Estimation of the line-to-line voltage phases in MV distribution faults using transformer inrush currents

1 Estimation of the Line-to-Line Voltage Phases in MV Distribution Faults using Transformer Inrush Currents Antonio Garc´ıa, F´elix Biscarri, Carlos Le´on, Member, IEEE, and Enrique Personal Abstract—This paper proposes an innovative and comprehensive method using transformer measured inrush current during the first breaker re-close to add usefull information as fault location problem input. It is divided into the main following sections: Introduction, Magnetizing Inrush Current- A Brief Analysis, Voltage Phase Estimation Method and Conclusion. The authors insist on the importance and the availability of the fault data recorded by the line protection relay (oscillogram, COMTRADE file), with the transformer impedances knowledge, to accurately estimate voltage phases. Index Terms—Inrush current, fault location methods, oneterminal lines. I. INTRODUCTION DURING the last decade, these has been much research on the fault location problem. There are a lot of bibliographical references about main research topics regarding this paper: fault location methods for distribution systems (see [4]) and inrush current in power transformers( [1], [2] and [3]). A classical research will be found in [5]. The IEEE offers a complete guide, [6]. Many new fault location algorithm are available [7], [8], but it authors usually assume that they knew of current and voltage values and also about its phases. The main contribution of this paper is to look for hidden information in a non linear event in lines: the inrush current, and its relation with phase voltages. Both transmition and distribution networks, the usual methods can be classified into three groups: methods that use traveling waves, methods that use harmonics and methods based on the apparent impedance calculated from fundamental components of voltage and current. The last mentioned methods use, as available inputs from measurements, the phase voltages, the phase currents and all of the system characteristics except the fault resistance (RF) But there is a practical problem using these methods. With the means available to distribution companies, almost in Europe, the full aforementioned input information is not usually available. Sometimes, voltages and currents are measured but line impedances are partialy unknown. Sometimes, only the RMS voltages is known, but voltage phases are unknown. The paper presents an innovative method to obtain the voltage phases from a set of registered current in three different conditions: pre-fault, fault and first breaker re-close. This method might come in useful for complement a lack of information, in order to put into practice any of the fault location methods. Medium voltage (MV) distribution feeders are normaly protected by overcurrent relays. Modern devices (IED), adding a microprocessor, not only measure the currents and trip the breaker switch when necessary, but also record the input currents. The stored oscillogram is reduced to a few tenth of seconds before and after a trip. The output oscillogram file is recorded in a standard file format (COMTRADE). These files are availables, usually few seconds after the fault, throught the utility communication system, in the Utility Dispatch Center. Most protection strategies include an automatic re-close operation on the breaker switch, from a half to a few seconds after the fault detection. This operation allows to solve nonpermanent faults. After the re closed operation, if there is still the fault, the relay switch-off the line permanently. As we has said, the relay records current’s data in three different conditions: pre-fault, fault and first breaker re-close. Inrush current, due to re-energization of transformers, are shows in the last situation. Two real oscillograms from a medium voltage relay are shown in Figure 1 and Figure 2 (courtesy by Endesa Company). Note the short-circuit earth current is limited to 300A. by the transformer neutral grounding resistor. Fig. 1. Pre-Fault and Fault Currents in a Real System This paper present a new method to accurately estimate, from recorded COMTRADE files, the voltage phases in trasnformers. This method is based in looking for the maximum value of the inrush current recorded, that accurate pointed the zero line input transformer voltage. 2 Fig. 2. The First Breaker Re-Close II. MAGNETIZING INRUSH CURRENT-ABRIEF ANALYSIS Magnetizing inrush current in transformers results form abrupt change of the magnetizing voltage. When a fault occurred, magnetizing inrush may be caused by: •Occurrence of an external fault. •Voltage recovery after clearing an external fault. •Change of the character of a fault. For example, when a phase-to-ground fault evolves into a phase-to-phase-to- ground fault. If the input transformer voltage is sinusoidal, the magnetic flux too, and the magnetization current is a bell-shaped curve (Figure 3). The current harmonic spectrum declines significantly from the third harmonic. If the magnetizing current is sinusoidal, the third harmonic of the flux will be significant and also the third harmonic in the induced core transformer voltage. Figure 4 shows this case. The proposed method will be applied in both cases. The most common case in the real electric distribution lines is the first one. Fig. 3. Bell-shaped Magnetization Current Fig. 4. Sinusoidal Magnetization Current A. Inrush due to switching-in Initial magnetizing due to switching a transformer is considered the most severe case of an inrush. When a transformer is de-energized (switched-off), certain remanent flux left in the core. When, afterward, the transformer is energized by an alternating sinusoidal voltage, the flux become also sinusoidal but biased in remanence. The flux-current trayectories for above the knee-point of the characteristic resulting in both large peak values and heavy distortions of the magnetization current. Figure 5 shows a typical inrush current shape. Fig. 5. Inrush Current in a Single Phase Transformer B. Inrush in three phase transformers Inrush current, or magnetization currents, measured in separated phases of a three-phase transformer (in the simple showed model: Iabmag,Ibcmag and Icamag, Figure 6) may differ considerably because of the following: •The angle of the energizing voltages are different in different phases. 3 Fig. 6. Simple l-g Fault •When the delta-connectedwinding is switched-in, the line voltages are applied as the magnetizing voltages. Line current on a giving phase is a vector sum of two winding currents. •Sometimes, depending of the core type and others conditions, only some of the core legs may get saturated. In real faults, is not a very common case. As a result of the aforementioned, the augment in a particular phase and in a grounded neutral may be similar to the single-phase inrush current pattern, or become distorted but oscillatory waveform. In the later case, the amount of the second harmonic may drop dramatically, creating problems for differential relaying. Figure 7 presents an example of energizing a three phase transformer. Magnetization currents and voltages are shown. In this example, the currents in the phases assume the typical inrush shape. Otherwise, only one current present a clear oscillatory waveform. Fig. 7. Inrush Current and Associated Line-to-Line Voltages III. VOLTAGE PHASE ESTIMATION METHOD This simple but innovative method to detect the phase of the voltages is based in the evolution of inrush currents with the re closed time. As it can easy be checked, the three magnetization currents, Iabmag,Ibcmag and Icamag, never overlap (see Figures 8 and 9). Thus, we can easy obtain magnetization current from the difference between fault current and inrush current. The following steps describe the proposed method. The input variables are fault currents and inrush currents, recorded Fig. 8. Magnetization Currents in Transformers, Balanced Systems Fig. 9. Magnetization Currents Contour Plot by the relay. The line voltages phases are unknown. Its are the target of this study. 1) Obtain the measured difference between fault and inrush currents: Ia −Iafault,Ib −Ibfault and Ic −Icfault. Figure 10 shows typical shapes of these differences. From these measured functions, estimation magnetization currents will be obtained. Fig. 10. Ia −Iafault,Ib −Ibfault and Ic −Icfault 2) If the function with the lowest abs(maximum)value , Ib −Ibfault in Figure 10, is a clear oscillatory waveform, the case is simple. A threshold will be fixed, and shapes A and B correspond,exactly, with the two smaller magnetization currents: Iabmag and Ibcmag (Figure 11 3) The third unknow magnetization current,Icamag, is estimated by IabmagESTIMAT ED −[Ia −Iafault]or by 4 Fig. 11. Estimation of Iab mag and Ibcmag from Ib −Ibfault IbcmagESTIMATED +[Ic −Icfault]. The redundancy in the estimation of Icamag will be used to reduce the error of the global estimation. 4) If the function with the lowest abs(maximum)value is not ’a clear’ oscillatory waveform (Figure 12), the method is the same, because the function with the lowest abs(maximum)value is always an oscillatory waveform. Simple, one of the magnetization current is very small that the others, an it influence in inrush current is limited. Once the two minor magnetization current are estimated, the other will be easily estimated from measured Ix−Ixfault. By this last point, authors emphasize the importance of the real limited measurement sampling in the estimation process. Fig. 12. Iabmag has limited influence 5) At last, once the magnetization currents are estimated, its maximum values pointed when input transformer voltages are zero, and it set the voltage phase. Impedances show in Figure 6 will be know. Once estimated voltage phases in transformer, known line currents and line impedances, it’s possible to estimate the voltage phases in the substation; or known voltages in the substation, it’s possible to estimate line characteristics. Perhaps the process is simple than could be thought at first. For example, Ixfault is not exactly the component of the postfault Iab, Ica and Ibc (figure 6), but the difference in real balanced system studied is not significant. IV. CONCLUSION This paper deals with the amount of the available measurements in power lines, to detect electric faults. A new phase voltage estimation method, applied in a simple balanced systems, is presented, to find relevant knowledge about the particular characteristics of a non-linear electric phenomenon: inrush currents in transformers. This method might come in useful for complement a lack of information, in order to put into practice any of existing fault location methods. ACKNOWLEDGMENT The authors would like to thank Endesa Company for providing the data support for this project. Also, this work has been funded by the research projects ref. C03-056, ref. CICYT/FEDER DPI2006-15467-C02-01 and ref. CICYT/FEDER DPI2006-15467-C02-02. REFERENCES [1] S. G. Abdulsalam and W. Xu, “Analytical study of transformers inrush current transients and its applications,” in International Conference on Power Systems Transients, IPST’05. IEEE/PES, 2005, pp. 140–145. [2] Y. Wang, S. G. Abdulsalam, and W. Xu, “Analytical formula to estimate the maximum inrush current,” IEEE Transactions on Power Delivery, vol. 23, no. 2, pp. 1266–1268, april 2008. [3] R. Girgis, “Inrush current tutorial session.” IEEE Standards Transformers Committee, 2001. [4] K. R. T. C. de Oliveira, R. H. Salim, and A. D. Filomena, “Unbalanced underground distribution systems fault detection and section estimation,” in ICIC 2007. Springer-Verlag, 2007, pp. 1054–1065. [5] T. Takagi and Y. Yamakoshi, “Development of a new type fault locator using the one-terminal voltage and current data,” IEEE Transactions on Power Apparatus ans Systems, pp. 2892–2898, 1982. [6] I. S. C37.114, IEEE Guide for Determining Fault Location on AC Transmision and Distribution Lines. 3 Park Avenue, New York, NY 10016-5997, USA: IEEE Power Ingineering Society, 2005. [7] M. Choi, S. Lee, D. Lee, and B. Jin, “A new fault location algorithm using direct circuit analysis for distribution systems,” IEEE Transactions on Power Delivery, vol. 19, no. 1, pp. 508–516, january 2004. [8] R. H. Salim, M. Resener, A. D. Filomena, and K. R. C. de Oliveira, “Extended fault-location formulation for power distribution systems,” IEEE Transactions on Power Delivery, vol. 24, no. 2, pp. 508–516, april 2009. Antonio Gar´ıa received the B.Sc. degree in Physical Electronics from the University of Seville, in 1982. He has been Professor of Electronic Engineering in the Electronic Technology Department since 1984. His areas of research include instrumentation, digital signal processing, power devices and fault location methods in power lines. F´ elix Biscarri received the B.Sc. degree in electronic physics and the Ph.D. degree in computer science from the University of Seville, Seville, Spain, in 1991 and 2001 respectively. He is currently a Coordinator Professor of Power Electronic with the Polytechnic University School of Seville. His research areas include electricity markets, fault and fraud detection in the power electric industry. 5 Carlos Le´ on received the B.Sc. degree in electronic physics in 1991 and the Ph.D. degree in computer science in 1995, both from the University of Seville, Seville, Spain. He is a professor of Electronic Engineering at the University of Seville since 1991 and currently CIO of the University of Seville. His research areas include knowledge based systems and computational intelligence focus on Utilities System Management. He is a Member of the IEEE Power Engineering Society. Enrique Personal received the B.Sc. degree in Industrial Technical Engineering the B.Sc. in Automatic Control and Industrial Electronics Engineering from the University of Seville, Spain, in 2006 and 2009, respectively. He is a member of the research group ”Electronic technology and Industrial informatics”, where he is currently pursuing his Ph.D. His main fields of interest are power systems and fault location methods in power lines.