Detection of High Impedance Fault

Transcription

1 Detection of High Impeance Fault Abstract: High impeance faults (HIFs) on transmission an istribution systems create unique problem to many power network relaying schemes. In this paper the high impeance fault etection issue is aresse. This paper presents a brief synopsis of what has transpire to ate in the effort. The work proposes a technique base on Hilbert Transform of current for HIF etection. The technique is evaluate for a system for ifferent HIFs. Keywors: High impeance fault, Fault etection, Hilbert transform, Digital relay H I. INTRODUCTION Premalata Jena Department of Electrical Engineering Inian Institute of Technology Kharagpur West Bengal premalata@ee.iitkgp.ernet.in Ashok Kumar Prahan Department of Electrical Engineering Inian Institute of Technology Kharagpur West Bengal akprahan@ee.iitkgp.ernet.in IGH impeance faults in transmission an istribution lines are cause by line coming into contact with a tree branch, fire or a construction crane [1]-[3]. These faults also occur when an overhea conuctor breaks an falls on high impeance surface such as asphalt roa, macaam, san, cement, grass or tree. The HIF causes the effective impeance to change constantly an ranomly ue to the fault s arcing behaviour an the fault current magnitue becomes very chaotic in nature [1]. HIFs are accompanie by variations in the 5-Hz an harmonic components. However, these variations are not stationary, but time-varying in nature ue to the ynamics of the fault arc [3]. High impeance faults are ifficult to be etecte through conventional protection relays such as istance an overcurrent relays because of small change in funamental component [1]- [1]. The fault current in a HIF may range from zero to the pickup setting of overcurrent relays. Therefore, the overcurrent relays will not be able to etect the HIF. When this type of fault happens, energize high-voltage conuctors may fall within reach of personnel. In aition, as the arcing often accompanies with these faults, it further poses a fire hazar. Therefore, from both public safety an operational reliability viewpoints, the etection of high impeance faults is critically important. Several researchers in recent years have presente many techniques aime for etecting HIF more effectively. The work in [1] presents an approach for quantization of power system behaviour which lies between strict fractal imension calculation algorithms an statistical estimation of ranomness of change of current values in the system. In [2], an approach base on Kalman filtering is propose to obtain the best estimation of the time variations of the funamental an harmonic components for etection of high impeance fault. Wavelet transform is use for the ecomposition of signals an feature extraction, feature selection is one by principal component analysis an Bayes classifier is use for classification [4]. The metho in [6] is use for iscriminating HIFs from insulator leakage current (ILC) an transients such as capacitor switching, loa switching (high/low voltage), groun fault, inrush current an no loa line switching. An application of Morlet wavelets to the analysis of highimpeance fault generate signals is propose in [5]. With the time-frequency localization characteristics embee in wavelets, the time an frequency information of a waveform can be presente as a visualize scheme for etection of high impeance fault. In [6], the measure values of current harmonics at a stage high impeance groun fault on sany soil are compare with 12 an 18 Hz frequency components for etection of high impeance fault. The measure low frequency spectrum is compare with current harmonics recore continuously for one week at the substation. In [7], the Discrete Wavelet Transform (DWT) as well as frequency range an rms conversion are use to apply a pattern recognition base etection algorithm for electric istribution high impeance fault etection. The aim is to recognize the converte rms voltage an current values cause by arcs usually associate with HIF. The analysis using Discrete Wavelet Transform (DWT) with the conversion yiels measurement voltages an currents which are fe to a classifier for pattern recognition. The classifier is base on the algorithm using nearest neighbour rule approach. In [8], a metho base on incremental variance of normalize even orer ratio measure is use for the etection of high impeance faults. An integrate high an low impeance fault etection metho is propose in [1]. In this work, for HIF etection, the propose technique is base on a number of characteristics of the HIF 1

