Fault Location in MV Unearthed Distribution Network Using the Undamped Frequency of the Transient Signal

Transcription

1 Fault Location in MV Unearthed Distribution Network Using the Undamped Frequency of the Transient Signal MOHD RAFI ADZMAN, MATTI LEHTONEN Department of Electrical Engineering, School of Science and Technology Aalto University Otakaari 5 FI Espoo FINLAND mohdrafi.adzman@aalto.fi, matti.lehtonen@aalto.fi Abstract: - This paper present a method to locate a fault location for single phase to earth fault in medium voltage (MV) unearthed network using undamped frequency of the transient signal. During the earth fault, a transient component in the fault signal provides valuable information for fault location purposes. The fault distance is formulated using the lumped parameter model. In this paper, the properties of Hilbert transformation are used to estimate the damping attenuation of transient signal. The networks considered are assumed to be radially operated and modelled using ATPDraw program. Furthermore, the accuracy of technique was tested on parameters such as a fault distance, fault resistance and the voltage inception angle. Some sources of error affecting the method accuracy, are explained. Finally, the possible correction technique is explained to compensate the computation error. Key-Words: - Fault location, earth fault, isolated network, charge transient, discharge transient, symmetrical component, damping attenuation 1 Introduction Many countries in Northern Europe including Finland, most of their medium voltage distribution networks are operated with an unearthed or compensated neutral. The feeders are radially operated. The most common fault type in electrical distribution network is single phase to earth fault. Reported from the earlier studies [1-3], for instance in Nordic countries, about % of all faults are of this type. When a single phase to earth fault happens, the fault current in these kind of distribution network is much less than that in solid ground neutral system. For example, in rural overhead networks, the earth fault current is typically around A [4]. The small earth fault current at the fault point is advantage to system operation. It improves power quality in these networks by reducing the outages caused by single phase to earth faults. However, in these kind of networks, the localization and detection of fault is much more difficult than in the system having low impedance or solid neutral earthing. Latest technology trend in fault location is required to locate a fault quickly, reliables and with less human intervention. It is possible by utilizing the transient that occurs at the moment of the fault initiation. The initial transients of the earth faults are of importance especially in unearthed and compensated networks. The fault signal consists of different frequency components, which result from charging and discharging of the network capacitances. When an earth fault happens, the voltages of the sound phases are rapidly increased. This initiates a charging transient. At the same time, voltage of the faulty phase rapidly fall to, or close to zero, and the charge stored in its capacitances is removed. This give rise to the discharge transient. An example of initial transient fault voltage is shown in Figure 1. Fig.1 Simulated phase voltage of the faulted feeder When earth fault happens at 20ms. ISBN:

2 Nowadays, fault localization and detection based on fault transient signal in transmission and distribution network has been a subject of interest to utility engineers and researchers [1,4,5,6]. In this paper, we discuss a method to locate a single phase to earth fault in medium voltage network with unearthed neutral. The method utilizes the initial transient of the earth fault signal. The frequency of the generated transient signal is used in estimating the fault location. All faults in this work are simulated using ATPdraw software. Some sources of errors affecting the method accuracy, are explained. 2 Fault Location Principle In this paper, the algorithm which calculates the earth fault location is based on the frequency of the transient charging process. The charging transient is usually of higher amplitude and lower frequency than the discharge transient. The advantage of this method is that only one measurement is needed per substation transformer. 2.1 Earth fault transient solution using symmetrical components The inductance of the faulty line from transformer to the earth fault location is the most important parameter in transient analysis. In symmetrical component method, the transients are resolved into positive, negative and zero sequence components. The unsymmetry of the fault circuit is removed by an appropriate interconnection of the sequence networks. In case of single phase to earth fault, the sequence networks are connected according to Figure 2. Fig.3 Network model for general solution of discharge and charge transients. For simplicity the line and transformer resistance are not shown. The sequence network and the corresponding connection for earth fault transient in Figure 3 is shown in Figure 4. Based on Thevenin s theorem, the earth fault transients can be produced by injecting into the model terminals, a voltage of equal magnitude with opposite sign compared to the phase voltage at the fault point. 2L 1,p +L 1,0 2L T C E 0.5(C E +3C) Fig.4 The composite connection for isolated network without damping included [1]. The charge and discharge transient frequency can be determined by writing the differential equation with the aid of voltage equation around the loops of the circuit in Figure 4. The natural frequencies of the circuit can be solved as the roots of the characteristic equation as follows [1]: Fig.2 Interconnection of phase sequence networks for a single phase to earth fault [8]. F refer to fault and N to the system neutral. We apply the symmetrical sequence method to the network as shown in Figure 3. The model neglected the inductances and resistances of sound lines, distribution transformers, low voltage loads and approximated the network capacitances as lumped components located at the substations. ω = a 2 a 2 b (1) where; a= L C +L C +L C L L C C 1 b= L L C C L =2L L =L, +L, +L, =2L, +L, ISBN:

