508 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 24, NO. 2, APRIL 2009

Transcription

1 508 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 24, NO. 2, APRIL 2009 Extended Fault-Location Formulation for Power Distribution Systems Rodrigo Hartstein Salim, Student Member, IEEE, Mariana Resener, Student Member, IEEE, André Darós Filomena, Member, IEEE, Karen Rezende Caino de Oliveira, Student Member, IEEE, and Arturo Suman Bretas, Member, IEEE Abstract In this paper, an extended impedance-based fault-location formulation for generalized distribution systems is presented. The majority of distribution feeders are characterized by having several laterals, nonsymmetrical lines, highly unbalanced operation, and time-varying loads. These characteristics compromise traditional fault-location methods performance. The proposed method uses only local voltages and currents as input data. The current load profile is obtained through these measurements. The formulation considers load variation effects and different fault types. Results are obtained from numerical simulations by using a real distribution system from the Electrical Energy Distribution State Company of Rio Grande do Sul (CEEE-D), Southern Brazil. Comparative results show the technique robustness with respect to fault type and traditional fault-location problems, such as fault distance, resistance, inception angle, and load variation. The formulation was implemented as embedded software and is currently used at CEEE-D s distribution operation center. Index Terms Fault diagnosis, fault location, power distribution faults, power distribution protection. I. INTRODUCTION E LECTRIC power systems are constantly exposed to faults, which affect the system s reliability, security, and delivered energy quality. Different stochastic events may cause systems faults, such as lightning, insulation breakdown, and trees falling across lines. Protection schemes are important to maintain system stability and minimize consumer and network damages as well as economical losses. In these aspects, fault-location techniques represent an important role in the fast and reliable power system restoration process. Fault location in electric power distribution systems (EPDS), however, due to its specific topological and operational characteristics, still presents challenges [1] [3]. Still today, fault lo- Manuscript received October 29, 2007; revised January 17, First published March 04, 2009; current version published March 25, This work was supported in part by Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) and in part by Companhia Estadual de Distribuição de Energia Elétrica do Rio Grande do Sul (CEEE-D). Paper no. TPWRD R. H. Salim and K. R. C. de Oliveira are with the University of São Paulo (EESC-USP), São Carlos, SP , Brazil ( salim@sel.eesc.usp.br; caino@sel.eesc.usp.br). M. Resener and A. S. Bretas are with the Federal University of Rio Grande do Sul, Porto Alegre, RS , Brazil ( mariana@ece.ufrgs.br; abretas@ece.ufrgs.br). A. D. Filomena is with the Companhia Estadual de Geração e Transmissão de Energia Elétrica (CEEE-GT), Porto Alegre, RS , Brazil ( afilomena@ece.ufrgs.br). Digital Object Identifier /TPWRD cation in EPDS is often made through visual inspection, field methods, and brute force methods [1]. These techniques are not feasible on underground systems and are very time consuming in large distribution networks. Therefore, a specific fault diagnosis method for EPDS becomes very important and useful. Recently, several fault-location methods for transmission [1], [4] [6] and distribution systems [1], [3], [7] [10] have been proposed. These methods, however, do not totally consider the characteristics of distribution systems (unbalanced operation, presence of intermediate loads, laterals/sublaterals, and timevarying load profile), which significantly affect the methods performance. These characteristics have already been considered before, but are not united in a single optimized fault-location formulation. The fault-location methods described in [3], [7], and [10] consider the existence of different laterals and sublaterals in the distribution system; however, only a single loading characteristic is considered. These methods also evaluate the different paths by the analysis of series and parallel impedances, which are not suitable for large distribution systems. In [8], a similar technique is applied by using direct circuit analysis and different loadings are considered for the same system, achieving high accuracy in the fault-location estimate. The load compensation technique applied in [8] considers measurements in each load point. These measurements, however, are not always available in distribution systems, especially in the aged ones. In this way, its practical implementation becomes limited. Still concerning