DISTANCE PROTECTION FOR SIX-PHASE TRANSMISSION LINES BASED ON FAULT INDUCED HIGH FREQUENCY TRANSIENTS AND WAVELETS M. M. Mansour H. E. A. Talaat S . 0. Faried Ain Shams University Cairo, E a p t mmsmansour@ieee.org King Saud University Riyadh, Saudi Arabia htalaat@kr;u.edu.sa Universityof Saskatchewan Saskatoon, Canada sherif_of aried@engr.usask.ca Ammar A. Hajjar Tishreen University Latakia, Syria aahajjar@hotmail.com Abstract Six-phase transmission is an optimal solution for increasing the power transmission capability of overhead power transmission lines over existing rights-ofiway. This new technology, however, causes some protection problems. This paper introduces a new technique based on wavelet transform for high-speed distance protection of six-phase transmission lines. The technique utilizes a wavelet to capture the fault induced high frequency transient currents superimposed on the power Pequency modal currents. The trip decision is based on the relative arrival times of these high frequency signals at the relaying point. The introduced technique is tested and validated through simulation studies. The results show that the technique is accurate and high-speed in fault detection irrespective of the fault vpe, fault inception angle, and fault resistance. Moreover, it is insensitive to the line terminals and transposition. Keywords: SLx-Phase Transmission Lines; Distance Protection; High Frequency Transient; WaveletAnalysis. 1. INTRODUCTION Three-phase transmission system encounters some problems such as increasing demand and restrictions on rights-of-way (environmental laws). In this respect, sixphase transmission system is introduced as an optimal solution for the aforementioned problems. The existing double-circuit three-phase transmission line can be successively converted to a singlecircuit six-phase line [1,2]. In this context, the first empirical six-phase line in the world was energized in USA in 1992 [2]. Therefor, there is a need to establish protection schemes for six-phase transmission system because the additional three-phases complicate the fault analysis and consequently the required protection. In this respect, the number of shunt faults is “120” in a six-phase system whereas it is only “1 1” in a three-phase system [3]. So far, there is a few protection schemes dedicated for sixphase transmission lines [4-61. In this paper, a new protection approach based on wavelet transform (WT) and distance protection is presented. The WT of a signal consists of measuring the similarity between the signal and a set of translated and scaled version of the basis function “mother wavelet”. The mother wavelet is a chosen fast decaying oscillatory function The WT maps a given nonstationary signal from the time domain into time-frequency (scale) domain. The main advantage of the WT is its ability to extract a tiny discontinuity on a disturbed signal [7]. T h s feature is used for detecting the abrupt disturbances such as faults in transmission lines [8-lo]. In this paper, a useful application of the WT is presented for distance protection of six-phase transmission lines. The technique is based on fault induced high frequencies (HF) transient modal currents captured using wavelets and the travelling wave theory. Extensive computer simulations of the proposed scheme show that the technique is accurate and hgh-speed irrespective of fault type, fault inception angle and fault resistance. Moreover, it is unaffected by line terminals and line transposition. 2. WAVELET TRANSFORM Wavelet transform is relatively a new mathematical techmque for a nonstationary signal analysis. In this respect, the WT ofa time dependent signalfor) consists of finding a set of coefficients C/a,b) that measure the similarity between the signal and a set of scaled (compressed or dilated) and translated (shifted) versions of a function y(t) called the mother wavelet that given by: where “a” and “b” represent the time dilation and translation respectively. The selection of the mother wavelet depends on the application. The coefficients C!a,b) , or the continuos wavelet transform, are defined by the following inner product: Proceedings of the 2002 IEEE Canadian Conference on Electrical & Computer Engineering 0-7803-7514-9/02/%17.00 0 2002 IEEE -7- The arrived