Placement of Fault Current Limiters in Power Systems by HFLS Sorting and HIGA Optimization Approach

Transcription

1 Placement of Fault Current Lmters n Power Systems by HFLS Sortng and HIGA Optmzaton Approach Hong-Tzer Yang, Wen-Jun Tang, Conne Wang, Potr Lubck and Steve Wang Abstract-- As the number of new power plants and the scale of transmsson systems sustanably grow, the probablty of exceedng the short crcut ratng of crcut breakers (CBs) ncreases, a fact whch leads the system's securty and stablty to a bg ssue. Fault Current Lmter (FCL), as one of the solutons to current surge attenuaton, attracts a lot of attenton from utltes. To make the best use of FCL, the placements must be optmally determned along wth the most sutable FCL parameters settngs. Ths paper proposes a method combnng Herarchcal Fuzzy Logc System (HFLS) method for pre-sortng the feasble solutons and Hashng-Integrated Generc Algorthm (HIGA) as an optmzaton tool to fnd the best soluton n the reduced search space. To verfy the proposed approach as an effectve means for placement of the FCLs, the proposed method s mplemented through Matlab and DgSILENT and tested n a practcal power system. The numercal results show that the proposed method may acheve a better soluton n less tme wth fewer placements of FCLs, thus reducng the cost, whle mantanng comparable system securty. Keywords: Fault Current Lmter (FCL), Short-crcut Current, Voltage Stablty, Search Space Screenng, Herarchcal Fuzzy Logc Control, Generc Algorthm, FCL Placement Optmzaton W I. INTRODUCTION th the constant ncrease n power demand, power system has been facng the challenges of securty and relablty [][]. Dstrbuted Energy Resources (DERs) and Independent Power plants (IPPs) are appled to solve energy shortages. As a result, the occurrence rate and magntude of fault current greatly ncrease [3]. The fault current may thus exceed the CB nterrupton capacty, leadng to fal of CB operaton and even cascadng ncdents, lke blackouts [4]. Accordngly, many researches have been focused on ths problem. The most drect soluton s to upgrade all devces n the system, whch n most cases s not economcally feasble. Alternatvely, power network reconfguratons have been Ths work was supported n part by AMAT and MOST E-006-0, Tawan Hong-Tzer Yang s wth the Research Center for Energy Technology and Strategy, Department of Electrcal Engneerng, Natonal Cheng Kung Unversty, Tanan, Tawan (e-mal: htyang@mal.ncku.edu.tw). Wen-Jun Tang s wth the Research Center for Energy Technology and Strategy, Department of Electrcal Engneerng, Natonal Cheng Kung Unversty, Tanan, Tawan (e-mal: monkatang354@qq.com). Conne Wang s wth Appled Materals, Inc., Santa Clara, CA, USA (e-mal: Conne_Wang@amat.com) Potr Lubck s wth Appled Materals, Inc., Gloucester, MA, USA (e-mal: Potr_Lubck@amat.com) Steve Wang s wth Appled Materals, Inc., Tanan, Tawan (e-mal: Steve_Wang@amat.com) Paper submtted to the Internatonal Conference on Power Systems Transents (IPST05) n Cavtat, Croata June 5-8, 05 suggested to elmnate overload or fault cases by changng the topology through sectonalzng swtches [5][6]. Although no addtonal devces are needed wth ths method [7], the control n real-tme s obvously problematc [8]. The current-lmtng reactor applcaton s another possble soluton, whch prevents CBs from over capacty current; however, t smultaneously consumes energy n normal operaton. Wth nearly zero power loss and smaller voltage drop durng normal state, as well as a short delay tme to ntroduce hgh reactance nto the crcut, when a fault occurs [9], FCLs are one of the more attractve solutons to current surge attenuaton. In the early stages, power electroncs structure was wdely used n FCLs. Wth the progress of materal technology, superconductor has been ntegrated n FCLs. Superconductor has the electrcal specfcaton that the resstance value s nearly zero, when operatng under a fxed scope of temperature and crtcal current densty; on the contrary, the resstance ncreases mmedately, as subject to a fault [0][]. In addton, the applcaton of FCL does not change the topology of an exstng network [] or mpact other power devces [3]. Taken together, these characterstcs make FCL an deal canddate for solvng exstng hgh faultcurrent problems [4]. The placements of FCLs have been dscussed n several studes. Most of studes have focused on specfc locatons, such as double-bus parallel, pont of common couplng (PCC), transformer s neutral lne or a smart grd n small areas [5]- [8], wth a few consderng overall system plannng [9][0]. The objectve functon to mnmze the number of FCLs nstalled and ther shunt reactors values has been employed wth respect to the whole-system FCL placement optmzaton []. However, advantages of FCL are not fully taken for the objectve functon employed. The plannng of FCL nstallaton s always a concern n large power systems. However, optmzaton of the FCLs placement becomes very dffcult wth so many canddates n the large power system. The computatons needed for the power flow and fault current estmaton n the systems consume much tme. Besdes, some locatons and areas to nstall FCLs may have much less mpact on the fault current than the other. The computng tme of these canddates wth less mpact s thus meanngless, and so a sortng tool s, therefore, requred. Teng and Lu recommend a senstvty factor for sortng canddates []. The senstveness of each canddate locaton s consdered by the fault current reducton, when ts fault mpedance value s changed. They suggest that the more senstve a canddate locaton s, the hgher the nstallaton potental. Accordngly, ths method saves tme of the optmzaton algorthm by reducng the search space;

2 nevertheless, t stll requres consderable tme va the tralerror assessment. The fuzzy logc system (FLS) has been wdely used to solve the mult-objectve problems []. When managng more than two nput control varables, a collecton of lowdmensonal systems s needed. Instead of usng a hghdmensonal system that may cause fuzzy rule exploson, a herarchcal fuzzy logc system (HFLS) [3][4] employs the low-dmensonal control systems based on dfferent rules whose number s the same as the herarchy number to ncrease the accuracy of results. HFLS s thus a good alternatve deal wth the multple-varable systems. Genetc Algorthm (GA), Partcle Swarm Optmzaton (PSO), Colony System (ACS) and Artfcal Immune System (AIS) are classfed as Populaton-Based Intellgent Search (PIS) algorthms [5]. PIS has been used by numerous studes as an optmzaton algorthm, and has been shown to have promse n many applcatons. Among such studes, GA has been employed n the research focusng on optmal placement of dfferent system devces [6], such as FCLs [], fault ndcators [7], and capactors [8][8][9], etc. In ths paper, the GA-based approach s used wth the ftness functon consderng more aspects, ncludng the FCL functons of voltage varaton restranng. Based on the proposed new measurng senstvty factors, HFLS s frstly presented n sortng the potental nstallaton locatons of FCL. Consdered are not only the over-current buses, but also the source buses of the fault current. The optmzaton further employs the HIGA [3] method to elmnate the computng tme of any repettve solutons under evaluaton. To verfy the proposed approach for the optmzaton of FCL placement, the method s analyzed and tested through Matlab and DgSILENT n a practcal power system. II. CIRCUIT BREAKER INTERRUPTION CAPACITY Before dscussng FCL placement optmzaton, defnton of fault current and CB nterrupton capacty s clarfed frst. A. Characterstcs of Fault Current When the short crcut current s calculated, the power system s assumed to be a Thevenn equvalent crcut (Fg. ), consstng of an AC source and seres R-L. The expresson for a KVL crcut equaton s derved as the functon below. Ld R( t) dt V sn( t ) t 0 () The soluton becomes where ( t) ac ( t) ( t) dc t V sn( t ) sn( ) e T () V ac sn( t ) (3) dc t T V sn( ) e (4) R ( L) R X (5) L X tan tan R R (6) L X X T R R fr (7) The total fault current, also called the asymmetrcal fault current, s shown n Fg., and s dvded nto ac n (3) and dc components n (4). The ac fault current named as the symmetrcal or steady-state fault current s snusodal, whle the dc element decays by tme constant T, as gven n (7). From the expressons presented above, each component s affected by the factors α and orsd mponent s sons presented e accordng to the dverse fault nstants (as llustrated n Table I). Fg.. Current n a seres R-L crcut wth ac voltage source Fault Occurrence 0 90 Fg.. Total Fault Current TABLE I DIFFERENT FAULT CURRENT VALUES Short Crcut Value Vm snt Vm ( e I ac Vm I ac R t L cost) 0 ( t) dt R t L ( e 0or 90 ( 0) ( 90) )