the previously analyzed works, it is noted that these methodologies were not fully extended to all fault types, such as line-to-line (L-L), double-line-to-ground (DLG), and three- faults (3PH). This extension is extremely important, since distribution systems are subjected to different faults types, and a unified technique must be provided in order to enhance the fault-location estimate reliability. Considering the aforementioned limitations of current fault-location techniques, in this paper, a unified and extended fault-location formulation considering unbalanced distribution feeders is proposed and discussed. The proposed fault-location method uses local voltages and currents as input data. The method is based on the apparent impedance calculation and considers the time-varying load profile of EPDS based on the sending-end measurements. Also, in order to apply the technique in large distribution systems, a power-flow-based analysis is executed to determine the possible fault locations in each path. In order to validate the proposed method, the formulation was implemented in MATLAB [11] and had its performance evalu /$ IEEE

2 SALIM et al.: EXTENDED FAULT-LOCATION FORMULATION FOR POWER DISTRIBUTION SYSTEMS 509 ated by using real data of a distribution feeder from CEEE-D, simulated under BPA s ATP/EMTP Software [12]. The formulation was built as embedded software, and is currently used as an analysis tool at the CEEE-D s distribution operation center. The remainder of this paper is as follows. Section II describes the state-of-art impedance-based fault-location formulation. The proposed extensions are presented and discussed in Section III. The case study and the results are presented in Sections IV and V, respectively. Section VI presents the conclusions of this paper. Fig. 1. Single line-to-ground fault. II. IMPEDANCE-BASED FAULT LOCATION In the following subsections, the state-of-art impedancebased fault-location method is explained in detail. A. Fault-Location Equations Consider the faulted system illustrated in Fig. 1. It is possible to show that for a single line-to-ground (SLG) fault in [10] the subscript and represent, respectively, the variables real and imaginary parts. The variables are sending-end voltage (in volts); fault point to local bus distance (in meters); fault current (in amps). Also, and are defined in (2) and (3) s, and ; impedance between s and (in ohms per meter); sending-end current (in amps). B. Fault-Location Algorithm The following algorithm estimates the fault location. Step 1) Load current during the fault is assumed to be the prefault load current. Step 2) Considering Fig. 1, the fault current is calculated by using (4) is the remote end current. Step 3) Fault location is estimated by using (1) (3). (1) (2) (3) (4) Step 4) Fault-point voltages are calculated by using (5) Step 5) Load current is updated by using the faultpoint voltages in (6) and (7) (5) (6) (7) is the line impedance (mutual or self) per unit length between s and is the load impedance (mutual or self) between s and, and is the total line length. Step 6) Check whether has converged by using (8) is a previously defined error tolerance (given in meters) and is the iteration number. Step 7) If has converged, stop the procedure; otherwise, return to Step 2). C. Intermediate Loads The application of the described algorithm in radial distribution systems with intermediate loads is made through an iterative search process. The process starts by using the sending-end voltages and currents. Using the presented procedure, the fault distance is initially estimated. If the fault distance obtained is beyond the analyzed section length, then the fault is considered external. In this case, the algorithm is applied again, but now by using the downstream bus estimated voltages and currents. Since the measured voltages and currents are available only from one terminal, the voltage at the downstream bus can be estimated by (9) bus three- voltages vector (in volts); impedance matrix of the line section between buses and (in ohms); bus three- currents vector (in amps). (8) (9)

