A New Hierarchical Approach for Modeling Protection Systems in EMT-type Software

Transcription

1 A New Hierarchical Approach for Modeling Proecion Syem in EMT-ype Sofware Henry Gra, Jean Maheredjian, Emmanuel Ruovic, Ula Karaagac, Aboualeb Haddadi, Omar Saad, Ilhan Kocar, and Ali El-Akoum Abrac-- Thi paper preen a new hierarchical approach for modeling and imulaion of proecion yem in an EMT-ype ofware. Variou proecive device, uch a relay and fue are aembled uing block diagram hrough a hierarchical rucure. The advanage of he choen developmen mehod are demonraed in hi paper. I i hown ha depie he complexiy level of proecive relay, accepable compuaional performance can be achieved when imulaing numerou proecion device in he EMT approach. Even beer compuaional performance are achieved by imulaing in a muli ime-ep environmen. Simulaing relay in ime-domain provide variou benefi: repreenaion of nonlineariie uch a ranformer auraion; incluion of harmonic; and accurae repreenaion of power elecronic-baed device. Keyword: Proecion, relay, fue, power wing, ou-of-ep, differenial proecion, muli-ime-ep, muli-core. T I. INTRODUCTION RADITIONALY, proecion yem udie are performed in pecialized ofware package where he power yem are imulaed in phaor-domain. Thi approach aume ha he power yem i linear and a fundamenal frequency only. Alhough accepable for many power yem udie by aking ino accouning he given aumpion, hi approach may encouner everal limiaion in he conex of modern power yem. Thi i epecially more imporan wih he increaing inegraion of power elecronic-baed device and wih he growing accuracy need in variou yem udie for proecion applicaion. The accuracy iue in phaor-domain imulaion mehod are highlighed in everal publicaion. Imporan accuracy problem for hor-circui calculaion for MOV-proeced erie-compenaed line are preened in [2]. The impac of curren ranformer (CT) auraion on proecion yem i udied in [3] and [4]. Problem relaed o proecion yem for renewable are dicued in [5], [6] and [7]. The applicaion of phaor-domain and ime-domain H. Gra, J. Maheredjian, A. Haddadi and I. Kocar are wih Monreal Polyechnique, Monréal, QC H3T1J4 Canada (grahenry@gmail.com jean.maheredjian@polyml.ca aboualeb.haddadi@polyml.ca and ilhan.kocar@polyml.ca). E. Ruovic i wih PowerSy Monréal, QC Canada (e.ruovic@poweryoluion.com). U. Karaagac i wih he Hong Kong Polyechnic Univeriy, Hung Hom, Kowloon, Hong Kong (ula.karaagac@polyml.ca). O. Saad i wih Iniu de recherche d'hydro-québec (IREQ), Monréal, QC Canada (aad.omar@ireq.ca). A. El-Akoum i wih Élecricié de France (EDF), Pari, France (ali.elakoum@edf.fr). Paper ubmied o he Inernaional Conference on Power Syem Tranien (IPST2017) in Seoul, Republic of Korea June 26-29, 2017 mehod for proecion yem udie are ummarized in [8]. The circui baed ime-domain approach ued in he imulaion of elecromagneic ranien (EMT) i very accurae for he imulaion of complex yem wih muliple nonlineariie, convenional and renewable generaion ource, and power elecronic-baed yem wih wiching device wih all relaed conrol yem. The implemenaion of relay model in EMT-ype ool i challenging. The meaured ignal are imilar o hoe capured in he field ince he nework model repone include acual ignal wih noie and harmonic. The relay model mu include ani-aliaing filer, and phaor calculaion algorihm mu be repreened ince he inpu of he model are inuoidal wih harmonic. I i poible o accoun for variou delay uch a breaker opening a curren zerocroing. A a conequence, he EMT-ype relay model algorihm can and mu cloely imiae hoe found in he acual device. Thi reul in numerically heavy model. Anoher imporan complexiy in deailed relay model i he capabiliy o inveigae cae where he relay device doe no operae a expeced. In uch cae, he model implemenaion mu allow he uer (or developer) o idenify he acual relay funcion/block ha did no operae properly or had eing error. The analyi of problem can be hen brough o he level of phyical relay eing comparion and dicuion wih manufacurer. Relay model have already been developed in EMT-ype ofware [9], [10], bu he approach preened in hi paper innovae in everal apec for addreing he aforemenioned challenge. An open-archiecure approach i preened uing a hierarchical deign wih block-diagram. The model conen are viible and allow o navigae and cuomize differen funcionaliie. The hierarchical implemenaion i opimized o improve numerical performance and memory. Thi apec i very imporan due o he very large number of block found in a ypical relay model and for imulaing large cale power yem wih muliple relay. Compuaional peed i furher enhanced hrough a muli ime-ep parallel programming environmen. The model preened in hi paper are implemened in he EMTP ofware in [1]. II. GENERAL ARCHITECTURE A. Deign approach When deigning proecion device model in a ime-domain ofware here are everal deign crierion. I i neceary o

2 opimize numerical performance and memory uage. Thi i a challenging problem due o he involved complexiy level. The proecion device behavior mu allow o inveigae i precribed performance, debug and adju i eing. The proecion device model mu be reconfigurable and cuomizable. Reconfiguraion allow o implemen differen algorihm for each funcion. For example, in a diance relay, he volage polarizaion [11] calculaion ued for Mho characeriic depend on he ype of polarizaion (elf, croed, memorized, ec.) and differ from one manufacurer o anoher. Thu, i i imporan o be able o reconfigure he volage polarizaion funcion according o elecion becaue hi ha a ignifican impac on he ize of Mho zone. The model mu be cuomizable by allowing he uer o include or exclude differen funcionaliie according o he ype of udy and available informaion. One approach for opimizing performance i o implemen he relay algorihm uing acual code hrough a dynamic link library (DLL) inerfaced wih he ofware package. Thi black-box approach doe no allow he uer o acce he variou relay funcion and manipulae hem uing a graphical uer inerface (GUI). Debugging and adjuing eing become a complex proce and i i almo impoible o underand how he model work. The black-box approach doe no repec he deign crierion pecified above. In he approach developed here, he relay model i baed on block diagram organized in a hierarchical manner (ee Fig. 1). A op level ubcircui i given a mak and conain everal layer of ubcircui organized in a op-down rucure wih ranfer of daa and eablihmen of funcion hrough a hierarchy. Thi approach alo include ome DLL for baic funcion or funcion ha can be be implemened (for efficiency and faciliy) hrough a programming language, like a Dicree Fourier Tranform, for example. The propoed approach provide an open archiecure (open-ource). I offer everal advanage, bu creae a very large hierarchical yem wih inerconneced block diagram. A complee line proecion relay, for example, can conain up o device. I i hu required o organize he hierarchy hrough an objec-oriened approach. Once a relay i placed in a deign, adding anoher one i merely creaing a new inance of i objec which minimize memory increae. I i no poible o ue a ingle hierarchical objec ince he relay ubcircui (objec) would hen have o include he block diagram for all poible relay configuraion and opion, and occupy unnecearily large memory. Inead, he relay are buil wih everal macro-objec (ubcircui in he hierarchy, ee Fig. 1) which are auomaically acivaed or deacivaed according o op block elecion. Therefore, wo relay wih differen opion elecion can hare macro-objec and ave memory. For example, all relay wih inpu having a Wye conneced Volage Tranformer (VT) have he ame volage acquiiion funcion (Signal Acquiiion block in Fig. 1). All he one wih inpu o dela-conneced VT have anoher acquiiion funcion. A anoher example, all line proecion relay uing Mho zone have a common objec, bu migh have differen polarizaion volage (Memory Manager block in Fig. 1) according o he manufacurer elecion. Fig. 1. Diance relay model inernal archiecure. In order o quanify he compuer memory gain in he above deign approach, i i propoed o conider hree benchmark, namely, Benchmark 1, Benchmark 2, and Benchmark 3. Benchmark 1 i hown in Fig. 2. Benchmark 2 i he ame a Benchmark 1 excep ha i ha 3 more idenical relay (wo for he op line and one more for he boom line). Benchmark 3 (ee Fig. 3) ha 4 differen relay ype. ir vr Relay1 Generic G P Fig. 2. Benchmark 1: parallel line beween wo nework wih one relay. ir vr Relay1 Generic G P ir vr Relay3 General Elecric G P i v ir vr Relay4 SEL G P i v ir vr Relay2 Siemen G P Fig. 3. Benchmark 3: parallel line beween wo nework wih 4 relay of differen ype. Table 1 ummarize he oal number of device in each Benchmark and he number of inance which i an image of he memory uage. Benchmark 2 add hree relay o

3 Benchmark 1, o here are 4 ime more device. However, becaue macro-objec are hared, he number of inance only increae by 54. A oal of 18 macro-objec are hared in hee line proecion relay. Beween Benchmark 2 and Benchmark 3, he number of device doe no increae ignificanly, bu he relay are of differen ype and conain differen macro-objec. Therefore, he number of inance increae bu remain far from being 4 ime more han ha of Benchmark 1. TABLE 1: COMPARISON OF THE NUMBER OF DEVICES AND INSTANCES Benchmark Number of device Number of inance Benchmark Benchmark Benchmark B. Implemenaion Several proecion yem model have been implemened in EMTP [1] following he above approach. Phae and Ground diance proecion (ANSI P, G) are baed on 4 polarized zone which can be eiher forward, revere, or non-direcional and Mho, Len or Quad-haped. Calculaion mehod of memorized volage polarizaion from differen manufacurer are available. Each zone can be upervied by direcional, overcurren, load encroachmen, faul idenificaion and reacance elemen. Thi level of deail allow o udy he proecion coordinaion of power yem and o conduc deailed inveigaion udie, a hown below. Phae, ground, neural and negaive-equence overcurren elemen (ANSI 50-51P, 50-51G, 50-51N, 50_2-51_2, 46) are e uing ANSI, IEC, manufacurer or uer-defined ime curve. Polarized direcional elemen are alo available (67P, 67G, 67N, 67_2). Proecion cheme and proecion coordinaion can be accuraely imulaed and udied. Thi i paricularly imporan wih fa-dynamic yem where faul curren magniude change before relay operae or when he level of renewable energy peneraion in he yem i high. Expulion and curren-limiing fue are modelled uing meling curve, pre-arc energy and clearing energy. Thee model provide he unique capabiliy o imulae he peak-lehrough curren of curren-limiing fue. Graphical ool allow o diplay fue and relay ime-curren curve. Tranformer and conducor damage curve are alo included. Phae and ground differenial proecion (ANSI 87, 87G) include differen manufacurer algorihm wih inernal and exernal faul deecion, harmonic rerain and harmonic blocking. They are baed on phaor or inananeou value of differenial curren. Scenario of faul inide or ouide he proeced zone can be imulaed conidering CT auraion. Tranformer energizaion and overexciaion can alo be udied a demonraed below. Power Swing and Ou-Of-Sep deecion funcion (ANSI, 78) are baed on operaing characeriic deecion uing 2 or 3 zone which can be Mho, Quad or Mho wih blinder. Coningency udie are ued o accuraely deermine he ime eing of each characeriic (ee Secion IV. ) and udy he coordinaion wih diance and lo-of-field (ANSI 40) proecion, he laer being alo included in he proecion oolbox. Coninuou impedance calculaion-baed deecion i anoher algorihm available for Power Swing and Ou-of-ep funcion. Phae undervolage (ANSI 27), phae, neural, negaiveequence and compenaed overvolage (ANSI 59, 59N, 59_2), overfrequency, underfrequency and rae-of-change-offrequency (ANSI 81O, 81U an 81R) are alo available and ued for udie uch a ranien abiliy, load-hedding, and ilanding deecion. Uing he oupu of hee proecion funcion, uer can wrie heir own ripping logic funcion. Thee oupu and oher quaniie calculaed by relay are alo acceible ouide he model in order o creae proecion cheme uing communicaion beween device. III. SIMULATION EXAMPLES AND PERFORMANCE A explained in he inroducion, an imporan advanage of an EMT-ype olver over phaor-domain mehod i he much higher accuracy level. In EMTP [1], a fully ieraive oluion i available for aaining very accurae reul wih nonlinear model, uch a he magneizaion branch model ued in ranformer and meauring device. A imple 500MVA, 315/120 kv ranformer energizaion udy i preened in Fig. 4 (Benchmark 4). Tranformer energizaion caue inruh curren due o he auraion of he ranformer core. The curren magniude depend on he wiching inan of each pole of he circui breaker and he reidual fluxe of he power ranformer prior o he even. Thi inruh curren coupled wih CT auraion during he energizaion can creae a differenial curren wih a magniude reaching few per-uni, enough o caue mioperaion of percenage differenial relay [12]. To preven unexpeced rip, he harmonic conen of he differenial curren, epecially he econd harmonic, i moniored. When he raio of he econd harmonic componen in he differenial curren over he fundamenal exceed a pre-pecified hrehold, he relay deec he energizaion and block he rip. The blocking can be done individually for each phae or in common, in which cae he blocking of one phae block he oher. The challenge i o find a hrehold which i ufficienly low o cover all energizaion cenario and ufficienly high no o block in-zone faul which can aurae CT and produce harmonic in he differenial curren. Fig. 4. Benchmark 4: Tranformer energizaion Fig. 5 how he fluxe inide he ranformer of Fig. 4

4 following an energizaion. Afer approximaely 8 cycle, CT1 aurae (Fig. 6) o he harmonic in he differenial curren calculaed by Relay 87 (Fig. 7) are caued by he auraion of boh ranformer. The raio of he econd harmonic in phae- B differenial curren over he fundamenal, while he differenial curren over he rerain curren i higher han he lope eing (SLP) of he percenage differenial, i hown in Fig. 8. To deermine he hrehold eing of he 2 nd harmonic blocking funcion, bach imulaion are performed where he wiching ime of each pole of he circui breaker i varied a well a he iniial fluxe in he ranformer. ofware provide a uiable environmen for he imulaion of auraion and heir impac on differenial proecion. Anoher imporan capabiliy wih EMT-ype ofware i he accurae imulaion of renewable energy inegraion wih proecion yem. Inverer-baed device, uch a wind or olar park, behave differenly from convenional generaion during faul. Some proecion package model hem he ame way a ynchronou (generaor) machine (SM), bu hi approach can lead o ignifican error in he reach of diance relay. Fig. 5. Tranformer fluxe (ee Fig. 4). Fig. 7. Rerain curren muliplied by SLP, 2 nd harmonic and fundamenal of he differenial curren, phae-b, calculaed by Relay 87. Fig. 6. Fluxe of CT1 in Fig. 4. Two cenario are demonraed: a) 50 energizaion wihou reidual flux; b) 50 energizaion wih reidual fluxe of 0.8 pu on phae-a, and -0.4pu on phae B and C. The econd harmonic conen when he differenial curren over he rerain i higher han SLP i analyzed. For individual blocking, he eing ha o be e o ha he curren raio (differenial econd harmonic componen over he fundamenal) of each phae remain above he eing. For common blocking, he condiion i applied on he maximum of each phae raio. Thee raio, wih and wihou reidual fluxe, are diplayed in Fig. 9 and Fig. 10, repecively. A value of 100 mean he differenial curren over he rerain never reached SLP. According o hee imulaion reul, he blocking eing o avoid mioperaion mu be lower han 15.1% for individual blocking and 23.2% for common blocking. The 4 h and 5 h harmonic blocking can be udied he ame way. Oher imulaion, like in-zone faul wih CT auraion or energizaion wih fauly phae can be ued o validae he final eing [12]. IV. RELAY SETTINGS AND PROTECTION SCHEME The previou example demonraed how a ime-domain Fig. 8. Raio of he 2 nd harmonic over he fundamenal componen differenial curren of phae-b, calculaed by Relay 87 when he differenial curren over he rerain i higher han SLP. Fig. 9. Minimum raio, differenial 2nd harmonic over fundamenal in each phae, energizaion of ranformer wihou reidual fluxe. Fig. 10. Minimum of he large phae raio, differenial 2nd harmonic over fundamenal in each phae, energizaion of ranformer wih reidual fluxe. Benchmark 5 (Fig. 11) i a 3-bu yem repreening a 400 km ranmiion line beween wo 345 kv nework

5 (Nework 1 and Nework 2) wih an infeed a BUS2 from a 150 MVA wind park. To illurae he impac of wind park model on he reul of proecion udie, he wind park ha been modelled eiher a an equivalen SM-ype (SMEQ in Fig. 11) generaor or an acual deailed WP. The SMEQ model i conneced o an excier (IEEE ST1). Typical value are aumed for all parameer. The WP model i baed on aggregaion of full-converer wind generaor and include all appropriae deailed conrol yem. The Q-conrol mode i ued. I alo ue he average value modeling echnique for he elecronic converer. Complee daa for all e cae preened in hi paper i available for download upon reque. WP Wind park 150MVA SMEQ locu, comparion of SMEQ and WP model. Benchmark 6 (ee Fig. 13) focue on he difference in he dynamic behavior beween he wo modeling approache (WP and SMEQ). Four 250 MVA generaor uni are conneced o a 230 kv nework hrough wo parallel 500 kv ranmiion line of 500 km. Subaion A i conneced o he generaion uni and Subaion B i locaed a 280 km from Subaion A. A 250 MVA Type-III WP i conneced o BUS2 (model opion SMEQ and WP). A in Benchmark 5, he aggregaed WP model include all ypical conrol and ue he average value model approach for converer. Figure 14 how he diance and power wing proecion relay of hi benchmark. Wind park SMEQ WP TPP BUS2 BUS 1 BUS 2 TPP BUS 3 Nework 1 line km Nework 2 Relay 1 Faul phae A-o-B-o-GND Fig. 11. Benchmark 5: impac of wind park inegraion on diance relay reach. Fir, he SMEQ model i included and he WP i excluded. Fig. 12 how he zone 1 and 2 of he diance relay (Relay 1) in an R-X diagram and he impedance rajecory een by he relay for a faul locaed a 120% of he proeced line impedance. When he SMEQ model i ued, he inducance een by he relay few cycle afer he faul i higher han ha wih he WP model. Thi i oberved in Fig. 12 where he orange line how he rajecory wih he SMEQ model and he red line how he rajecory wih he WP model. The impedance locu keep moving due o he SM ocillaion wherea he locu of he WP model reache a new eady-ae wihin a few cycle. Wih EMTP, i i poible o perform coningency udie and, in hi cae, o vary he faul locaion uing biecion echnic o preciely deermine he reach of he relay zone. The reach of zone 2 uing he WP model i 1% wherea he one wih he SMEQ mode i 112.8%. Thi i he reach conidering only a few cycle afer he faul, ince he locu in ha cae i conanly moving. Wihou any infeed, he reach i 1.8%. Thi benchmark demonrae he imporance of accurae model for uch yem proecion analyi and eing. AVR AVR AVR ou ou in ou in in SM SM SM / / /500 BUS1 I1 I2 C Relay communicaion chanel O1 O Line 2 Subaion A Line faul I1 I2 C O1 O2 Line Subaion B Line /230/50 500/230/50 Fig. 13. Benchmark 6: impac of WP inegraion on power wing. Relay12 Relay34 Fig. 14. Benchmark 6: proecion relay in Subaion A and Subaion B. C BUS3 Nework Fig. 12. Zone 1 and 2, line diance proecion Relay1 impedance rajecory Fig. 15. R-X diagram: comparion of wor able wing, wih WP and SMEQ model. A phae-a o phae-b faul i applied a 5 km from Subaion A. When he faul occur, diance proecion in boh ubaion deec he faul and ue Permiive Over Reach raegy o clear i. The criical clearing ime of he faul i found for he WP and SMEQ cae uing bach proceing (everal imulaion) in EMTP. The criical clearing ime i he maximum ime during which a faul can be applied wihou he

6 yem loing abiliy. In hi cae, he nework loe abiliy 2.5 cycle earlier wih he WP han wih SMEQ. Wih boh model, he wor able (limi) wing een by he relay in Subaion A i diplayed in Fig. 15 on he ame graph a he Power Swing deecion zone. The difference in he impedance oberved by he relay for boh modeling approache are noiceable a oon a he faul i iniiaed. Once cleared, he fir wing doe no go a deep inide he deecion zone for boh cae, which i imporan o conider when eing up line proecion [13] and udying he limi of deecion beween able and unable wing. V. MULTI-TIME-STEP IMPLEMENTATION In Benchmark 6, boh Subaion A and B have 2 relay for a oal of 4 relay. In he cae where he wind park i repreened by a ynchronou machine (o iolae he performance iming for relay from he compuaional burden of WP conrol), he imulaion wih relay model ake 78% of he CPU ime. However, relay have a ampling frequency and heir inernal algorihm only need o be olved afer a pree ampling period. The muli-ime-ep and muli-core compuaion opion available in EMTP can be ued o ignificanly improve compuaional performance. In hi example, each relay ha a ampling frequency of 20 ample per cycle, o a ampling period of 833 µ. Sufficienly good accuracy can be achieved when inerfacing wih relay model uing a ime-ep of 400 µ while he re of he circui i imulaed a 50 µ. Wih he muli-ime-ep approach, he relay can be imulaed on differen compuer core wihou lo of accuracy. The gain in CPU uage are ubanial (3.2 ime) (ee Table 2). The imulaion have been performed on a i7 compuer wih 4 core. Table 2: Compuaional performance, Benchmark 6, imulaion uing a ingle-ime-ep and 1 core or muliple ime-ep on everal core. The imulaion period i 2. Simulaion CPU ime () Single-ime-ep and ingle-core Muli-ime-ep and muli-core VI. CONCLUSIONS Thi paper preened a new approach for advanced and accurae modeling of proecion yem in an EMT-ype imulaion environmen. A hierarchical and block-diagram baed approach ha been ued. I deliver open-archiecure model ha can be eaily analyzed and modified by uer uing a high-level graphical inerface. Due o he complexiy of reuling model (more han device in he hierarchy of ome model), i wa neceary o apply an objec-oriened approach. The defined objec are inaniaed according o relay ype and proecion opion eleced by uer. Thi approach minimize memory requiremen and reul in an overall opimized deign. I become alo uiable for imulaing large cae wih numerou relay model. The implemenaion of advanced proecion yem model in an EMT-ype package allow o achieve very accurae imulaion and become eenial epecially for udying he inegraion of power elecronic-baed device, uch a wind generaor. Thi paper alo demonraed ha ignifican compuaional gain can be achieved by imulaing he relay model uing muli-ime-ep and muli-core compuaion. I ha been hown ha he relay model can ue much larger numerical inegraion ime-ep han he proeced power yem, wihou compromiing accuracy. VII. REFERENCES [1] J. Maheredjian, S. Denneière, L. Dubé, B. Khodabakhchian and L. Gérin-Lajoie, On a new approach for he imulaion of ranien in power yem, Elecric Power Syem Reearch, vol. 77, no. 11, pp , Sep [2] J. Maheredjian, P.J. Lagace, S. Lefebvre, and A. Charrand, Superpoiion echnique for MOV-proeced erie capacior in horcircui calculaion, IEEE Tran. Power Del., vol. 10, no. 3, Jul [3] Lj.A. Kojovic, Impac of curren ranformer auraion on overcurren proecion operaion, IEEE PES Summer Meeing, [4] D. Angell, Inpu ource error concern for proecive relay, h Annual Conference for Proecive Relay Engineer. [5] I. Kocar, T. Kauffmann, U. Karaagac, M. Elamahy, H. Gra, J. Maheredjian, and E. Faranao, Impac of renewable on yem proecion: Shor-circui phaor model of converer inerfaced renewable reource and performance of ranmiion line diance proecion, EPRI echnical repor, Dec. 2015, available online a hp:// [6] I. Kocar, A. Haddadi, U. Karaagac, T. Kauffmann, H. Gra, J. Maheredjian, and E. Faranao, Impac of renewable on yem proecion: Shor-circui phaor model of renewable and impac of renewable on power wing deecion and diance proecion, EPRI echnical repor, Nov. 2016, [7] I. Kocar, A. Haddadi, H. Gra, J. Maheredjian, and E. Faranao, Advanced hor-circui modeling, analyi and proecion cheme deign for yem wih renewable TVA cae udy: Reviion of line diance proecion cheme uing deailed model for renewable and digial relay in elecromagneic ranien program, EPRI echnical repor, Jun [8] P. G. McLaren, K. Muaphi, G. Benmouyal, S. Chano, A. Girgi, C. Henville, M. Kezunovic, L. Kojovic, R. Marila, M. Meiinger, G. Michel, M. S. Sachdev, V. Skendzic, T. S. Sidhu, and D. Tziouvara Sofware model for relay, IEEE Tran. Power Del., vol. 16, no. 2, pp , Apr [9] J. B. Mooney, D. Hou, C. F. Henville, and F. P. Plumpre, Compuerbaed relay model implify relay-applicaion udie, 20 h Annual Weern Proecive Relay Conference, Spokane, Wahingon, Oc. 19-, [10] M. Kezunovic, J. Ren, and S. Lofifard, Deign, Modeling and Evaluaion of Proecive Relay for Power Syem, Springer Inernaional Publihing, [11] E. O. Schweizer and J. Rober, Diance relay elemen deign, Schweizer Engineering Laboraorie, 46 h Annual Conference For Proecive Relay Engineer, Texa, Apr [12] K. Behrend, N. Ficher, and C. Labuchagne, Conideraion for Uing Harmonic Blocking and Harmonic Rerain Technique on Tranformer Differenial Relay, 33rd Annual Weern Proecive Relay Conference, Oc [13] Power wing and Ou-Of-Sep conideraion on ranmiion line, IEEE PSRC WG D6,