2 Voltage (V) current. These characteristics are: fault current magnitue, magnitue of the 3r harmonic current, magnitue of the 5th harmonic current, the angle of the thir harmonic current, the angle ifference between the thir harmonics current an the funamental voltage, negative sequence current of HIF. In [11], a scheme which uses the istorte voltage waveforms uring arcing is iscusse. The propose scheme implemente on two ifferent moels of HIF in extra high voltage mutually couple ouble ene transmission lines. The high pass filter tap yiels three phase voltage in the high frequency range which is fe to Clarke s transformation to ecouple the travelling waves of the mutually couple lines an prouces groun moe an aerial moes voltage components to the classifier for pattern recognition. The classifier is base on an algorithm that uses recursive metho to sum the absolute values of the high frequency signal generate over one cycle an shifting one sample. This work proposes an alternative solution for etection of high impeance fault in a transmission system. It is foun that the magnitue of the etector obtaine from the Hilbert transform of instantaneous phase current is capable of proviing the occurrence of high impeance fault. The propose technique is an attractive option as it oes not require fault voltage component avoiing istortion ue to CCVT an eliminating the issue of line or bus sie voltage selection. The metho is teste using ata simulate with MATLAB Simulink for single circuit configuration. The performance of the algorithm is also evaluate. II. THE SYSTEM A 4 kv, 5 Hz three phase system as shown in Fig. 1 is consiere. In the system, line-1an line-2 segments are of 1 km. Detail system ata are provie in appenix-a. The fault voltage an current ata are collecte at a sampling rate of 1 khz. M Line1 Line2 F ~ ~ Source-1 L Fig. 1. The three phase power system. HIF N Source-2 resistance is varie ranomly between 5 an 1. The uration of each resistance value is ranomly varie between 1 s an 5 ms. This way the true ranomness of HIF s can be capture. When the line voltage is greater than the positive DC voltage Vp, the fault current starts flowing towars the groun. The fault current reverses backwar from the groun when the line voltage is less than the negative DC voltage Vn. In case of the line voltage being in the value between Vp an Vn, line voltage is counter-balance by Vp or Vn so that no fault current flows. The typical fault current an voltage are shown in Fig x 15 D p V p R p F Fig. 2. Simplifie 2-ioes fault moel of high impeance faults Fig. 3. Typical current an voltage waveform uring high impeance fault. D n V n R n A simplifie 2-ioes moel [5] of HIF is use in the simulation. The circuit of the HIFs moel is shown in Fig. 2. This HIF moel is base on arcing in sany soil. The moel inclues two DC sources, Vp an Vn, which present the arcing voltages of air in soil an/or between trees an line. Two resistances, Rp an Rn, between ioes an DC voltages present the resistance of trees an/or the earth resistance. In orer to simulate asymmetric current, ifferent values of Rp an Rn are use. The high impeance fault moel in [11]-[14] relies on ranomly varying the magnitue of the fault resistance an its uration. The III. PROPOSED TECHNIQUE There are several techniques available for high impeance fault etection using current-voltage information [8], [9] at transmission an istribution levels. In this work a etector obtaine from the Hilbert Transform of instantaneous phase current is use for the etection purpose. The Hilbert transfer function in the time omain is expresse as in (1) 2

3 1 h(t)= t (1) πt While in the frequency omain it is expresse as in (2) j H( )=-jsgn( )= (2) j Since there is no negative frequency in the real application, the Hilbert transform can be regare as an all-pass filter, with a phase shift of 9 for all positive frequency components. A real function x(t) an its Hilbert transform h(t) are relate to each other in such a way that they together create a so calle strong analytic signal. The strong analytic signal can be written with an amplitue an a phase where the erivative of the phase can be ientifie as the instantaneous frequency. It is not har to see that a function an its Hilbert transform also are orthogonal. This orthogonality is not always realize in applications because of truncations in numerical calculations. However, a function an its Hilbert transform have the same energy an therefore the energy can be use to measure the calculation accuracy of the approximate Hilbert transform. The computation of the Hilbert transform is a convolution of input signal xt () with ht (), since ht () is not an integrable function. The Hilbert transform is efine by using the Cauchy principal value as expresse in (3) 1 1 HT( x(t))= ( ) π x t 1 1 x() t a jb π t = a b 2 2 an In the above equation x(t) is the instantaneous current signal of each phase. The etector is use for etection of high impeance fault. In the propose technique, the current ata are sample with a sampling frequency of 1 KHz. The Hilbert Transform of the instantaneous current is obtaine. The etector is estimate using (3). The etector will eclare the occurrence high impeance fault when the magnitue of is more than the threshol value. After simulation of ifferent situations, the threshol value is set at 32. IV. RESULTS The system with istribute parameter line moel as shown in Fig. 1 is simulate using MATLAB Simulink an the results are provie in the following. The ata sampling rate was maintaine at 1 khz. (3) A. Results for uncompensate line A high impeance fault is create at position F in line-2 at.4 s with the values of V p an V n are of 155 kv an for R p an R n are of 7. Each phase current is sample at a sampling frequency of 1 khz. Corresponing performance plot is shown in Fig. 4. The instantaneous current of phase-a contains other frequency components than the funamental one. Hilbert transform of the instantaneous current of each phase is carrie out an the etector is shown in each figure. From Fig. 4(a) it is clear that the magnitue of the etector is increasing after.4 s. The threshol value is maintaine at 32. When the value of is crosse 32 then the high impeance fault is eclare. The high impeance fault is etecte after 3 ms. The magnitue of for other two phases are shown in Fig. 4(b) an (c). For other phases magnitue of the etector is less than (a) (b) (c) Fig. 4. Performance for high impeance fault in phase-a. (a) For phase-a. (b) For phase-b. (c) For phase-c ( with V p =V n= 155 kv, R p =R n = 7 ). 3