phase signals are first transformed to their modal signals. The first mode (mode 0) is frequency dependent and refers to the ground mode that is usually used to distinguish grounded faults. The other five modes (model ...mode5) are frequency independent and refer to the aerial modes that are present for any kind of fault. Accordingly, the fault distance is determined based on the aerial rnodes and their velocity. I-- CWT(a,b)= If (t).;t,,dt).dt (2) -00 where “*” refers to complex conjugate. Wavelet transform of a sampled signal can be obtained by using the following discrete wavelet transform (DWT): where, uom,k represent a time exponential dilation and’ time shift, respectively, n, k, a. are integers; % is some selected spacing factor (usually chosen equal to “2” for dyadic grid), and the dilation index (scaling) m is 0,1,2,3 ,.... The DWT analysis involves successive pairs of lowpass and high-pass filters at each scaling stage of the WT. The first scale covering a large frequency range at the high frequency end of the spectrum with the highest time resolution. The higher scales covering the lower end of the frequency spectrum with progressively shorter bandwidths with increasinglylonger time interval [7-91. 3. FAULT INDUCED TRANSIENTS AND 4a’WAVELET FOR SIX-PHASE LINES DISTANCE PRQTECTION The proposed approach uses wavelet transform as a fault induced HF transient currents detector. For this purpose, it is tuned to extract the HF currents superimposed on the modal power frequency currents at freque.ncy ranges between 50 and 100 IcHz. In this respect, the so-called“D4” mother wavelet, which is more suitable for transient analysis, time-frequency localizing, is used [9]. The fault distance calculation is based on the reflections times of the HF signals either from the fault point only or from both the fault point and far-end bus. This depends on the existence of a connection between the fault and the ground. MODAL TRANSFORM 4.1. Ungrounded Faults When a fault occurs on a power transmission line, high frequency transient signals of currents and voltages are induced at the fault point. These HF signals travel toward the line ends then they reflect back and forth between the fault point and the line ends until the post fault steady state is reached. However, these HF signals, which contain a wealth information about the fault type, distance, and its direction are superimposed on the power frequency signals of the faulted and unfaulted phases due to the mutual coupling between phases. Therefore,modal transformation is used to decouple the phase signals into their respective aerial and ground modes. The relation between the phase currents and the modal currents given by: Iphase = T Imde When the fault is ungrounded or symmetrical, the reflection from the remote end is insignificant and the fault distance x is determined by measuring the time interval between the first two consecutive peaks of similar polarity of the WT coefficients of the considered signal as follows: x=- V .z (6) 2 wherc: V is the wave velocity of the aerial mode and z is the time interval between the two consecutive peaks of the WT coefficients. On the other hand, if the first two consecutive peaks are of opposite polarities, then fault is considered out of protection zone. (4) where Tis the modal transformationmatrix and Iphme,Imode are the phase and modal current vectors, respectively. In this study, the six-phase transmission line is assumed ideally transposed; therefore the Clarke’s constant and real transformationmatrix is used [4]: (5) 4.2. Grounded Faults For a grounded fault the reflections from the fault point and Rrom the remote end buses will be observed at the sendhg end of the faulted line. Depending on the fault distance, the reflections from the remote end buses may arrive before or after those reflections from the fault point. It c m be easily verified by using the lattice diagram ill arrive method, that the remote end bus reflections w later than the fault reflections if the fault occurs within the first half of the line. The opposite will be true if the fault occurs in the second half of the line. Another problem arises due to a ground connection is that the reflections may be from the adjacent (faultedunfaulted) line. -8- Therefor, a suitable