3 B. Crcut Breaker Interrupton Capacty Accordng to ANSI IEEE Std. C [3], the standard ratng structure for AC hgh-voltage crcut breakers rated on a symmetrcal current bass s taken as a reference. The requred symmetrcal and asymmetrcal nterruptng capabltes are descrbed. The abltes are related wth onehalf cycle relay tme, or sometmes a pesble trppng delay tme s employed nstead. The hghest value of the symmetrcal component of the poly-phase or phase-to-phase short crcut currents n amperes determnes the rated short-crcut current of a CB. In lne wth ths value, the CB should normally close, latch, and nterrupt. The relatonshp between rated short-crcut current and other necessary capabltes can be found n detal n ANSI IEEE Std. C The hghest fault current value of the symmetrcal component n amperes s taken nto consderaton at the specfc operatng voltage. And usually, ths capablty s equal to K tmes the rated short-crcut current. The K value of a modern CB s defned as.0, whle older CBs requre K to be more than.0. The asymmetrcal nterruptng capablty s desgned based on the value of the total short-crcut current amperes at the separaton tme of the prmary arcng contact. The partng tme s defned as one-half of a cycle or some other operatng tme partcular to the breaker, lke.0,.5,.5 or 3.5 cycles for rated nterruptng tme of, 3, 5, or 8 cycles, respectvely. The rated nterruptng current standard used by Tawan Power Company (TPC), n Tawan, s the ANSI IEEE Std. C , whch uses three-phase to ground faults to smulate the maxmum fault current. The functons to determne the asymmetrc fault current are used to estmate whether the CB nterrupton capactes are suffcent (as shown n (8) and (9)) : K D S t ( e ) S as ( where K Multplyng Factor D DC Effect Multpler S Asymmetrc Capabltes (././.3 for CBs wth 5/3/ cycles) t Partng Tme as Asymmetrcal Fault Current Symmetrcal Fault Current ds III. 377 t e X / R ) S (8) K sy (9) THE PROPOSED FCL PLACEMENT OPTIMIATION METHOD The proposed optmzaton method manly comprses two parts, the HFLS for pre-sortng the solutons and HIGA for subsequent optmzaton n the reduced search space. The soluton flowchart of the proposed method s presented n Fg. 4. As mentoned, the system uses DgSILENT and Matlab as two smulaton tools, wth DgSILENT for power flow analyss as well as fault current calculaton (yellow squares), and Matlab for HFLS and HIGA (red squares). The connecton between these two programs s based on the.csv fle. From the flow chart, we can see that not only HFLS but also HIGA are all on account of the data obtaned from DgSILENT, whch nclude the values of current, voltage, shunt reactors, and generators. A. HFLS Sortng Method As shown n Fg. 5, the HFLS sortng process s dvded nto two levels. Mendel s Type- fuzzy system s used here wth the addton of a new set of membershp functons to ensure flexblty. The frst two nputs go through the frstlevel fuzzy system; then, the output of the frst-level s operated n the hgher-level fuzzy system wth the secondstage nput. Ether the components or the evaluatng varables are ndependent from each other. Amng to completely apprase the senstvty of each bus, the proposed method employs the herarches representng the probablty of one bus beng the fault current resource, the bus wth over-capacty CB or both roles. The peak-current data for the bus are then used to defne the senstvty ndces of Max Contrbuton Current (MCC) when a fault occurs on the connected bus, Generator Connected Number (GCN) as the number of connected generators to the bus, and Max Fault Current (MFC) as the fault current calculated based on (9). Fg. 4. Placement Optmzaton Method Flowchart Fg. 5. HFLS Sortng Structure