3 510 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 24, NO. 2, APRIL 2009 Considering a constant impedance load model, the downstream load current is given by (10) (10) is the bus three- load current and is its admittance matrix. Therefore, the downstream line current can be obtained by using (11) (11) After the voltages and currents at the downstream bus are estimated, the proposed fault-location algorithm is once again executed to the respective line section. The process is repeated until the fault-location algorithm converges to a distance inside the analyzed section. After the convergence, the total distance between the substation and the fault point is obtained by (12) Fig. 2. DLG fault. expressions for the fault distance. The expression is presented in matrix form by (12) is the distance between the sending end and the analyzed bus. III. PROPOSED EXTENSION The fault-location method described in Section II was already studied in previous works [10]. This method, however, was proposed only for SLG faults, which limits its practical application. Also, the method calculates the distribution system load profile through remote measurements, which are hardly available in practical applications. Still, in this method, the equivalent systems are obtained through the calculation of series and parallel impedances, which is not convenient in large distribution systems. In this paper, the consideration of these aspects is proposed as an extension of the method. The following subsections describe, in detail, the proposed extensions. A. Different Fault Types The application of the proposed fault-location formulation to all remaining fault types is made through specific fault-location equations. The fault-location algorithm remains the same; however, the fault-location equation for SLG faults, given by (1), is replaced by the equations presented in the following subsections. It should be noted, however, that a fault classification algorithm, such as those described in [13] [15], must be previously used to identify the correct fault type and to choose the correct fault-location equation. Since the development of such algorithms is beyond the scope of this paper, only the fault-location problem is analyzed, and the correct fault type is considered to be already known. 1) DLG Fault: Consider a generic DLG fault in s and [16], as illustrated in Fig. 2. Following the same procedure presented in Section II, it is possible to obtain the mathematical (13) the subscript and represent, respectively, the real and imaginary parts, and the superscript denotes a matrix transpose. The variables represent sending-end voltage (in volts); fault distance (in meters); fault current (in amps); fault resistance (in ohms); s and fault resistance (in ohms); s, and and and are given by (2) and (3). 2) Line-to-Line Fault: Consider an L-L fault between s and, as illustrated in Fig. 3. Following the same procedure presented in Section II, it is possible to obtain a fault-location expression, given by (14) (14) and are given by (15) and (16), respectively, with (15) (16) 3) Three-Phase Fault: Consider a three- fault as illustrated in Fig. 4. Following the same procedure presented in

4 SALIM et al.: EXTENDED FAULT-LOCATION FORMULATION FOR POWER DISTRIBUTION SYSTEMS 511 Fig. 3. Line-to-line fault. Fig. 5. General distribution feeder. Fig. 4. Three -to-ground fault. Section II, it is possible to obtain the fault-location expression, given by (17) and are given by (2) and (3), respectively. (17), and B. General Feeders Distribution systems are typically composed by a main feeder, laterals, sublaterals, and end loads. Laterals are main feeder branches not always composed by three- connections. Fig. 5 illustrates such a system. To consider each lateral, the proposed method calculates equivalent systems to each possible power flow path (PPFP), resulting in equivalent radial systems, is the number of laterals. In the presented system, illustrated in Fig. 5, for example, there are a total of 10 PPFPs, and the node groups (1, 2, 3, 4, 5, 6) and (1, 2, 16, 21, 22, 24, 25) comprise two of them. The equivalent systems are obtained by the transformation of lines and loads outside the path being analyzed into constant impedances along the radial system. Since the method analyzes the system in the first fault cycle, this assumption is a reasonable approximation [17]. The consideration of equivalent impedances has already been proposed in previously published papers [10]. This consideration was based in a systematic computation of parallel and series impedances, representing lines and loads. This type of approach, however, is not well suited for large distribution systems, especially in systems with a high degree of coupling between s. Instead, a power-flow algorithm could be used. The advantage of using this type of approach is that three- power-flow algorithms are well known, have great precision, and are easily applied in general distribution systems. The proposed formulation applies a ladder technique-based three- power flow [18], which is an iterative process developed for radial distribution systems applications. The implemented power flow considers the distribution system nonlinearities and asymmetrical coupling. Therefore, the proposed fault-location method can be applied to each equivalent system. The procedure is proposed as follows. Step 1) Run a three- power-flow algorithm, such as the one described in [18], considering the prefault conditions of the system. Step 2) Calculate the equivalent impedances in each node by using (18) (18) prefault voltage at bus (in volts); prefault current flowing from bus to bus (in amps). Step 3) Determine the PPFP. Step 4) Select one PPFP and determine the nodes with laterals. As an example, in path (1, 2, 3, 4, 5, 6), only nodes 2 and 4 have laterals. Step 5) For each node with lateral, determine an equivalent load, considering only the previously calculated equivalent impedances outside the path being analyzed. As an example, node 2 in the path (1, 2, 3,