algorithm is developed to distinguish the faulted line and the fault distance as follows: If the first two consecutive reflections are with same polarity, the fault is in the first half of the line. Moreover, if the time interval between these reflections plus the time interval between the first one and the first significant reflection of opposite polarity is equal to twice the traveling time of the considered line, the fault is on the considered line. Otherwise the fault is on the adjacent line. If the first two consecutive reflections are with opposite polarities, the fault should be in the second half of the line. Moreover, if the time interval between these reflections plus the time interval between the first one and the first significant reflection of same polarity is equal to twice the traveling time of the considered line, the fault is on the considered line. Otherwise the fault is on the adjacent line. 0 If the two peaks of the WT coefficients are equal this refers to coincidence between the reflected signals from the fault and from the far-end bus. Fig. 1 shows a schematic block diagram of the proposed distance protection approach 5- SIMULATION RESULTS A verification of the developed approach with practical cases is carried out with the help of electromagnetic transient program (PSCADEMTDC) [ 111. The algorithm of the proposed approach is programmed under the MATLAB environment and using its wavelet toolbox [ 121. The study is conducted on a 230 kV six-phase power transmission system shown in Fig. 2. The line parameters are considered to be frequency dependent. The source’s impedances Zsl, Zs2 and Zs3 are 20, 10 and 20 ohms respectively. The arc resistance is included in the fault model. L1= 200 km I L,2=1OOkm I Fig. 2. A one line diagram of a six-phase power transmission system. The system is simulated under various types of faults at different locations, fault resistances, source configurations and fault inception angles. Moreover, the line transposition and untransposition are also considered. The relay is located at bus 1. A sampling frequency of 200 kHz is used and the CWT at scale 2 is used to capture the HF current signal superimposedon the aerial modes. 5.1 Source Configuration Transform Wavelet Transform Ungrounded fault A Out of zone fault Trip decision Fig. 1. A flowchart of the proposed distance protection approach. Fig. 3 depicts the CWT coefficients of the modal current signal corresponding to phase “a” to ground fault at 180 km from source S1 on line L1 for fault inception angle of 90” of phase “a”. Fig. 3.a corresponds to S1, S2, S3 impedances of 20, 10 and 20 i2 respectively, whereas Fig. 3.b corresponds to equal sources impedances of 20 R. As shown in Fig. 3.a and 3.b, the first and second HF transient signals arrive at busbar 1 with positive and negative polarities at times tl = 5.122 ms, t 2 = 5.255 ms respectively, this signifies that the fault is at the second half of the line. The time interval z1between tl, t2 is 0.133 ms and the time interval z2between tl and tlois 1.2002 ms, where tlois the arrival time of the first significant positive reflection, tlo= 6.3222 ms. Hence, z = 21+22= 1.3332 ms which correspond to the 399.98 km which equals to the twice of the line length of L,. Therefore, the faulted line is L1and the fault distance is 180.05 km.Accordingly a trip decision from the relay near bus1 should be issued to the dedicated circuit breaker. This assures that the technique is insensitive to the system source configuration. This is due to the busbar capacitances, which act as a short circuit at the considered range of frequency. -9- 3-15 15 C 2 IO. 3 n 2 w- 0 0.15 I a L L 2 m + + 50-5- -a - O.O!j 1 ' - - w- 0 iv),-0.15 v) e-10. 