4 MCC and GCN are the evaluaton ndces of the frst stage (Input ). Through the membershp functons and rule base, the output s acheved mathematcally as the nputs of the second stage along wth MFC (as shown n Fg. 5). Illustrated n Fg. 6 are the membershp functons of HFLS sortng. The x-axs range of MCC and MFC s changeable and refers to the vrtual fault current of the under-plannng systems. The rule base used n the paper s gven n TABLE II and Fg. 7. To defuzzfy the data, Center of Maxmum (CoM) method s employed as shown n (0). x u xu xnun x fnal (0) u u u where x Typcal numercal value for the scaled membershp functon n n u The degree of membershp at scaled membershp functon n n n G C N TABLE II HFLS RULE BASE (a) FIRST STAGE MCC LC MC HC SHC NG Fourth Thrd Second Frst OG Thrd Second Second Frst TG Second Second Frst Frst (b) O U P U T SECOND STAGE MFC S M L S S S L M S M L L M L L Fg. 8. Hash Table Create an empty Hash Table wth the sze of (GA s teraton no. * popsze)*varable no. GA generates the st populaton and cost. All are put nto Hash Table Fg. 6. Membershp Functons of HFLS Sortng If the GA teraton tme < max, then generate the new populaton. yes Compare the new populaton wth the Hash Table Value. The same value? no Pck the cost value from extng Hash Table Calculate the cost value agan by DgSILENT Add the new chromosome and cost value to Hash table Fg. 7. Rule Base Surface of HFLS Sortng no GA teraton s fnshed? yes Optmzed Result Fg. 9. HIGA Flowchart

5 Fg. 0. Chromosome Structure Senstvty ndces n the frst-stage have an unbalanced evaluaton mechansm, emphaszng that MCC should have more weght. For example, f a bus data has a Super Hgh Current (SHC)-level MCC wth no generator connected (NG), the output s the value more lkely to be approached frst. Then, n the second-stage, MFC acts n the leadng role. The hgher the value the MFC has, the more t nfluences the fnal output. Once the fnal output s determned, the senstvty sortng of buses s completed. Wth above results, the frst m buses are chosen for the followng GA optmzaton step. B. HIGA for Optmzaton The tradtonal structure of the GA has the dsadvantage that the same chromosome/soluton may be generated more than once. A repettve ftness value s thus calculated agan; however, t costs computaton tme and decreases the effcency, especally when appled n a large system or dealng wth large quanttes of data. To elmnate ths problem, ntegratng hashng technque wth the GA s proposed to solve the problem. Apart from the normal GA, HIGA employs a hash table to save the chromosome and ts ftness value [3]. As shown n Fg. 8, the frst chromosome s generated and saved n the hash table wth ts ftness value. Each chromosome s compared wth the saved components n the hash table. Therefore, when the same chromosome appears a second tme, the ftness value s obtaned mmedately, as marked n red n the fgure. The HIGA flowchart s shown n Fg. 9, and the detals of each step are lsted as follows: A hash table s ntaled wth a sze whose length s the same as the teraton number multpled by populaton sze and wdth as a varable number plus one. The last row contans the ftness value of each populaton column. The frst GA populaton s generated, and the cost s then calculated. Populaton and results are nput to the hash table. The exstng populaton s lsted by the ftness value n descendng order. The next generaton s created based on the crossover or mutaton mechansm and rate. 3 The new generaton s compared wth the exstng chromosome n the hash table to check whether t s the same as the prevous one. 3. If the same, the ftness value s adopted drectly wthout any further calculaton. 3. If dfferent, the ftness value s calculated and then nput to the hash table wth ts chromosome. 4 Stoppng rule s checked. 4. If max. teraton number s exceeded, the algorthm ends, 4. If t s not exceeded, go back to Step 3. Each varable has a fxed length bnary number n the chromosome (as shown n Fg. 0). To smplfy the optmzaton, every length refers to the FCL shunt reactor s value at the bus. However, f the value s zero, t denotes that there s no FCL nstalled. The objectve functon s the key to an optmzaton algorthm. The proposed method s amed at optmzng overcurrent-capacty systems wth economc solutons. The problem s formulated as a mult-objectve functon n () wth constrants that restran the fault current wthn the CB nterrupton capactes. A penalty factor here s used for the constrant volaton. Subject to Where, FCL mn J N B, FCL mn max, FCL, FCL, FCL I SC j I SC,max j N V () j B Shunt reactor s value of th FCL N The number of FCLs V Voltage varaton at fault state, mpact factor IV. M CASE STUDIES N FCL To verfy the proposed approach for optmzng FCL placement, the method s mplemented through Matlab and DgSILENT. The optmzaton