5 512 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 24, NO. 2, APRIL , 5, 6) would have an equivalent impedance given by the parallel of and. Step 6) For each node with lateral and loads, conduct the parallel between loads and the equivalent load determined in Step 5). This is the final equivalent load for the analyzed nodes in the path. Step 7) Go back and follows Step 4) until all of the equivalent systems have been determined. Step 8) Execute the fault-location algorithm for each equivalent system. A total of fault locations will be determined. Step 9) Determine the correct fault location and path by using a fault section determination algorithm, as proposed by the authors in [19]. C. Load Profile The load demand in EPDS is dependent on the consumer type (residential, industrial, or commercial) and may vary with many factors [18]. Also, impedance-based fault-location methodologies are dependent on the system loading during the fault period, as one of the required inputs is the load in each node of the system. If the system has a different load profile during the algorithms execution, this will lead to errors, as is presented in Section V. To overcome this limitation, due to the time-varying load characteristics, a procedure to update the load value is proposed. Some previous works propose a load compensation based on measurements in each load node [8]. In practical distribution systems, however, such measurements are hardly available. In this way, the proposed extension is based on local measurements, available at the substation terminals. The proposed procedure is described as follows. 1) A standard equivalent impedance is defined as the equivalent impedance seen at the substation terminals, obtained by using the local voltages and currents measured when the system is operating in a previously defined normal operation load profile, and is given by (19) is the standard equivalent impedance (in ohms); terminal S normal operation load profile voltages (in volts); terminal normal operation load profile currents (in amps); (19) s, and. 2) When acquiring the faulted voltage and current data, the measured equivalent impedance seen from the substation is calculated by using the prefault data. The measured impedance represents the system s present operation state and is given by (20) (20) Fig. 6. AL-1PL distribution feeder. ohms); prefault terminal prefault equivalent impedance (in voltages (in volts); prefault terminal currents (in amps). 3) The measured equivalent impedance is compared with the standard equivalent impedance. A load variation rate is obtained by (21) (21) 4) The load matrices used in the power flow and in the proposed fault-location formulation are updated by multiplying them by the obtained load variation and considering an equal percentage change for all loads. 5) A three- power flow is executed to obtain the new equivalent systems. 6) The proposed fault-location technique is executed with the updated load information. The procedure just described uniformly considers changes for all loads. However, due to the stochastic nature of the timevarying load, each spot (or distributed) load can have different characteristics. If the load pattern is known for each load group, the load estimation procedure can use this information to weigh the load changes measured at the substation differently to each load group. IV. CASE STUDY To validate the proposed formulation, data from a real distribution system were used for numerical test simulations. The distribution system called AL-1PL from CEEE-D, illustrated in Fig. 6, was chosen as a test system. It has a total line length of 4015 m, with 3 three- laterals. A tape shield of 750 MCM and 4/0 AWG aluminum conductors compose the three- system. The line length in each section and the normal operation load in each node are presented, respectively, in Tables I and II. In order to model the distribution lines, Carson s equations [20] and the Kron s reduction method [21] were used. The test system was simulated by using BPA a ATP/EMTP Software [12]. The proposed fault-location formulation was implemented in MATLAB [11]. The simulations were executed considering the following fault conditions: 31 different fault locations (covering all system laterals and sections);