5 Time,ms 3-15L I k -o.o!j - -0.25 - 'tl -1c t2 Time,ms 5.4 Fault Types 5.2 Fault Resistance Fig. 4 shows the WT coefficients of the modal transient current for a single line to ground fault 'f-g' at 25 km from busl for a 135' fault inception angle and 400 i2 fault resistance. Despite the WTC magnitude reduction, the wave shapes of the captured signals remain unchanged and accurate fault distance is calculated as follows: tl = 4.6 ms, tz = 4.764 ms and t4 = 5.772 ms, thus z 31-4. Hence, L1 is the faulted line and the fault distance is 24.6 lan. The obtained result proves that the proposed approach is insensitive to the fault resistance. Fig. 6 shows the WT coefficients for symmetrical 3-phase fault '"b-d-f' at 130 km of SI. Since WO= 0 and the first two consecutive reflections are with similar polarities, the fault is at L1and the fault distance is130.5 lan. = In, U) Fig. 6 . A 3-phase "b-d-f' fault at 130 km of line Ll. Fig. 7 shows the WT coefficients for a 4-phase fault "cdAlso, since (WO= 0) and the first two e-f' at 60 km of L. consecutive reflections are with opposite polarities the fault is on the adjacent line and the relay will reset. 4.5 5 5.5 6 Fig. 4. A line-to-ground "f-g" fault at 25 km from S1 with ~p = 135" and Rf = 400 a. 5.5 Untransposed Line Due to untransposition, mutual couplings arise between the phases which consequently affect significantly the ground mode signal, whereas the aerial mode signals are - 10- References 5.6 5.4 5.8 6 6.2 6.4 Fig. 7. A 4-phase “c-de-f” fault at 60 km of line L2. slightly affected. Since the proposed method is based on the aerial modes, the WT coefficients will not affected by l i e untransposition as shown in Fig. 8, which corresponds to a six-phase fault at 130 km of the untransposed line L1. Therefore, the proposed approach is suitable and accurate for both transposed and untransposed lines. E 20‘ ? 15. -3 1 0 . 5. t tl B -5. -10. 3 - -20 -D+ t3 L o $ -15. J + c - v 4 1 Tme,ms Fig. 8. A six-phase fault at 130 km fiom S1 of untransposed line LI. an Finally, it is worth noting that the proposed distance protection technique is complemented by the authors with a fault classification and phase selection technique 161. 6- CONCLUSION This paper presents a new distance protection technique based on the WT and a fault induced HF transients to be used for six-phase transmission. The technique utilizes the WT for extracting the fault induced HF transient signals, and the traveling waves theory for determining their consecutive arrival times. Studies show that the technique is high-speed and accurate irrespective of the fault type, fault inception angle and fault resistance. Moreover, it is not affected by the line tennmls and line transposition. The accuracy of the technique is proportional to the sampling frequency rate; however the error in determining the fault distance in the worst case does not exceed 1.5 km. This accuracy could be improved either by increasing the sampling frequency rate or by using a smoothing algorithm [ l ] J. R. Stewart, and D. D. Wilson, “High phase order transmission part I and 11,”IEEE Trans. on PAS Vol. 97, No. 6, Nov./Dec. 1978,pp. 2300-2317. [2] R Brown, and J. R. Stewart, “Six-phase successfully applied to utility transmission system,” CZGRE paper 22/33136-01Paris, 1998. [3] S. S. Venkata and et al, “Six phase (multi phase) power transmission systems fault analysis,” ZEEE Trans. on PAS, Vol. 96, No. 3, MayIJune. 1977, pp. 758-767. [4] A. A. Hajjar, M.M. Mansour, and H. A. Talaat, “Travelling wave-based protection of six-phase transmission lines,” The 41h Conference of Arab CZGRE National Committees, Mars 18-21,2001, TripoliLibya. [5] R. V. Rebbapragda et al, “Selection and application of relay protection for six phase demonstration project,” IEEE Trans. on Power Delively, Vol. 7, No. 4, October 1992,p.p, 1900-1911. [6] A. A. Hajjar, M.M. Mansour, and H. A. Talaat, “Wavelets for six-phase transmission line relaying: fault classification and phase selection,” The I l l h ZEEE MELECON 2002 Conference, May. 27-29. 2002, Cairo, Egypt, accepted. [7] 0. Rioul, and M. Vetterli, “Wavelet and signal processing,” ZEEE SPM, Vol. 8, No. 4, October 1991, pp. 14-38. [8] 0. Chaari, and M. Meunier, ‘Wavelets: a new tool for resonant grounded power distribution systems relaying,” IEEE Trans. On Power Delivety, Vol. 11, No. 3, Ju& 1996, pp. 1301-1308. [9] D. C. Robertson, and 0. I. Camps, “Wavelets and electromagnetic power system transients,” ZEEE Trans. On P K W , Vol. 11, No. 2, April 1996, pp. 1050-1058. [lo] A. A. Hajjar, M.M. Mansour, and H. A. Talaat, “Signal processing using wavelet transform for power transmission lines protection,” The I” ZEEE ZSSPZT Symposium, Dec. 28-30. 2001, Cairo, Egypt. [l I] EMTDC, Manitoba HVDC Research Center, Canada, 1999. [ 121 ‘Wavelet toolbox for use with MATLAB”, The Math Work Inc, 1996. - 11 -