method s programmed n Matlab wth an nterface developed to allow the power flow and fault analyss to be conducted n DgSILENT. The mplemented approach s tested on the practcal 83-bus power system of a large manufacturng factory n Tawan to evaluate the qualty of the acheved soluton and the computatonal effcency of the proposed approach (as shown n Fg. ). The results from the proposed method are compared wth those by usng a Reference Method whch employs MFC as space reducton ndex and HIGA as the followng optmzaton algorthm. Based on the comparson, the advantages of the three-factor HFLS sortng can be demonstrated. The teraton tme, populaton sze, mutaton rate and selecton rate of HIGA are gven to be 00, 6, 0.35 and 0.5, respectvely. There are 83 buses, 48 loads and 4 generators n ths factory system, ncludng three voltage levels (6kV, 33kV and.5kv). Due to the over-load problem and expanson of the factory, the fault current exceeds the extng CB nterrupton capactes (50kA for 6kV and 40kA for 33/.5 kv). The frst 4 hgh fault currents can be found n Table III wth ther names before the FCL placement. To lmt the overcapacty fault currents and leave a margn to ensure safety, the current lmtaton n constrant s gven as 40kA. The optmzaton algorthm also employs the sortng results obtaned from the sole ndex of MFC to perform the comparson.

6 Fg.. Fault Current Comparson Fg.. Test System Topology TABLE III FIRST 4 HIGH FAULT CURRENTS BEFORE FCL PLACEMENT Fault Current (ka) Fault Current (ka) Bus Bus 3 3. Bus Bus Bus Bus Bus Bus Bus Bus 7.98 Bus Bus Bus 3. Bus TABLE IV CANDIDATES OBTAINED FROM THE OPTIMIATION METHODS MFC wth HIGA Proposed Method (Reference) Method Shunt Reactance (Ω) Shunt Reactance (Ω) Bus 8 0 Bus 8.49 Bus 7 0 Bus Bus Bus 7.49 Bus 5 0 Bus Bus 3 0 Bus 5 0 TABLE V FIRST 0 HIGH FAULT CURRENTS MFC wth HIGA Proposed Method (Reference) Method Fault Current (ka) Fault Current (ka) Bus Bus Bus Bus Bus Bus Bus 3.60 Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus Bus Table IV shows the 5 canddates obtaned from the proposed and the Reference Method, respectvely. As can be seen, 5 FCLs are suggested by the Reference Method to be nstalled versus FCLs done by the proposed method. Based on the canddates, the proposed algorthm optmzes the FCL placement and shunt reactors wth zero meanng no FCL connected to that bus n TABLE IV. Whle the non-zero value ndcates the FCL optmzaton locaton as well as ts shunt reactance value. To verfy the advantages of proposed method, the fault currents are presented and compared by the Reference Method n Table V wth the frst 0 hgh fault currents after the optmzaton of FCL placement. Bounded by the nterrupton capacty of 40 ka, Fg. further shows the fault currents comparsons of all the buses for the systems wthout FCL (ntal), wth the FCLs suggested by the proposed method, as well as wth the FCLs by the Reference Method. It reveals n Fg. that the proposed method does restrct the fault currents to the lmtaton wth only two FCLs nstalled. In contrast, the referenced MFC ntegrated wth HIGA method restrans fault currents to the lmtaton usng fve FCLs, whch s obvously much less economcal. The effectveness of the hash table s also demonstrated by the optmzaton wth HIGA and normal GA referrng to the same HFLS sortng canddates. It s noted that the computng tme s decreased from 59 mns 7 secs to 80 mns 9 secs wth the teraton number reduced from,600 to,459. V. CONCLUSIONS Ths paper has proposed an effectve method based on HFLS sortng and HIGA optmzaton to obtan the best soluton for the optmzaton problem of FCL placements and ts shunt reactance values, especally for large power systems. HFLS renders the search for optmal canddates operatonal and effectve, the results have been well verfed n the case studes. The hash table has also been embedded n the GA to accelerate the optmzaton process wth mproved computng tme demonstrated. The advantages of the proposed method are beleved to be more apparent n applcaton of the optmzaton approach n a large power system. Moreover, voltage varaton s taken nto account n the ftness functon by usng the sgnfcant functon of FCL. Overall, the system securty and relablty can thus be enhanced through the proposed method for optmal placement wth less number of FCLs nstalled.

7 VI. REFERENCES [] P. M. Duggan, "Utlty perspectve on fault current lmters and expected synerges from ntegratng fault current lmters wth superconductng cables," Power and Energy Socety General Meetng - Converson and Delvery of Electrcal Energy n the st Century, 008 IEEE, pp. -, July 008. [] J. Jager, "Interacton between fault current lmters and protecton - a glance at the fnal report of CIGRE WG-A3.6," Power and Energy Socety General Meetng - Converson and Delvery of Electrcal Energy n the st Century, 008 IEEE, pp. -3, July 008. [3] W. Lee, J. Sm, K. B. Park, and I. S. Oh, Practcal applcaton ssues of superconductng fault current lmters for electrc power systems, IEEE Transactons on Appled Superconductvty, vol. 8, no., pp , 008. [4] M. Noe, M. Steurer, S. Eckroad, and R. Adapa, "Progress on the R&D of fault current lmters for utlty applcatons," Power and Energy Socety General Meetng - Converson and Delvery of Electrcal Energy n the st Century, 008 IEEE, pp. -4, July 008. [5] M. E. Baran and F. F. Wu, Network reconfguraton n dstrbuton systems for loss reducton and load balancng, IEEE Transactons on Power Delvery, vol. 4, no., pp , 989. [6] H. Falagh, M. Haghfam, and C. Sngh, Ant colony optmzaton-based method for placement of sectonalzng swtches n dstrbuton networks usng a fuzzy multobjectve approach, IEEE Transactons on Power Delvery, vol. 4, no., pp , 009. [7] R. A. Jabr, R. Sngh, and B. C. Pal, Mnmum loss network reconfguraton usng Mxed-Integer Convex Programmng, IEEE Transactons on Power Systems, vol. 7, no., pp. 06-5, 0. [8] V. Farahan, B. Vahd, and H. A. Abyaneh, Reconfguraton and capactor placement smultaneously for energy loss reducton based on an mproved reconfguraton method, IEEE Transactons on Power Systems, vol. 7, no., pp , 0. [9] L. Qngmn, L. Hongshun, J. Lou, and. Lang, Impact FCL on the rate of rse of recovery voltage wth crcut breakers, IEEE Transactons on Power Delvery, vol. 3, no. 4, pp , 008. [0] T. Ise, N. H. Nguyen, and S. Kumaga, Reducton of nductance and current ratng of the col and enhancement of fault current lmtng capablty of a rectfer type superconductng fault current lmter, IEEE Transactons on Appled Superconductvty, vol., no., pp , 00. [] M. Ichkawa and M. Okazak, A magnetc sheldng type superconductng fault current lmter usng a BI thck flm cylnder, IEEE Transactons on Appled Superconductvty, vol. 5, no., pp , 995. [] J. Lee and S. K. Joo, Economc assessment method for superconductng fault current lmter (SFCL) n fault current-constraned power system operaton, IEEE Transactons on Appled Superconductvty, vol. 3, no. 3, pp , 03. [3] S. B. Rhee, J. K. Lee, and B. W. Lee, Impacts of superconductng fault current lmters on the recloser operaton n dstrbuton electrc power systems, IEEE Transactons on Appled Superconductvty, vol., no. 3, pp , 0. [4] Y. Shahbaz, K. Nayesh, and H. Mohsen, "Fnte element method analyss of performance of nductve saturable-core fault current lmter," 0 st Internatonal Conference on Electrc Power Equpment - Swtchng Technology (ICEPE-ST), pp , Oct. 0. [5] Ioka, Y. Yokomzu, and T. Matsumura, "Influence of fault current lmter on solated operaton of customer system wth synchronous generator," Power and Energy Socety General Meetng - Converson and Delvery of Electrcal Energy n the st Century, 008 IEEE, pp. - 5, July 008. [6] I. G. Im, H. S. Cho, and H. W. Cho, Lmtaton of fault current and burden of superconductng element appled to neutral lne of the transformer, IEEE Transactons on Appled Superconductvty, vol. 4, no. 3, pp. -4, 04. [7] Y. C. hang and R. A. Dougal, "State of the art of Fault Current Lmters and ther applcatons n smart grd," Power and Energy Socety General Meetng, 0 IEEE, pp. -6, July 0. [8] S. M. Cho, H. S. Shn, and J. C. Km, Study on coordnaton of protectve relays between prmary feeder and nterconnectng transformer grounded by SFCL of wnd farm, IEEE Transactons on Appled Superconductvty, vol., no. 3, June 0 [9] K. Hongesombut, Y. Mtan, and K. Tsuj, Optmal locaton assgnment and desgn of superconductng fault current lmters appled to loop power systems, IEEE Transactons on Appled Superconductvty, vol. 3, no., pp , 003. [0] J. H. Teng and C. N. Lu,, "Optmum fault current lmter placement," Internatonal Conference on Intellgent Systems Applcatons to Power Systems, pp. -6, Nov [] J. H. Teng and C. N. Lu, Optmum fault current lmter placement wth search space reducton technque, IET Generaton, Transmsson & Dstrbuton, vol. 4, no. 4, pp. 485, 00. [] Y. C. Huang, Enhanced genetc algorthm-based fuzzy mult-objectve approach to dstrbuton network reconfguraton, IET Transactons on Generaton, Transmsson and Dstrbuton, vol. 49, no. 5, pp , 00. [3] C. S. Chang,. X. Wang, F. Yang, and W. W. Tan, Herarchcal fuzzy logc system for mplementng mantenance schedules of offshore power systems, IEEE Transactons on Smart Grd, vol. 3, no., pp , 0. [4] M. Mohammadan, Supervsed learnng of fuzzy logc systems, vol. I., Unversty of Canberra, Australa, 009, p [5] L. F. Wang and C.N. Sngh, Populaton-based ntellgent search n relablty evaluaton of generaton systems wth wnd power penetraton, IEEE Transactons on Power Systems, vol. 3, no. 3, 008 [6] Ghahreman and I. Kamwa, Optmal placement of multple-type FACTS devces to maxmze power system loadablty usng a generc graphcal user nterface, IEEE Transactons on Power Systems, vol. 8, no., pp , 03. [7] W. F. Usda, D. V. Coury, R. A. Flauzno, and I. N. da Slva, Effcent placement of fault ndcators n an actual dstrbuton system usng evolutonary computng, IEEE Transactons on Power Systems, vol. 7, no. 4, pp , 0. [8] M. A. S. Masoum, M. Ladjevard, A. Jafaran, and E. F. Fuchs, Optmal placement, replacement and szng of capactor banks n dstorted dstrbuton networks by genetc algorthms, IEEE Transactons on Power Delvery, vol. 9, no. 4, pp , 004. [9] K. H. Km, S. B. Rhee, S. N. Km, and S. K. You, Applcaton of ESGA hybrd approach for voltage profle mprovement by capactor placement, IEEE Transactons on Power Delvery, vol. 8, no. 4, pp. 56-5, 003. [30] J. D. Glover, M. S. Sarma, and T. Overbye, Power system analyss & Desgn, 5th ed., U.S.A, January 3, 0. [3] IEEE Standard Ratng Structure for AC Hgh-Voltage Crcut Breakers Rated on a Symmetrcal Current Bass, ANSI IEEE Std C , May, 009 [3] W. S. Chen, H. T. Yang, and H. Y. Huang, Optmal desgn of support nsulators usng hashng ntegrated genetc algorthm and optmzed charge smulaton method, IEEE Transactons on Delectrcs and Electrcal Insulaton, vol. 5, no., pp , 008.