4 Another set of high impeance fault current ata are obtaine by ecreasing the magnitue of the variable resistance (R p =R n = 35 ). At this situation the magnitues of the DC source voltages are kept constant. During this situation the fault is create at position F in line-2 at.4 s. Corresponing results are shown in Fig. 5. The instantaneous current of phase-a contains other frequency components in comparison to the phase-a current of previous case. Hilbert transform of each phase is evaluate an the performance plots are shown below. Here for phase-a, when the value of etector is increasing more than 32, the fault is eclare as shown in Fig. 5(a). For other two phases, the magnitue of the etector is well within 32 as shown in Fig, 5(b) an (c). Another set of high impeance fault current ata are store by varying the magnitue of DC source voltage to 145 kv an keeping the variable fault resistance constant. For this case, performance plot is shown in Fig. 6. The magnitue of etector is more than 32 for fault in phase-a as shown in Fig. 6(a). For other two phases, the magnitue of the etector is within 32 with confirmation that there is no fault in phase-b an phase-c (a) (b) (c) Fig. 5. Performance for high impeance fault in phase-a. (a) For phase-a. (b) For phase-b. (c) For phase-c( with V p =V n= 155 kv, R p =R n = 35 ). B. Results for series compensate line The three phase power system as shown in Fig. 7 is consiere an line-2 is compensate by 7% where the capacitor is place at the relay en as shown. At times the capacitors may be in the circuit or bypasse. A high impeance fault is create at position F in line-2 at.4 s with the values of V p an V n are of 155 kv an for R p an R n are of 7. Each phase current is sample at a sampling frequency of 1 khz. Corresponing performance plot is shown in Fig. 8. For this case the fault is etecte correctly as the magnitue of the etector is increase than the threshol value. For other two phases the magnitue of the etector is well within the pre-set value. C. Results for single-pole tripping situation M Line1 Line2 F ~ ~ Source L Fig. 7. The three phase power system. Fig. 6. Performance for high impeance fault in phase-a. (a) For phase-a. (b) For phase-b. (c) For phase-c( with V p =V n= 145 kv, R p =R n = 35 ). HIF Single-pole tripping (SPT) operation imposes problem to many power network relaying schemes. To test the performance of the propose N Source-2 4

