges that AEP engineers have faced when providing protection for harmonic filter banks have been shown. It has been highlighted that some implementations of protection methods which exist for harmonic filter banks cannot be implemented on some modern conventional protection devices. This is largely due to the fixed nature of the signal processing tasks onboard these devices. Towards this end, the authors encourage manufacturers to allow for the modification to these pillars. Additionally, a conventional protective relay platform s flexibility was extended to create a newly formed capacitor or filter bank main section unbalance method, which can be used on externally fused bank types. Element analysis is shown and simulations illustrating Pg. 8 of 9

9 APPENDIX A 1 Can Out Threshold 3Icalculated 3Imeasured 135 Angle(3Icalc. 3Imeas.) Can Out Threshold Highset Threshold 5P ΦA ΦB ΦC ΦA/C ΦA/B ΦB/C H A T Other Trips Target eset Target eset Target eset S Q 1 Can Out Detected S Q A Phase Target eset S Q B Phase Target eset S Q C Phase Target eset S Q 2 Can Out Detected Trip Logic S Q Highset Trip Target Pg. 9 of 9