6 SALIM et al.: EXTENDED FAULT-LOCATION FORMULATION FOR POWER DISTRIBUTION SYSTEMS 513 TABLE I LINE LENGTHS IN AL-1PL TABLE II THREE-PHASE LOADS IN AL-1PL five different fault resistances: 0, 10, 20, 50, and 100 ; ten fault types: A-g, B-g, C-g, AB, BC, AC, AB-g, AC-g, BC-g, ABC-g. The simulated faults were divided in the following three different test sets, which are summarized as: Set 1: fault inception angle: 90 ; normal operation load profile; total: 1550 faults. Set 2: four different fault inception angles: 0,30,45, and 90 ; normal operation load profile; total: 6200 faults. Set 3: fault inception angle: 90 ; two different load profiles % load variation on six buses (2, 7, 8, 9, 10, and 11); total: 3100 faults. The percentage errors were calculated by using (22) estimated fault distance (in meters); simulated fault distance (in meters); total line length (in meters). % (22) V. RESULTS The effects of the fault resistance, the fault distance, the inception angle, the load variation, and the fault type on the proposed extended formulation are analyzed in the following subsection. A. Fault Resistance Effect Set I test results were used to analyze the fault resistance effect. Table III shows the average and maximum errors for four different fault types. The results show that the error slightly increases with increasing fault resistance. The highest difference between the average errors for 0 and 100 was close to 1% of the total line length. The maximum average error obtained was 0.98% for a three -to-ground fault (ABC-g), with, which represents an error of 39 m, approximately. The average error in all fault types was lower than 1%, considering faults up to 100, and in line-to-line faults, the error never increased to more than 0.1%, which represents about 4 m. The obtained errors show the efficiency and robustness of the methodology, though maximum errors of less than 3% were achieved for a 100- fault. The fault resistance effect may be explained by the erroneous estimation of the fault current for high resistances [22], which is dependent on the load current estimation during the fault period. For faults without resistance, the current divider circuit of the faulted system is constituted by the load impedance and the negligible fault resistance. Therefore, the substation current will mainly feed the fault, and the fault current will be close to the first. B. Fault Distance Effect To analyze the fault distance effect, Set I test results were used. Fig. 7(a) (c) illustrates, in detail, some obtained test results. Test results for SLG (B-g) faults with and 90 of the fault inception angle are presented, respectively. The obtained results show that the proposed formulation is independent of the fault location. The proposed formulation performance shows that there is no significant variation of the fault distance estimate as the simulated fault distance is modified. From Fig. 7, it can be observed that for faults without resistance, the errors were close to 0% in all feeder lengths. For faults with 20 of fault resistance, the errors were less than 0.25% of the total line length, which represents an error of lower than 10 m, approximately. For the worst case, with 100 of fault resistance, the maximum error was 1.6%. Fig. 7 also illustrates that the proposed formulation has a similar errors range in all line lengths, according to the fault resistance. This shows that the fault distance does not significantly affect the proposed extensions. C. Fault Inception s Angle Effect To verify the fault inception angle effect on the proposed formulation, Set II test results were analyzed. Table IV presents the obtained results for fault inception angles of 0,30,45, and 90, for four different fault types with 50- fault resistance. Table IV shows that the highest difference between average errors was 0.02%, for a -to- fault (BC) and angles of 0 and 45. This is not a significant difference; in this way, the fault-location estimates are not significantly affected by different fault inception angles. D. Load Variation Effect In order to verify the load variation effect, Set III test results were used. The average errors results for test load variations are shown in Tables V and VI. For comparative purposes, the proposed fault-location formulation was also executed without the load variation compensation.

7 514 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 24, NO. 2, APRIL 2009 TABLE III FAULT RESISTANCE EFFECT TABLE IV FAULT INCEPTION S ANGLE EFFECT FOR DIFFERENT FAULT TYPES WITH R =50 TABLE V LOAD INCREASE EFFECT Fig. 7. Single line-to-ground (B-g) fault-location results in different fault points. As can be seen from Tables V and VI, for faults without resistance, the fault-location formulation with and without load compensation is not significantly affected by the load variation. This is due to the substation current that will mainly feed the fault; thus, the fault current will be close to the substation current. However, for resistive faults, significant errors were obtained by using the formulation without the load variation compensation. For, an error of almost twice the total line length was obtained. And for faults with a resistance greater than 20,

8 SALIM et al.: EXTENDED FAULT-LOCATION FORMULATION FOR POWER DISTRIBUTION SYSTEMS 515 TABLE VI LOAD DECREASE EFFECT the average errors obtained are greater than 6%, which are not suitable for practical applications. On the other hand, by using the proposed load compensation extension, the error for the same fault resistance was always less than 0.29%, which represents about 12 m, a significant improvement. Also, the maximum error obtained with was close to 2.3% for a load increase and 4.7% for a load decrease. It is noted that comparing these results with the ones obtained without load deviation, that there is a fault-location estimation error increase. The cited errors, however, are still suitable for practical applications and show the robustness of the method in relation to load variation. Also, for faults limited to 50, the average error was always lower than 1% for a load increase and 1.5% for a load decrease. Considering that the load compensation is based only on voltage and current measurements at the substation, these results show great improvement in the fault-location method. E. Different Fault Types Tables III VI show results for different fault types, considering different effects. Based on these results, a slight difference between the effects and results in each fault type is noted. The fault resistance, as shown in Table III, has a lower effect on L-L faults and a more pronounced effect on faults involving all three s. Considering faults with ground (SLG, DLG, and 3PH), it is noted that as the number of s involved in the fault increases, the fault resistance effect also increases. On the other hand, the load variation effect was shown to be more pronounced on faults without ground (L-L), considering load decrease and increase. For faults to ground, the same profile can be noted that as the number of s involved in the fault increases, the load variation effect also increases. These statements show that there is a difference between the results obtained for different fault types as well as the effects for each of them. Based on the overall results, however, the feasibility is noted for practical applications of the methodology with the proposed extensions. VI. CONCLUSION This paper proposes and discusses an extended fault-location formulation to be used in general distribution systems. The method is based on the apparent impedance calculation and fundamental quantities. Furthermore, the method considers the specific characteristics of distribution systems, being capable of locating faults in systems with intermediate loads and laterals/sublaterals. The load profile variation is also considered in this formulation, considering only local measurements. This characteristic was not presented in the recent proposed impedance-based fault-location methods. The formulation is also suitable for large distribution systems and different fault types. A real distribution feeder from Brazil was simulated and had the technique s performance analyzed. Simulation results were provided to validate the formulation and illustrated its accuracy on fault location, showing the method s robustness and potential for real-time applications. The formulation was implemented as an embedded system and is used as an analysis tool at the operation center of CEEE-D. By the time this paper was published, no real fault cases have been analyzed through the software, and such discussions are to appear in future publications. REFERENCES [1] IEEE Guide for Determining Fault Location on AC Transmission and Distribution Lines, IEEE Std. C , Jun [2] S. Jamali and H. Shateri, Robustness of distance relay with quadrilateral characteristic against fault resistance, presented at the IEEE/ Power Eng. Soc. Transm. Distrib. Conf. Exhibit.: Asia and Pacific, Dalian, China, [3] J. Zhu, D. L. Lubkeman, and A. A. Girgis, Automated fault location and diagnosis on electric power distribution feeders, IEEE Trans. Power Del., vol. 12, no. 2, pp , Apr [4] K. Srinivasan and A. St.-Jacques, A new fault location algorithm for radial transmission lines with loads, IEEE Trans. Power Del., vol. 4, no. 3, pp , Jul [5] T. Takagi, Y. Yamakoshi, J. Baba, K. Uemura, and T. Sakaguchi, A new algorithm of an accurate fault location for EHV/UHV transmission lines: Part I Fourier transformation method, IEEE Trans. Power App. Syst., vol. PAS-100, no. 3, pp , Mar [6] T. Takagi, Y. Yamakoshi, M. Yamaur, R. Kondow, and T. Matsushima, Development of a new type fault locator using the one-terminal voltage and current data, IEEE Trans. Power App. Syst., vol. PAS-101, no. 8, pp , Aug [7] M.-S. Choi, S.-J. Lee, D.-S. Lee, and B.-G. Jin, A new fault location algorithm using direct circuit analysis for distribution systems, IEEE Trans. Power Del., vol. 19, no. 1, pp , Jan [8] M.-S. Choi, S.-J. Lee, S.-I. Lim, D.-S. Lee, and X. Yang, A direct three- circuit analysis-based fault location for line-to-line fault, IEEE Trans. Power Del., vol. 22, no. 4, pp , Oct [9] R. Das, Determining the location of faults in distribution systems, Ph.D. dissertation, Univ. Saskatchewan, Saskatoon, SK, Canada, [10] S.-J. Lee, M.-S. Choi, S.-H. Kang, B.-G. Jin, D.-S. Lee, B.-S. Ahn, N.-S. Yoon, H.-Y. Kim, and S.-B. Wee, An intelligent and efficient fault location and diagnosis scheme for radial distribution systems, IEEE Trans. Power Del., vol. 19, no. 2, pp , Apr [11] Mathworks Matlab. Natick, MA, [Online]. Available: [12] Alternative Transient Program: ATP/EMTP. Bonneville Power Administration, [Online]. Available: [13] H. S. Horowitz and A. G. Phadke, Power System Relaying, 2nd ed. London, U.K.: Research Studies Press, [14] B. Das, Fuzzy logic-based fault-type identification in unbalanced radial power distribution system, IEEE Trans. Power Del., vol. 21, no. 1, pp , Jan [15] R. H. Salim, K. R. C. Oliveira, M. Resener, A. D. Filomena, and A. S. Bretas, Hybrid fault diagnosis scheme implementation for power distribution systems automation, IEEE Trans. Power Del., vol. 23, no. 4, pp , Oct

9 516 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 24, NO. 2, APRIL 2009 [16] P. M. Anderson, Analysis of Faulted Power Systems, ser. IEEE Press Power Syst. Eng. Ser. New York: IEEE, [17] P. Kundur, Power System Stability and Control. New York: McGraw- Hill, [18] W. H. Kersting, Distribution System Modeling and Analysis. Boca Ratón, FL: CRC, [19] K. R. C. Oliveira, R. H. Salim, A. D. Filomena, M. Resener, and A. S. Bretas, Unbalanced underground distribution systems fault detection and section estimation, in Advanced Intelligent Computing Theories and Applications. With Aspects of Artificial Intelligence, ser. Lect. Notes Comput. Sci. Berlin, Germany: Springer, 2007, vol. 4682, pp [Online]. Available: [20] J. R. Carson, Wave propagation in overhead wires with ground return, Bell Syst. Tech. J., vol. 5, pp , [21] G. Kron, Tensorial analysis of integrated transmission systems. Part I: The six basic reference frames, AIEE Trans., vol. 71, [22] A. D. Filomena, R. H. Salim, M. Resener, and A. S. Bretas, Fault resistance influence on faulted power systems with distributed generation, presented at the 7th Int. Conf. Power Systems Transients, Lyon, France, [Online]. Available: Rodrigo Hartstein Salim (S 07) was born in Porto Alegre, Rio Grande do Sul, Brazil, on September 15, He received the E.E. and M.Eng. degrees from the Federal University of Rio Grande do Sul (UFRGS), Porto Alegre, Brazil, in 2006 and 2008, respectively, and is currently pursuing the D.Eng. degree in power systems at the University of São Paulo (EESC-USP), São Carlos, Brazil. His research interests include power system protection, control, and distributed generation. Mariana Resener (S 07) was born in Passo Fundo, Rio Grande do Sul, Brazil, on April 9, She received the E.E. degree from the Federal University of Rio Grande do Sul (UFRGS), Porto Alegre, Brazil, in 2008, she is currently pursuing the M.Eng. degree in power systems. Her research interests include fault detection and location. André Darós Filomena (M 09) was born in Porto Alegre, Rio Grande do Sul, Brazil, on October 27, He received the E.E. and M.Eng. degrees from the Federal University of Rio Grande do Sul (UFRGS), Porto Alegre, Brazil, in 2005 and 2008, respectively. He is currently with Companhia Estadual de Geração e Transmissão de Energia Elétrica (CEEE-GT), Porto Alegre. His research interests include power system protection, modeling, and reliability. Karen Rezende Caino de Oliveira (S 07) was born in Aracaju, Sergipe, Brazil, on December 18, She received the electrical engineering degree from the Federal University of Rio Grande do Sul (UFRGS), Porto Alegre, Brazil, in 2007, and is currently pursuing the M.Eng. degree in power systems at the University of São Paulo (EESC-USP), São Carlos, Brazil. Her research interests include power system stability and fault location. Arturo Suman Bretas (M 98) was born in Bauru, São Paulo, Brazil, on July 5, He received the E.E. and M.Eng. degrees from the University of São Paulo, Brazil, in 1995 and 1998, respectively, and the Ph.D. degree in electrical engineering from Virginia Polytechnic Institute and State University, Blacksburg, in Currently, he is an Associate Professor of the Federal University of Rio Grande do Sul (UFRGS), Porto Alegre. His research interests include power system protection, control, and restoration.