5 Fig. 8. Performance for high impeance fault in phase-a with 7 % series compensation in line-2. (a) For phase-a. (b) For phase-b. (c) For phase-c ( with V p =V n= 155 kv, R p =R n = 7 ). technique, SPT situation is create in line-2. Line-to-groun fault of bg-type is create at F an the corresponing results are shown in Fig. 9. During this situation the bg-fault is etecte correctly by using the propose technique as shown in Fig. 9(a) x Fig. 9. Performance for high impeance fault in phase-b with single-pole tripping in line-2. (a) For phase-a. (b) For phase-b. (c) For phase-c ( with V p =V n= 155 kv, R p =R n = 7 ). D. Results for loa change Loa changes continuosly in a power system an therefore voltages an currents are affecte at the relay bus. If fault etection unit ientifies the situation as a fault, many relaying schemes may take ecision that there is a fault. During loa change, the current in each phase is very less may be same as in case of high impeance fault. To test the performance of the algorithm uring loa change, suenly the loa angle between the sources is change from 1 to 2. Corresponing results are shown in Fig. 1. It is foun that uring loa change the magnitue of the etector is within the threshol value. As the etector value is not crossing the threshol then at loa change the etector unit will remain silent Fig. 1. Performance for loa change. (a) For phase-a. (b) For phase-b. (c) For phase-c ( with V p =V n= 155 kv, R p =R n = 7 ). V. CONCLUSION The issue of high impeance fault situation for a transmission system is aresse. An alternate solution is propose using the Hilbert transform of the instantaneous current of each phase. The problem of selecting line or bus sie voltage for the etection 5

6 process can be overcome by using the propose technique, as it oes not require fault voltage component. The metho is teste for three types of fault situations. The performance of the algorithm is also teste for series compensate line, single-pole tripping situation an loa change cases. The performance of the algorithm shows its consistency an accuracy in etecting the high impeance fault. REFERENCES 1. A. V. Mamishev, B. D. Russell, an C. L. Benner, Analysis of high impeance faults using fractal techniques. IEEE Transactions on Power Systems, vol. 11, no. 1, pp , A. Girgis, W. Chang, an E. Makram, Analysis of high-impeance fault generate signals using a Kalman filtering approach, IEEE Transactions on Power Delivery., vol. 5, no. 4, pp , High impeance fault etection technology, Rep. Power Syst. Relaying Committee (PSRC) Working Group D15, Mar A. Seighi, M. Haghifam, O. Malik, an M. Ghassemian, High impeance fault etection base on wavelet transform an statistical pattern recognition, IEEE Trans. Power Del., vol. 2, no. 4, pp , A. E. Emanuel, D. Cyganski, J. A. Orr, S. Shiller an E. M. Gulachenski, High impeance fault arcing on sany soil in 15kV istribution feeers: contributions to the evaluation of the low frequency spectrum, IEEE Transactions on Power Delivery, vol. 5, no. 2,pp , T. M. Lai, L. A. Snier, E. Lo, an D. Sutanto, High-impeance fault etection using iscrete wavelet transform an frequency range an RMS conversion, IEEE Transactions on Power Delivery, vol. 2, no. 1, pp , V. L. Buchholz, M. Nagpal, J. B. Neilson, R Parsi-Feraioonian an W. Zarecki, High impeance fault etection evice tester, IEEE Transactions on Power Delivery, vol. 11, no. 1, pp , W. H. Kwon Gi Won Lee Young Moon Park Man Chul Yoon Myeong Ho Yo, High impeance fault etection utilizing incremental variance of normalize even orer harmonic power, IEEE Transactions on Power Delivery, vol. 6, no. 2, pp , Davi C. Yu an Shoukat H. Khan, An aaptive high an low impeance fault etection metho, IEEE Transactions on Power Delivery, vol. 9, no. 4, pp , Essam M. Aboul-Zahab, El Saye Tag Elin, Doaa khalil Ibrahim, Saber Mohame. Saleh, High impeance fault etection in mutually couple ouble-ene transmission lines using high frequency isturbances, 12 th International mile-east power system conference, Aswan, N. I. Elkalashy, M. Lehtonen, H. A. Darwish, M. A. Izzularab, an A. I. Taalab, Moeling an experimental verification of a high impeance arcing fault in MV networks, IEEE Transactions on Dielectrics an Electrical Insulation, vol. 14, no. 2; pp , A. R. Seighi an M. R. Haghifam, Simulation of high impeance groun fault in electrical power istribution system, International conference on power system technology, Hangzhou, E. Sortomme, S. S. Venkata, an J. Mitra, Microgri protection using communication-assiste igital relays, IEEE Tansactions on Power Delivery, vol. 25, no. 4, pp , S. R. Nam, J. K. Park, Y. C. Kang, an T. H. Kim, A moelling metho of a high impeance fault in a istribution system using two series time varying resistances in EMTP, in Proc. IEEE Power Eng. Soc. Summer Meeting, Vancouver, BC, Canaa,, vol. 2, pp , Jul