3 C =0.5C +1.5C C =C L T =substation transformer inductance L 1 =inductance of the faulty line L 1,p, L 1,n, L 1,0 =sequence inductances of the faulty line length C E =earth capacitance of the network C=phase to phase capacitance of the network transform (FFT). The spectrum of fault current and voltage is shown in Figure 5. 3) Removing of undesired frequency with FIR filter If, the fault happens at the substation busbar, L 2 in equation 1 should be set to zero and the equation should be modified as follow [1]; 1 ω = 3L (C+C ) (2) Fig. 5 A typical frequency spectrum of fault current and voltage. Equation (2) is same for the frequency of charge transient as in [7]. Assume that l be the distance of the earth fault from the substation and the total inductance of the faulty line will be lxl 2.Then, equation (1) can be rearranged to estimate the fault distance of the faulty line from the transformer to the earth fault location: = ( ) ( + ) 1 L L C C ( ) L C (3) 3 Signal Pre-processing Fault transient signal is non-stationary signal and it consists of many frequency components. The main purpose of pre-processing is to extract the charge transient that mixed with other signals such as noise and fundamental frequency component. In a real network, the fault transient signal is recorded using the fault recorder installed in substation. In this paper, we use only charge component for fault location calculation. The reason why we use only charge component is because it has lower frequency and in most cases a larger amplitude than the other transient component. The pre-processing of the transient signal is divided into two parts. First is the identification of charge transient components. Second part is the estimation of the damping attenuation. First part of the signal processing is made in the following steps: 1) Removal of the fundamental frequency component using comb filter 2) Spectrum analysis for estimating the charge transient frequency using fast Fourier In second part, the main steps needed to obtain an estimation of the damping attenuation (δ) through Hilbert transformation can be summarized as follows [9]: 1. In first part of pre-processing, we obtained the interesting part of the transient signal which is the charge component. 2. Then, the Hilbert transformation was used to obtain the envelope of the interesting part as a function of time. 3. Finally, the damping attenuation (δ) was estimated from the plot of ln A x (t) versus time. In this case A x (t) is the instantaneous amplitude of the filtered transient signal. The damping attenuation (δ) is the slope of curve plotted of In A x (t). If the damping is linear, a first order polynomial will be sufficient to fit the curve. Fig.6 Example of instantaneous current amplitude, envelope and the linear decaying of the filtered charge transient signal The undamped charge transient frequency can be calculated as follows [10], = + (4) where, δ = damping attenuation, ω d = angular frequency of the transient signal ISBN:

4 4 Simulation Analysis and Result A single line diagram of the simulated unearthed medium voltage (MV) network and its simulation model is shown in Figure 7. The simulation model is based on the sequence component networks for the exact solution of the earth fault transient. The system is 21 kv overhead lines unearthed MV radial network and the line parameters are as follows: positive sequence resistance is 0.6 ohms/km, zero sequence resistance is 1.3 ohms/km, positive sequence inductance is 1.0 mh/km, zero sequence inductance is 5.0 mh/km, positive sequence capacitance is nf/km, zero sequence capacitance is 6.12 nf/km. We assumed the negative sequence parameters equal to the positive ones. The lines were modeled as a chains of Π-type cells (1 cell per 10km of the line).the network was 5 feeders and the earth fault was simulated at feeder 1. The total length of the network is 98km. The substation transformer positive sequence resistance and inductance is ohms and 2.8 mh respectively. All faults with several different fault conditions were simulated with the software package ATP (Alternative Transients Program), version of EMTP program where the circuit was realized using ATPDraw. The sampling frequency was 20 khz. In the simulation analysis, the load was either 100Ω, 500Ω,1000Ω or zero. All loads was located at the end of the line. The simulated fault distance with the corresponding fault distance computation error are shown in Table 1, 2, 3 and 4. Table 1 Transient fault location versus fault distance without load. l l err l err (km) (km) ( %) Table 2 Transient fault location versus fault resistance without load. L R f l err l err (km) (Ω) (km) ( %) Table 3 Transient fault location versus inception angle without load. Fig. 7 The simulated, 21kV, 98km, unearthed MV network (above) and its simulation model (below). M is the measurement point, F refer to fault location. Z F is the fault impedance.z 1T and Z 2T are positive and negative sequence impedance of the substation transformer. j=2 5 refer to the impedances of the parallel sound lines. l Angle l err l err (km) ( o ) (km) ( %) ISBN:

5 Table 4 Transient fault location versus load l Load l err l err (km) (Ω) (km) ( %) In overall, the computation error for cases studied without load have distance error less than 2km. The errors are primarily due to the simplification of the lumped parameter model used in this algorithm. From the result, the fault distance error is found to be increased when the load is also increased. From the result in Table 2 and 3, the fault angle and fault resistance has small effect to the resistance. From the result in Table 1, the error is high when the fault happens close to the substation. Also, the percentage error is between 3 to 11 percent if the fault occurred between the middle to the end of faulted line. It seems that if the fault happen close to the substation the algorithm will estimate higher than the exact fault location. While, if the fault happens is close to end of the line, it will estimate lower than the exact fault location. From this observation, we can make assumption that the computation error from this algorithm can be compensated by multiplying the estimation fault distance with the estimated correction factor. In this paper, the correction factor is defined as, = (5) where l is exact fault distance and l cal is calculated fault distance. Assume that we know the total length of faulty feeder, in this case is 20km. The estimated correction factor can be applied as following condition, 1. If the estimated fault distance is close to the substation. In this case 10 % of the total faulty length. Recalculate using the correction factor between If the estimated fault distance is close to the end of the faulty length. In this case 90% of the faulty length. Recalculate using the correction factor between Conclusion In this paper, an algorithm to locate the single phase to earth fault in unearthed neutral network using undamped frequency of the transient signal has been presented. The algorithm used the lumped model for the distance from the transformer to the earth fault location. The main error is due to the simplification of the lumped parameter model. The computation error can be compensated with the estimated correction factor. 6 Acknowledgements The first author gratefully acknowledges the studentship given by the Universiti Malaysia Perlis (UniMAP), Malaysia, for the PhD work in this are. References: [1] Lehtonen, Matti., Transient analysis for ground fault distance estimation in electrical distribution networks, Espoo 1992, Technical Research Centre of Finland, VTT Publications 115. [2] K. Winter, The earth fault problem and the treatment of the neutral in distribution network, ERA2,1987, (Original in Swedish) [3] H. Paulasaari, P. Järventausta, P. Verho, M. Kärenlampi, J. Partanen, T. Hakola, E. Vähätalo, Methods to study earth fault phenomena by using a residual overvoltage relay module, IEEE/KTH Stockhom Power Tec. Conference, Stockholm, Sweden, June 18-22, 1995, 55 pp [4] A. Nikander, E. Lakervi, J. Suontausta, Applications of transient phenomena during earh faults in electricity Distribution networks, Proc. Conf. Energy Management and Power Delivery, pp ,1995. [5] Hänninen, Seppo, Single phase earth faults in high impedance grounded networks. Characteristic, indication and location, Espoo 2001, Technical Research Centre Finland. VTT Publication 453, 78p.+app.61p [6] Imriš, P., Lehtonen, M, Transient based ground fault location using wavelets, The 4th IASTED International Conference on power and energy systems, Rhodes, Greece, June , pp [7] Pundt, H, Untersuchungen der Ausgleichsvorgänge bei Erdschluß in Hochspannungsnetzen mit Sternpunkt und induktiver Sternpunkterdung als Grundlage zur selektiven Erdschlußerfassung. Dissertation, TU Dresden, 1963, 167 pp. + app ISBN:

6 [8] Greenwood. A, Electrical transient in power systems, John Wiley & Sons, New York 1979, 515p [9] Mohd Rafi Adzman, Matti Lehtonen, The correlation of fault distance and transient frequency, Methods and techniques for earth fault detection, indication and location Seminar, 15 February 2011, Aalto University, Otaakari 5, Espoo Finland. [10] Arieh L. Shenkman, Transient analysis of Electric power circuits handbook, Springer, 1 edition, 2005 ISBN: