Proceedngs of the 4 th Internatonal Mddle East Power Systems Conference (MEPCON, Caro Unversty, Egypt, December,, Paper ID. A Classfcaton Technque for Protecton Coordnaton Assessment of Dstrbuton Systems wth Dstrbuted Generaton A. F. Naem adelnaem@yahoo.com Y. Hegazy Yasser.hgaz@guc.edu.eg A. Y. Abdelazz almoatazabdelazz@hotmal.com M. A. Elsharkawy masharkawy@yahoo.com Department of Electrcal Power & Machnes, Faculty of Engneerng, An Shams Unversty, Caro, Egypt Abstract In ths paper a novel approach s presented to study the mpact of dstrbuted generaton on recloserfuse coordnaton. Ths approach s based on the assessment of the recloserfuse coordnaton due to the penetraton of dstrbuted generaton wth specfed locaton and capacty for a gven fault locaton. Ths assessment process acts as a classfer, based on fndng the operatng sequence of all reclosers and fuses n the path from the faulted node to the substaton. Ths sequence s compared wth a prerequred sequence obtaned from the protecton coordnaton losoy, and then provdes the dstrbuton system operator wth a result ndcatng whether the recloserfuse coordnaton stll holds or lost. Consequently, the operator can take the proper decson accordng to ths classfcaton process. One of the proper decsons that can be taken nto account n case of recloserfuse mscoordnaton s to adapt the recloser settng such that the coordnaton can be reattaned. Ths new approach has been mplemented on the IEEE 3node test feeder usng MATLABbased developed software and the obtaned results are presented and dscussed. Keywords: IEEE 3node test feeder, recloserfuse coordnaton, Dstrbuted Generaton, Dstrbuton Systems. I. INTRODUCTION Electrc Dstrbuton Systems (EDSs are usually radal n nature and are suppled from dstrbuton substatons. The man advantages of these radal systems are ther low cost and smplcty n protecton schemes. Ths protecton scheme uses smple protecton devces such as reclosers, fuses and, n some cases, overcurrent relays. The coordnaton between these protecton devces s well establshed and s, usually, done assumng system radalty [, ]. Reclosers are used n man feeders of EDSs to protect them aganst temporary selfclearng faults, whle fuses are located at the begnnng of laterals and sublaterals to protect the EDS aganst persstent faults. The man dea behnd the recloserfuse coordnaton losoy s that when a fault occurs, the recloser should operate n ts fast mode and dsconnects the crcut to gve the fault a chance to clear. If the fault stll exsts after reconnecton, then the nearest fuse to the fault should open. If, for any reason, ths fuse fals to operate, then the next upstream fuses should act n a descendng order as a backup protecton and fnally f all fuses fal to operate, then the recloser should operate agan, but n ts slow mode, as a fnal backup element. In ths way, fuses are saved as they wll operate only n case of persstent faults, whch are less than % of EDSs faults. Recently dstrbuted generaton (DG s attractng both dstrbuton utltes and electrcty consumers and can be employed for; backup generaton, power loss reducton, power qualty mprovement, envronmental concerns, and peak load servce [3]. Hgh penetraton of DG may lead to false trppng of feeders, blndng of protecton, unwanted slandng, unsynchronzed reclosng, protecton devces mscoordnaton [4]. The man concern n ths paper s to present a novel approach to: Assess the effect of DG penetraton on the recloserfuse coordnaton problem that sgnfcantly affects the relablty of typcal radal EDSs [], and Propose solutons to mscoordnaton problems, f they appear. The proposed approach s based on classfyng the coordnaton between protecton devces n the fault path to ether coordnaton holds or coordnaton lost. If the system s classfed as coordnaton lost, then a soluton s appled by changng the recloser settng, whch results n a sgnfcant reducton n the number of cases that can be usually classfed as coordnaton lost. II. SYSTEM UNDER STUDY In ths paper the IEEE 3node test feeder, whch s an actual feeder n Calforna, has been selected as a study system. The data of ths feeder are obtaned from the IEEE s Dstrbuton System Analyss Subcommttee []. Ths feeder s shown n Fg. a snglelne dagram of ths feeder s shown after beng modfed, by removng the regulator, to clearly see the effect of DG on the system. Also the nodes are renumbered for the sake of smplcty. Fnally a protecton scheme shown n Fg. s mplemented based on the method gven n []. In ths scheme one recloser s added at the begnnng of the man feeder and fuses are added at the begnnng of each lateral and sublateral. Ths system s characterzed by the followng modelng ssues: A. Lne Model Ths system contans underground and overhead three ase lnes wth dfferent spacng between ase conductors. The seres mpedance of each lne secton s represented by a 3x3 matrx as follows: zaa zab zac z lne zba zbb zbc ( z ca zcb zcc 44
3 3 Fgure. Modfed IEEE 3 Node Test Feeder B. Load Model System loads are snglease or three ase ones. Threease loads are balanced or unbalanced, and star or deltaconnected. Loads are modeled as constant power (PQ, constant mpedance (Z or constant current (I type wth modelng equatons as shown n Table I []. C. DG Model The DG can be modeled as constant PQ or PV nodes. For PQ model, t s the same as the constant power load models except that the current s njected nto the system. For the PV model, the reactve power generaton Q s calculated to mantan the specfed power and voltage for the DG and f t s out of reactve generaton lmts, then t wll be set to the lmt and the DG wll act as a PQ node. In ths research, the PV model s adopted the magntude of the postve sequence voltage s set at p.u. F Fuse R Recloser 4 3 3 3 4 Fgure. Modfed IEEE 3 Node Test Feeder wth protecton devces 4 3 33 34 3 F 4 F 3 F F F F F R 3 F F F4 F3 3 4 F 3 F F 3 F 3 F 3 3 33 34 F 3 3 F 4 F4 F3 3 Load type Constant PQ Constant Z Constant Star S V TABLE I Load Models V Z V Z ( S I ( δ θ Delta S V V Z V Z ( S ( δ θ D. Protecton Devces Model Accordng to the mplemented protecton scheme, only reclosers and fuses are used to protect the system. Fuses have nverse currenttme characterstc that s usually plotted as a loglog curve, whch s better approxmated by a second order polynomal functon. The part of nterest n ths curve approaches a straght lne and a lnear equaton can be used to reduce the calculaton task as expressed n Eqn. ( []. log( t a.log( I + b ( t : fuse operatng tme. I : fault current seen by the fuse. a & b : fuse constants to be determned as n []. Reclosers are normally equpped wth nversetme overcurrent trp devces and the general characterstcs of such devces are expressed as n Eqn. (3 [3]. A t( I TD + B (3 p M t : recloser operatng tme I : fault current seen by the recloser TD : tme dal settng M : rato of I/I pckup I pckup : relay current set pont A, B, p : constants of the selected curve characterstcs The recloser was set to have one fast trp to account for selfclearng faults and one delayed trp for fuse backup protecton by settng proper values for the TD. III. OUTLINES OF THE PROPOSED APPROACH A bref descrpton of the proposed approach s presented n the followng steps: 4
A. Load flow analyss The backward/forward sweep method s presented by Shrmohammad et al. [4] for load flow analyss and t s wdely accepted as one of the most relevant methods used n ths aspect. Ths method s capable of solvng the load flow problem for unbalanced dstrbuton systems wth DG modeled as a PV bus n two steps. The frst step s a backward sweep n whch Krchoff s current law s used to fnd load branch currents at all nodes startng from the end nodes. The second step s a forward sweep whch s started n the opposte drecton to fnd nodal voltages by applyng Ohm s law. In ths paper a load flow program based on the backward/forward sweep method s developed usng MATLAB as a platform. The developed program s able to deal wth radal unbalanced dstrbuton systems wth nbuses and wth dfferent DG penetraton levels and locatons. B. Short crcut analyss Short crcut analyss of EDSs s essental because protecton devces are selected, nstalled and coordnated based on ts results. For symmetrcal three ase EDSs, the symmetrcal component method provdes acceptable results for short crcut currents calculatons. However, for unsymmetrcal EDSs, ths method s naccurate, and other methods based on the actual ase representaton should be appled[]. One of these methods s the hybrd compensaton method [] t uses the power flow soluton as prefault condton and uses a compensaton technque to fnd the njected node currents at DG, fault and loops breakpont nodes. Then, a backwardforward sweep teraton s performed once to fnd the short crcut currents and the node voltages mmedately after fault. In ths paper, a short crcut program based on the hybrd compensaton method s developed usng the MATLAB as a platform. Ths program s desgned to handle threelne to ground, doublelne to ground, snglelne to ground, and lne to lne fault. The DG s smulated as a PV node wth constant nternal voltage at the fault nstant. C. Protecton coordnaton settng Protecton coordnaton settng for fuses and reclosers s made based on equatons ( & (3 for approprate devces, assumng that there s no DG connected ntally. For settng the reclosers, t s assumed that they are equpped wth relays havng extremely nverse characterstcs, the recloser pckup current I pckup s found as n [] usng Eqn. (4. I pck up OLF Inom (4 OLF : Overload factor depends on the protected equpment I nom : recloser current obtaned from the load flow results On the other hand, fuse settng s based on the concept that all fuses n the fault path,.e. the path from the fault locaton to the substaton, should operate slower than the recloser fast mode and faster than the recloser slow mode. Fuse settng mples the determnaton of fuse constants 'a' and 'b'. The constant 'a' represents the slope of the straght lne I t loglog plot and s fxed at a specfed value for all fuses n the system. Ths condton s practcally acceptable because all fuses n the system should be of the same type. The constant 'b' s calculated usng the value of 'a' and the coordnates of one operatng pont of the fuse (fuse fault current and fuse operatng tme. Fuse fault current s obtaned from short crcut results whle fuse operatng tme s obtaned by dvdng the tme range of the recloser (.e. the dfference between the operatng tmes of the slow and fast operatng modes by the number of fuses n the fault path, usng Eqn ( whch s developed by the authors. ( trec slow trec fast t t + t fuse n t recslow t recfast fuse rec fast ( n + :operatng tme for the th fuse n the fault path for the fuse nearest to the faulted node. :total number of fuses n the fault path. :the recloser slow mode operatng tme. :the recloser fast mode operatng tme. D. Protecton coordnaton assessment After dong protecton coordnaton between fuses and reclosers n the system wthout DG, t s requred to assess ths coordnaton after the penetraton of DG, to avod takng dscplnary actons lke dsconnecton of DG each tme a fault takes place even for the cases the coordnaton s not lost. In the assessment process, t s requred for a fault at a certan node to check the operatng sequence of the protecton devces n the fault path after DG connecton, by fndng the operatng tmes of these protecton devces from Eqns. ( and (3 after substtutng nto them the calculated short crcut currents. Havng found the operatng tmes of protecton devces, the operatng sequence s determned and then compared wth the prerequred sequence. If a close match between the obtaned sequence and the requred sequence occurs, then the coordnaton holds and no further acton s requred, otherwse the coordnaton s lost and the DS operator should take a proper decson to avod the consequences of mscoordnaton between protecton devces. One soluton for the mscoordnaton problem s proposed n ths paper. It s based on changng the characterstcs of the recloser by changng the TD parameter n Eqn. (3. Ths acton s practcally acceptable nowadays thanks to the avalablty of mcroprocessorbased reclosers n the market. Mcroprocessors can be easly used to adjust recloser currenttme characterstcs accordng to system protecton requrements. To evaluate the effectveness of ths soluton on the coordnaton problem, dfferent cases are studed by changng DG penetraton level and locaton for a fault at a specfed node. The number of cases the coordnaton holds wth respect to the total number of studed cases s montored for dfferent values of the TD parameter. For ths purpose a program has been developed usng MATLAB to use t as a classfer to assess the protecton coordnaton and evaluate the effectveness of the proposed soluton. Fg. 3 shows a flow chart for ths program. 4
IV. RESULTS AND DISCUSSIONS In ths secton some results related to each step n the proposed approach are presented. Read system data and specfy the fault locaton Select a DG locaton Select a DG penetraton level Run the load flow and short crcut programs Apply the classfcaton process All DG penetraton levels No are consdered Yes No All DG locatons are consdered Yes Plot the classfed cases for dfferent DG locaton and level Fgure 3. Flow chart for FuseRecloser coordnaton assessment A. Load flow results The developed load flow program s appled to the IEEE 3 node test feeder and some of the results are presented. Table II shows the magntude of the branch currents n the base case (wthout DG penetraton. These results are very close to the load flow results publshed n [] for the same system, whch assures the valdty of the developed program. Fg. 4 shows the lne voltage V ab for dfferent cases of the modfed IEEE 3node test feeder. In case the test feeder has no DG penetraton. In cases ( and (3 one DG s connected to the system at node wth % and % penetraton level, respectvely. From these two cases, t s clearly shown that the presence of DG n the system mproves the voltage profle for the whole system and especally at the node at whch the DG s connected. In case (4, two DGs wth % penetraton level are connected at nodes and. From ths case, t s clearly seen that dvdng the same penetraton level on more than one node n the system has a better effect on the voltage profle compared wth concentratng t at one node. B. Short crcut results Some of the results of applyng the developed short crcut program on the IEEE 3node feeder are presented here. Fg. shows the magntude of fault currents at the faulted node for ase (A for two cases. In case (, the system has no DG and a three ase fault s appled at dfferent locatons. In case (, one DG s connected at node wth % penetraton level and agan a three ase fault s appled at the same locatons as n case (. It s clearly shown from Fg. that when the fault takes place at nodes 4 or whch are very TABLE II Branch Currents of the IEEE 3node feeder wthout DG penetraton Branch From Phase currents, A No. Nodeto Node Ia Ib Ic 3 4 3 4 3 3 4 3 3 3 33 3 3 3 3 4 4 4 3 34 4 4 3 3 33 34 3 3 3 3 333 3334 343 343 33.4.43. 4..3 4.4 4.4 3.34....4. 43.4 3...4 4.4 33.3 33.3. 4..4.3.3.. 3.3....43.4.4.4 4.4.4 3.3.3.3.3..3.3 3.3 3.343 3.343..4.3.3.....44.44 3. 3.4 34....34. 44.3.4 3.3.3.3 4.4 34...4 34.333..43.43.4...4..33.33.43.44.3 3.3 3.3 3.3.. close to node at whch the DG s connected, an apprecable ncrease n the fault current, compared wth case (, s remarked. In general, t can be concluded from Fg. that the severty of the effect of DG penetraton on fault currents, and consequently on the protecton system, becomes less as the electrcal dstance between fault locaton and the DG locaton ncreases. Voltage profle of Vab (p.u....... wthout DG DG at node wth % penetraton DG at node wth % penetraton two DGs at node & wth %penetraton for each DG 3 3 Bus Number Fgure 4. Lne Voltage (V ab for dfferent cases of the modfed IEEE 3node test feeder 4
Magntude offault Current for Phase A (Amp.... 3 x 4 o Case ( Case ( 4 4 4 3 3 34 3 3 Faulted Node Fgure. Magntude of fault currents of ase A for dfferent fault locatons, case ( wthout DG, case ( wth DG at node and %penetraton level. Table III shows the magntudes of the nonzero branch fault currents for ase (A wthout the presence of DG when a three ase fault s appled at dfferent nodes n the system. From the results, t s concluded that the fault currents decreases when the faulted node moves away from the substaton as t s clear when comparng the fault currents at nodes, 4 and. C. Protecton settng results Reclosers havng standard extremely nverse characterstcs are used. The parameters A, B, and p n Eqn. (3 are taken equal, respectvely, to.,., and [3]. The recloser nomnal current I nom s 33.4 A as obtaned from Table II and the OLF parameter n Eqn. (4 s taken equal to.. The parameter TD s arbtrary set to be. and., respectvely, for the slow and fast trppng modes of the recloser as n []. For fuse settng, the fuse constant 'a' n Eqn. (, s chosen equal to. and wll be fxed for all fuses n the system. TABLE III Branch fault Currents of the IEEE 3node feeder wthout DG penetraton Branch Faulted node No. 4 3 4 4 3 4 4 4 3 4 3 4 44 44 3 4 4 3 34 34 33 4 33 33 3 3 3 3 33 4 3 44 44 3 3 3 3 3 4 4 4 4 44 44 44 4 For calculatng the constant 'b', a three ase fault s appled at each end node successvely and the obtaned short crcut currents are substtuted nto Eqn. ( to obtan one operatng pont for the fuse. Consequently the constant 'b' s found by rearrangng Eqn. (. For example, to fnd the constant 'b' of fuses F, F4, and F, a three ase fault s appled at node and the short crcut currents n the branches contanng the recloser and these fuses are found by runnng the short crcut program as n Table III. From the recloser current, t recslow and t recfast are found from Eqn. (3 equal to.s and.s, respectvely. Usng Eqn. (, the operatng tmes for F, F4 and F are found to be.s,.s and.4s, respectvely. These operatng tmes wth the correspondng fuse fault currents are used n Eqn. ( after rearrangement to fnd the constant 'b' for each fuse. Table IV summarzes the values of the constant b for all fuses n the system. Usng these results the fuse characterstcs can be constructed. Fg. shows the operatng curves for the recloser and fuses F, F4 and F. D. Protecton coordnaton assessment results The coordnaton assessment process s appled to the IEEE 3node test feeder usng the developed program, one DG s connected n turn to all system nodes except the faulted node and the substaton node,.e. 3 dfferent locatons. The DG penetraton level s also changed from kw to kw n steps of kw resultng n dfferent penetraton levels wth total dfferent possble cases equal to 3x3. Fg. shows the results of the classfcaton process when a fault s appled at node whle changng the DG penetraton level and locaton. TABLE IV Fuse constant Fuse number Fuse Constant 'b' Fuse number Fuse Constant 'b' Tme(seconds 3 4.33.4...3.3.4.334..34 3 4 Reclosure Slow Reclosure Fast F F4 F.4.3..4....3..3 3 4 current(ampers Fgure. Operatng curves for the recloser and fuses F, F4 and F. 4
The whte crcles represent the cases coordnaton holds and the black crcles represent the cases coordnaton s lost. The number of cases coordnaton holds as a percentage from the total number of cases s 4/3 4.%. As a result, applyng the classfcaton process dscrmnates between the cases an acton s requred aganst the DG penetraton at fault condtons and the cases no need for an acton s requred, and, consequently, the system relablty wll be mproved. To decrease the number of cases coordnaton s lost, a soluton based on changng the recloser characterstcs s proposed and appled. Accordng to ths soluton the TD parameter of the recloser fast operaton s changed from ts ntal value at. to a value of. n steps of.. Fg. and Fg. show the new classfcaton pattern for TD equals.3 &. respectvely. The number of cases coordnaton holds as a percentage, for these two values of TD s 3/3 % and /3.%, respectvely, whch shows the effectveness of the proposed soluton to mprove the protecton coordnaton behavor. DG penetraton level(kw fault at node( Tme Dalng of fast Rec operaton. 4 4 3 3 4 4 4 3 3 34 3 3 DG locaton Fgure. Classfcaton pattern for a fault at node wth TD equals. for recloser fast operaton. DG penetraton level(kw fault at node( Tme Dalng of fast Rec operaton.3 4 4 3 3 4 4 4 3 3 34 3 3 DG locaton Fgure. Classfcaton pattern for a fault at node wth TD equals.3 for recloser fast operaton. DG penetraton level(kw fault at node( Tme Dalng of fast Rec operaton. 4 4 3 3 4 4 4 3 3 34 3 3 DG locaton Fgure. Classfcaton pattern for a fault at node wth TD equals. for recloser fast operaton. V. CONCLUSION A protecton coordnaton assessment process s proposed and appled to the IEEE 3node test feeder to evaluate the effect of the DG penetraton on the protecton devces coordnaton by classfyng the system to ether coordnaton holds or coordnaton lost. Dfferent DG penetraton levels and locatons are studed. To decrease the number of cases coordnaton s lost, a soluton s proposed based on changng the recloser characterstcs by changng the TD parameter for the recloser fast mode. Ths soluton s evaluated by reassessng the system for dfferent DG penetraton levels and locatons whch shows that the number of cases coordnaton s lost s sgnfcantly reduced. Ths study s a mle stone for developng an effcent and smple soluton for the recloserfuse coordnaton problem due to DG penetraton. REFERENCES [] J. M. Gers and E. J. Holmes, Protecton of Electrcty Dstrbuton Networks, London, U.K., IEE, 4. [] P. M. Anderson, Power System Protecton, New York: IEEE Press,, pp. 4, 4. [3] W. ElKhattam and M. M. A. Salama, "Dstrbuted Generaton Technologes, Defntons and Benefts", Electrc Power Systems Research, Vol., Issue, Oct. 4, pp.. [4] P. P. Barker, R. W. de Mello, "Determnng the Impact of Dstrbuted Generaton on Power Systems:part IRadal Dstrbuton Systems," IEEE Transactons on Power Delvery, Vol., Aprl, pp. 443. [] S. M. Brahma and A. A. Grgs, "Impact of Dstrbuted Generaton on Fuse and Relay Coordnaton: Analyss and Remedes," n Proc. Int. Assoc. Sc. Techno. Develop, Clearwater, FL,, pp. 343. [] K. Kauhanem, L. Kumpulanen, "Impact of Dstrbuted Generaton on the Protecton of Dstrbuton Networks," Developments n Power System Protecton, th IEE Internatonal Conference. Aprl 4, Vol.. [] S. Chatusaney, and A. Yokoyama, Conflct Between Backup Generaton and Recloserfuse Mscoordnaton n Relablty Aspect from Dstrbuted Generaton, IASTED Internatonal Conf. on Energy and Power Systems, March, Chang Ma, Thaland, pp.. [] Radal Dstrbuton Test Feeders.[Onlne].Avalable: http://www.ewh.eee.org/soc/pes/dsacom/testfeeders.html. [] J.A.slva, H.B.Funmlayo, K.L. BulterPurry, Impact of Dstrbuted Generaton on the IEEE 34 Node Radal Test Feeder wth Overcurrent Protecton, Power Symposum 3th North Amercan,, pp. 4. [] S. Khushalan and N. N. Schulz, Unbalanced Dstrbuton Power Flow wth Dstrbuted Generaton, IEEE Proceedng Transmsson and Dstrbuton Conf., Dallas, TX, May. [] S. Javadan and M. Haghfam, "Mantanng the RecloserFuse Coordnaton n Dstrbuton Systems n Presence of DG by Determnng DG s Sze", The th IET Internatonal Conf. on Developments n Power System Protecton, March, Glasgow UK, pp. 4. [] S. Chatusaney and A. Yokoyama, Preventon of Relablty Degradaton from Recloser Fuse Mscoordnaton Due To Dstrbuted Generaton, IEEE Transactons on Power Delvery, Vol. 3, No. 4, October, pp. 4 4. [3] IEEE Standard InverseTme Characterstc Equatons for Overcurrent Relays, IEEE Standard C3.. [4] C. S. Cheng and D. Shrmohammad, A ThreePhase Power Flow Method for RealTme Dstrbuton System Analyss, IEEE Transactons on Power System, Vol., No., May, pp.. [] P.M. Anderson: Analyss of Faulted Power Systems, IEEE Press Power Systems Engneerng Seres, New York,, pp. 3. [] X. Zhang, F. Soud, D. Shrmohammad, C.Cheng: A Dstrbuton Short Crcut Analyss Approach usng Hybrd Compensaton Method, IEEE Transactons on Power System,, Vol., No.4, pp 3 [] S. Khushalan, Solank and N. N. Schulz, Development of ThreePhase Unbalanced Power Flow Usng PV and PQ Models for Dstrbuted Generaton and Study of the Impact of DG Models, IEEE Transactons on Power Systems, Vol., No. 3 August, pp.. 4