An Ellipse Technique Based Relay For Extra High Voltage Transmission Lines Protection

Proceedings of the 14th International Middle East Power Systems Conference (MEPCON 10), Cairo University, Egypt, December 19-21, 2010, Paper ID 162. An Ellipse Technique Based Relay For Extra High Voltage Transmission Lines Protection Ali M. El-Rifaie Sohair Fakhry Alaa M. Hamdy S. M. Moussa E.H.Shehab El_din National Institute of Standards (NIS) -Egypt Faculty of Engineering, Helwan University-Egypt alisystem11@gmail.com, sohairfakhry@hotmail.com {sabrymmm1966&shehab_eldin01}@yahoo.com Abstract -- This paper introduces a new protective relay for Extra High Voltage (EHV) transmission lines, the relay is based on the new Ellipse Technique (ET) and is used with a transmission line carrying the Egyptian 500 Kv unified network parameters. The relay has two operating criteria that work together to detect fault presence and identify fault current direction. The capabilities of the ET are used to obtain a reliable operation of the relay during fault conditions in a time of about 5 msec (F=50 Hz). The simulated results showed that, one of the features of the produced image can be safely used to detect fault presence, identify the faulted phase(s) and fault type besides differentiating between forward and reverse faults. Index Terms Extra High Voltage (EHV) transmission lines, digital relaying, Ellipse Technique (ET), Two Dimension (2D) centre I. INTRODUCTION Transmission line is one of the main components in every electric power system, it is exposed to all possible environmental conditions where the possibility of experiencing faults is generally higher compared to the other components. Line faults are the most common faults triggered by lightning strokes, short circuit fault, and over voltages, When a fault occurs in a transmission line it is important to detect the fault in order to make necessary repair and to restore power as soon as possible without any maloperation, the time required to detect the fault in the line will affect the quality of power. The speed and accuracy of digital relays for transmission lines can be improved by accurate and fast fault detection and classification [2]. The high frequency based relays seem to be a good solution for obtaining a fast as well as reliable protection, several methods are used including travelling waves [3], wavelet transforms [4,5] as well as the Cos-Sin [6,7]. Image features analysis [1] is a new scope that can be used to fulfill both the reliability and speed requirements for transmission lines digital protective relays. In this paper a new technique based on the image features measurements and named Ellipse Technique "ET" [1] is used in obtaining a new fast and reliable digital relay for EHV transmission line protection, the ET produces an image that describes the current versus voltage signals plot, the produced image has a uniform elliptical shape during normal operating conditions and has irregular shapes during fault ones. The suggested relay has two operating criteria, the first criterion is used to detect fault presence, while the second detects fault direction, the relay needs 5 msec starting from the instant of fault occurrence to produce an output signal. The relay operation is examined with a simulated [8] extra high voltage transmission line with the Egyptian 500 Kv network parameters. The results showed the suggested relay capability of detecting all fault types and differentiating between internal and external faults. II. EHV NETWORK USED IN SIMULATION The EHV transmission network used in simulation is shown in Fig. 1, the line parameters are those of the 500 Kv unified network in Egypt. The relay is fed by the line current via a C.T and bus X voltage via a V.T and is aiming to protect the line connecting buses X and Y. The short circuit capacity behind each busbar is also given, a sampling frequency of 10 KHz is used in this simulation with a 50 Hz power frequency. The relay has two operating criteria, the first is the fault detection criterion which uses the deviation in the value of the 2D centres being updated every received sample together with a fault detection threshold to detect fault presence, while the second is the fault direction criterion that determines the deviation direction of the irregular shapes immerging from the normal elliptical ones during fault occurrence in differentiating between forward and reverse faults [1]. 20 GVA G 3 Z 50 GVA 30 GVA X C.T Y G 1 G 2 200 Km 300 Km F 2 V.T Digital Relay Load Fig. (1) 500 kv, typical network used in simulation F 1 271

III. FAULT DETECTION CRITERION The fault detection criterion operates in two consecutive stages, the first stage works during normal operating conditions, where it applies the ellipse technique to determine the value of the two dimension (2D) centre of the elliptical image and detects the instant of fault occurrence [1], while the second stage starts with any abnormal condition to insure fault presence where 13 represents the value of X L /R ratio of the simulated 500 Kv transmission line. A. Operation of Stage One The current and voltage signals fed to this stage as well as the output elliptical images are shown in Figures 2, 3 and 4 respectively where it needs at least 20 msec (F=50 Hz) to obtain a full ellipse. Normalization is done by dividing the current signal on the maximum short circuit current value,and dividing the voltage signal by its peak value; the image obtained in Fig.5 will keep all the features of that plotted in Fig. 4 but it will be plotted between (-1) and (1) for both axes. The 2D centre is calculated following the four coming equations: X 0 = (1) Fig. 2 Three phase current signals during normal operating conditions Y 0 = (2) 2D = (X 0,Y 0 ) (3) K 0 =..(4) Fig. 3 Three phase voltage signal during normal operating conditions Where N: is the total number of samples in one complete cycle (200 samples with sampling frequency 10 KHz), the magnitude of the 2D centre K 0 is given by equation (4) Table -1- shows the numerical values of K 0 for all phases during normal operation after normalization. Table-1- numerical values for K 0 in normal operating conditions X 0 Y 0 K 0 Phase (A) 1.20 10-6 2.85 10-5 2.85 10-5 Phase (B) 4.28 10-6 3.83 10-5 3.85 10-5 Phase (C) 3.08 10-6 3.34 10-5 3.35 10-5 Fig. 4 Three phase elliptical shape of during normal operating conditions As long as the EHV transmission network is working normally, the calculated value of K 0 for at each sample should have the same value of that of its previous one; however a slight difference is naturally to appear due to sampling error. The difference K 0diff is calculated according to equation (5) K 0diff = K 0(N+i) K 0(N+i-1) (5) The maximum value of K 0diff detected during normal operation is 3 10-5, thus a fault detection threshold ξ is chosen as given in equation (6) ξ = 13 3 10-5 = 3.9 10-4. (6) Fig.5 Three phase normalized elliptical shapes during normal operation 272

B. Operation of Stage Two At any time if the value of K 0diff exceeded ξ that might indicate fault presence, at this instant stage Two starts to work where it is going to calculate K 0diff for the 50 coming samples (5 msec, F = 50 Hz,sampling frequency 10 KHz) according to equation (5) every two consecutive samples; at the end of the 50 samples an average value K oavg is calculated according to equation (7) K 0avg = (7) Where j: is the second stage initiating instant Fig.6, Angle of inclination theta during normal forward operation At the end of the 50 samples K 0avg is compared with ξ, where fault presence is declared if and only if K 0avg ξ at any phase. Fault declaration is done by changing the criterion state from 0 to 1 ; however this declaration is not complete as fault direction should be detected to avoid tripping for reverse faults, that is the job of the second criterion. IV. FAULT DIRECTION CRITERION The fault direction criterion works in two consecutive stages, the first stage aims to detect the elliptical image direction by calculating the inclination angle {Theta Ө } at the end of the first normally operating cycle (20 ms). The second stage starts at the instant when K 0diff ξ where the fault current direction is detected for the next 50 samples. A. Operation Of Stage One The polarity of the C.T used in Fig.1 is such that the current flowing from bus X to line XY is forward and vice versa. Every point sketched on the elliptical shape has a 2D value, whose magnitude can be calculated as previously done for K 0 in equation (4). Two optimum (maximum) values (K opt ) are detected, where Ө is calculated from the values of the optimum positive value of X X opt+ and its corresponding Y value Y opt, following equation (8) Ө = Tan -1 (X opt+ /Y opt ) (8) The inclination angle is shown in Figs. 6 and 7 for forward and reverse operation respectively, where it is placed in the first quadrant for forward operation (current flowing from bus X to line XY) and placed in quadrant Two for reverse operation (current flowing from line YX to bus X ). When Ө lies in quadrant one (forward direction), stage one in the fault direction criterion will produce a 0 output and produces an output of 1 otherwise B. Operation Of Stage Two The operation of stage two starts at the instant when stage one of the first criteria detects that K 0diff ξ and with a pre-fault status of 0, the second stage will detect the deviation direction of the irregular shapes from the normally elliptical Fig.7, Angle of inclination theta during normal reverse operation ones within 2 samples (0.2 msec using 10 KHz sampling frequency) [1]; For ground faults the faulty current signal is tending to increase in magnitude, consequently the deviation direction within the two samples will appear moving towards the direction of current (X-axis) increase for forward faults and towards the current decrease for reverse ones, this rule will remain observed up even with high fault resistances and for all possible fault inception angles. Figs.8, 9 and 10 show the three phase currents, voltages and elliptical shapes during a forward SLG fault (on Line XY) through a fault resistance of 50 ohms, while Fig. 11,12 and 13 show that for a reverse SLG one (on Line XZ). For Phase faults, the faulted phases currents are still tending to increase; however they will oppose one another, that appears in a diverting deviation direction of the irregular shapes of the faulted phases in forward faults, while in reverse faults they seem to approach (move towards one another). Figs 14 up to 17 show the three phase elliptical shapes together with the deviating irregular shapes for phase faults at different fault inception angles for both forward and reverse fault cases. The deviation direction is stored for all phases until the first criterion ends its work (after 50 samples) and the direction of the faulted phase(s) is only considered. A reverse fault keeps the output of 0, while an output of 1 is produced for forward faults. The relay s final decision will depend on the response of its two criteria, where there are Three possible cases for the relay operation as shown in Table-2-273

Fig.8, Three phase current signals during forward SLG-AG, 250 Km from bus X, through fault resistance of 50 ohms Fig12, Three phase voltage signals during reverse SLG-AG, 50 Km from bus X, through fault resistance of 50 ohms Fig 9, Three phase voltage signals during forward SLG-AG, 250 Km from bus X, through fault resistance of 50 ohms Fig.13, Deviation direction during reverse SLG-AG shown for 0.2 msec Fig.10, Deviation direction during forward SLG-AG shown for 0.2 msec Fig.14, deviation directions during forward LL-AC fault shown for a time of 0.2 msec, 250 Km from bus X Fig.11, Three phase current signals during reverse SLG-AG, 50 Km from bus X, through fault resistance of 50 ohms Fig.15, deviation directions during reverse LL-AC fault shown for a time of 0.2 msec, 50 Km from bus X 274

Start Read the three phase voltages of Bus X and the three phase currents of line XY Plot the current versus voltage to obtain the Elliptical shapes every 20 ms Apply the ET to find K 0 and Ө for the first 20ms Fig.16, deviation directions during forward DLG-AB shown for a time of 0.2 msec250 Km from bus X, through R=50 ohms Update the value of K 0, and find the Value of K 0diff for every received sample No K 0diff ξ? Yes Calculate the value of K 0avg and detect the Current direction for 50 samples No K 0avg ξ? Yes Direction State=1? Fig.17, deviation directions during reverse 3P fault, shown for a time of 0.2 msec, 50 Km from bus X TABLE-2- Relay responses during loading and fault conditions Declare reverse fault Yes Output of First Criterion Output of Second Criterion Relay Response Notes 0 0 Nothing Loading 1 1 3-Phase Trip Forward Fault 1 0 Stand By Reverse Fault For a load current flowing in an opposite direction (Ө lies in quadrant 2), the previous rules for the deviation directions are reversed for all fault types, it appears moving towards a decrease in X-axis for forward SLG faults and in the increase directions for reverse ones, while for phase faults the diverting deviation direction appears for reverse faults and the approaching for forward ones, that is why it is necessary for stage one to determine and update the inclination angle Ө every 20 msec of healthy operating conditions. Fig. 18 shows the flow chart of the ET based relay describing its response during normal, forward and reverse faults. The relay response for load switching, fault inception angles, fault types, fault resistances as well as fault location are discussed in the next section. Send a Trip signal and declare a forward fault on line XY END Fig.18, flow chart of the suggested ET based relay V. RELAY RESPONSE DURING LOAD SWITCHING One of the most confusing operations to the chosen threshold should be load switching that might be seen as a fault case if ξ was not chosen correctly, the load at bus Y is having a value of 785 MVA and is switched at different instants [1]. The three phase currents as well as the three phase elliptical shapes during load switching are shown in Figs 19 to 22. The presence of D.C components in the current signals caused K 0diff to exceed the working threshold ξ at certain switching instants; however the value of K 0avg calculated for 50 samples is still below the working threshold value (ξ=3.9 10-4 ), in this case the value of K 0avg will be calculated once more for one complete cycle (K 0avg(1) at N= 200 Samples) and the obtained values will be compared once more with the value of ξ to avoid any wrong decision. Table-3- shows the value of K 0diff, K 0avg and K 0avg(1) for all load switching instants of, where all the values are multiplied by 10-4. 275

TABLE-3- K 0avg and K 0diff calculated at different switching instants Switching Instants K 0diff K 0avg for N=50 K 0avg(1) for N=200 A B C A B C A B C T=t 1 2.1 2.3 1.2 T=t 2 1.6 2.2 2.1 T=t 3 2.4 4.1 3.8 1.9 2.4 2.2 0.5 0.7 0.3 T=t 4 3.0 3.2 2.6 2.3 2.3 1.8 0.4 0.7 0.5 Fig.19, three phase current signals during load switching at t=t 3 T=t 5 3.1 3.6 4.1 2.4 2.5 2.8 0.8 0.3 1.2 T=t 6 2.0 2.0 1.8 T=t 7 2.4 1.5 2.6 T=t 8 3.0 2.2 3.1 2.1 1.7 1.9 0.6 0.5 0.8 T=t 9 2.1 2.3 1.1 Fig. 20 three phase elliptical shapes during load switching, at t=t 3 As shown in Table-3, at some switching instants (t 3 &t 5 ) K 0diff exceeded the value of ξ that initiated the second stage of both criteria to operate ; however the values of K 0avg are still beyond the value of ξ, that is also ensured by the value of K 0avg(1) calculated for one complete cycle (20 msec) starting from the initiation instant of stage 2 of both criteria. VI. RELAY RESPONSE DURING FAULT CASES During fault occurrence, the value of K 0diff as well as K 0avg will change rapidly exceeding the value of ξ; however the effect of fault location, fault resistance and fault inception angles should be investigated to insure the relay s reliability. The three phase elliptical shapes during different fault types at different fault locations are shown in the Figs. 23 to 26, while Fig.21, three phase current signals during load switching at t=t 5 Fig. 22 three phase elliptical shapes during load switching, at t=t 5 Table-4- shows the values of K 0avg during different fault types at different fault locations. Table-5- shows the values of K 0avg during SLG-AG faults through different fault resistance values up to 50 ohms, while Table-6- shows the values of K 0avg calculated during different fault inception angles for a SLG fault located at 150 Km on line XY. The results showed that the relay is capable of fault detection for all fault types at all fault inception angles and through ground fault resistances up to 50 ohms, remote end faults as well as nearby ones were easily detected. The results also showed the capability of determining fault types from the values of K 0avg(1) calculated for one complete cycle after fault occurrence where the relay will use these values after sending a trip signal during the first cycle quarter following fault occurrence, these values are shown in Table-7- where all the values of K 0diff as well as K 0avg are multiplied by 10-4 in Tables 4, 5,6 and 7 276

TABLE-4- K 0diff and K 0avg for different fault types at different locations along line XY Location From Bus X 25 Km 150 Km 275 Km Fig. 23 three phase elliptical shapes during SLG-AG, 275 Km from bus X Fault Type K 0diff K 0avg K 0diff K 0avg K 0diff K 0avg SLG AG L-L, AB DLG ABG 3Phase A 10.0 15.0 7.0 8.88 6.0 7.15 B 0.0 5.42 2.0 5.96 3.0 3.89 C 0.0 2.43 2.0 2.43 3.0 0.18 A 5.0 12.5 6.0 13.0 6.0 9.67 B 5.0 10.1 6.0 8.18 6.0 10.0 C 0.0 0.69 0.0 2.86 0.0 3.60 A 10.0 14.25 10.0 14.0 9.0 10.0 B 6.0 8.95 6.0 8.92 8.0 11.0 C 0.0 0.37 1.0 2.91 1.0 4.18 A 12.0 12.0 12.0 18.0 12.0 14.0 B 5.0 7.86 5.0 12.0 7.0 14.0 C 7.0 5.44 8.0 4.65 5.0 5.58 TABLE-5- K 0diff and K 0avg for SLG-AG faults throughout line XY Fig. 24 three phase elliptical shapes during LL-AB, 75 Km from bus X Location From Bus X 25 Km 150 Km 275 Km Fault Resistance Value K 0diff K 0avg K 0diff K 0avg K 0diff K 0avg A 10.0 15.0 7.0 8.88 6.0 7.15 R = 0 Ω B 0.0 5.42 2.0 5.96 3.0 3.89 C 0.0 2.43 2.0 2.43 3.0 0.18 A 5.0 13.9 4.0 8.40 4.0 6.78 R = 25 Ω B 2.0 5.13 2.0 5.66 2.0 3.61 C 1.0 0.91 2.0 2.31 2.0 0.06 A 4.0 12.4 4.0 7.80 4.0 8.84 R = 50 Ω B 1.0 4.92 2.0 5.25 2.0 0.72 C 0.0 0.67 2.0 2.13 2.0 0.76 TABLE-6- K 0diff and K 0avg for SLG fault, 150 Km from bus X at different switching instants Fig. 25 three phase elliptical shapes during DLG-ABG, 150 Km from bus X K 0diff K 0avg Fault Instant A B C A B C T=t 1 7.0 0.0 0.0 8.88 5.96 2.43 T=t 2 4.0 0.0 1.0 5.52 3.55 1.51 T=t 3 4.0 1.0 1.0 5.96 0.98 0.64 T=t 4 5.0 1.0 2.0 6.54 4.86 1.91 T=t 5 4.0 1.0 1.0 8.48 5.98 2.42 T=t 6 6.0 1.0 1.0 5.18 3.59 1.54 T=t 7 4.0 1.0 1.0 8.91 0.89 0.61 T=t 8 4.0 2.0 2.0 6.80 4.74 1.83 T=t 9 7.0 1.0 1.0 8.95 6.01 2.42 TABLE-7-K 0avg(1) calculated during different fault types for one complete cycle time after fault occurrence Fig. 26 three phase elliptical shapes during 3P, 225 Km from bus X 75 Km 225 Km K 0avg (1) SLG-AG K 0avg (1) L-L, AB K 0avg (1) DLG,ABG K 0avg (1) 3 Phase A B C A B C A B C A B C 38 2 2 193 193 0 144 242 2 50 336 285 12 3 2 68 68 0 50 86 1 14 123 108 277

VII. CONCLUSION VIII. REFERENCES In this paper a new digital relay for EHV transmission networks was introduced, the relay is based on the so called Ellipse Technique ET and has two operating criteria that work together to detect fault presence and fault directionality. The relay algorithm is tested during load switching at different switching instants and showed complete success in avoiding any wrong trips, the relay algorithm was also tested during all fault types, at different fault inception angles, at different locations and through fault resistances up to 50 ohms, where high reliable operation is observed during all fault cases; complete fault detection is done in only 5 msec without any problems regarding the relay s reliability. The introduced relay is characterized by being simple, without any complicated mathematical approaches, uses a fault detection threshold that is not affected in any way by harmonics presence, load value or power flow direction. The fact that the ET uses both the current and voltage signals without any need to filtering certain frequency values, makes it easy to apply the ET practically and reduces the overall protection time. There is no need to use high sampling frequencies to obtain the ET and that makes it more easy to be practically applied. The introduced relay is characterized by being fast, simple, reliable, and easy to be applied. [1] Ali M. El-Rifaie,Sohair Fakhry, Alaa M. Hamdy, S. M. Moussa and E.H.Shehab El_Din, A New Fault Detection Technique Based On Features Measurements Of Current Versus Voltage Image For Extra High Voltage Transmission Lines,MEPCON 2010. [2] M. Kezunovic and I. Rikalo, Detect and classify faults using neural nets, IEEE Computer. Appl. Power, vol. 9, pp. 42 47, Oct. 1996. [3] Xiao an Qin, Xiangjun Zeng, Yijie Zhang, and Zhihua Wu, HHT Based Non-unit Transmission Line Protection Using Traveling Wave, IEEE Transactions on power delivery, April 2008 [4] S. M. Moussa, M. M. Eissa, E. H. Shehab-eldin & M. Masoud Real Time Modeling For EHV Transmission Line and Fault Analysis Using a Modified Wavelet Function MEPCON 0221, 29-31 December, Helwan University. [5] Nan Zhang, and Mladen Kezunovic Transmission Line Boundary Protection Using Wavelet Transform and Neural Network, IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 22, NO. 2, APRIL 2007 [6] M.E. Masoud, E.H. Shehab El_din and Ali M. El-Rifaie " A New Transient Detection Tool For Extra High Speed Relays" MEPCON 2003,VOL. 2, NO.2,PP.(559-563), DECEMBER 2003 [7] M.E. Masoud, E.H.Shehab El_din and Ali M. El-Rifaie " A Protection Technique For Extra High Voltage Transmission Lines Using A New Supervising Relay", MEPCON 2003, VOL.1,NO.1,PP.(377-383),DECEMBER 2003 [8] ATP DRAW version 3.5 for Windows 9x/NT/2000/XP Users'Manual 278