Transmission Loss Minimization Using SVC Based on Particle Swarm Optimization

Transcription

1 0 IEEE Symposum on Industral Electroncs and Applcatons (ISIEA0), September 5-8, 0, Langa, Malaysa Transmsson Loss Mnmzaton Usng Based on Partcle Sarm Optmzaton St Amely Jumaat Faculty of Electrcal and Electroncs Engneerng Unverst Tun Hussen Onn Malaysa Part Raja, Batu Pahat, Johor, Malaysa Ismal Musrn, 3 Muhammad Murtadha Othman, 4 Hazle Mohls,3 Faculty of Electrcal Engneerng Unverst Tenolog MARA Malaysa Shah Alam, Malaysa I_musrn@yahoo.co.u, 3 mamat505my@yahoo.com 4 Department of Electrcal Engneerng Faculty of Engneerng Unversty of Malaya 4 hazl@um.edu.my Abstract Ths paper descrbes optmal szng of statc var compensator () based on Partcle Sarm Optmzaton for mnmzaton of transmsson losses consderng cost functon. Partcle Sarm Optmzaton (PSO) s one of the artfcal ntellgent search approaches hch has the potental to solve such a problem. For ths study, statc var compensator () s chosen as the compensaton devce. Valdaton through the mplementaton on the IEEE 6-bus system shos that the PSO s found feasble to acheve the tas. The smulatons results are compared th those obtaned from the Bee Algorthm (BA) technque n the attempt to hghlght ts mert. Keyords- FACTS devces, optmal szng, partcle sarm optmzaton, transmsson loss, mnmzaton, statc var compensator. I. INTRODUCTION Noadays, the demands for electrcal energy have ncreased contnuously yearly. The nstallaton of ne transmsson netors or ne poer plants can solve these demands. Hoever, there are some lmtatons to develop ne system,.e. nstallaton cost, envronment mpact and polluton control, and the land acquston. One of alternatve solutons to respond to the demands s by usng flexble alternatng current transmsson system (FACTS). The FACTS s a concept proposed by N.G. Hngoran [] as a ell-non term for hgher controllablty n poer systems by means of poer electroncs devces. FACTS devces can provde benefts n ncreasng system transmsson capacty and poer flo control flexblty and rapdty []. Populaton base, cooperatve and compettve stochastc search algorthms are very popular n ths recent year n the research area of computatonal ntellgence. Genetc Algorthm (GA) [6] and Evolutonary Programmng (EP) [7] technques are found to been establshed search algorthms and has been successfully mplemented n solvng the complex problem effectvely. PSO algorthm orgnally s developed by Kennedy and Eberhart based on the socal behavors of anmal sarms (e.g. brd blocs and fsh schools) [7]. PSO appled for solvng varous optmzaton problems n electrcal engneerng [, 3, 8-0]. Optmal locatons of dfferent types of FACTS devces n the poer system has been attempted usng dfferent EP technques such as Hybrd Tabu Search and Smulated Annealng (TS/SA), GA, Repettve Poer Flo method (RPF), BA and Fuzzy decson mang and PSO. The mum ncrease n system loadablty s acheved by GA and PSO technques th an optmal numbers of fve TCSCs devces n the system. From the results t s shon that TCSC devce has mproved the lne flos even to ther thermal lmts [3]. Wth multtype of FACTS devces nstalled; the reducton n total generator of fuel cost s more than the ndvdual nstalled FACTS devces [4]. The hybrd TS/SA converges at a faster computaton tme. In [5], BA does not requre external parameters such as cross over rate and mutaton rate. BA gves better result n terms of speed of optmzatons and accuracy of the results. BA needs the large number of trals. On the other hand, GA based approach s proposed to determne the sutable types of FACTS devces and evaluate the total costs system [6]. EP n [7] s used to dentfy the locaton of four FACTS devces. Optmal Poer Flo usng GA can also used to obtan the optmal locatons of s. The results shon that ths method can be used to mnmze the total cost functon, ncludng generatons cost of poer plants and nvestments costs n [8]. In [9], GA and PSO are used to optmze the parameters of TCSC. Hoever, PSO have more advantageous than the GA. PSO gves a better balanced mechansm and better varaton to the global and local exploraton abltes. Moreover, t can be appled to solve varous optmzaton problems n poer system such as poer //$ IEEE 49

2 0 IEEE Symposum on Industral Electroncs and Applcatons (ISIEA0), September 5-8, 0, Langa, Malaysa system stablty enhancement and capactor placement problems [0]. Ths paper presents PSO technque for loss mnmzaton n poer system by usng. PSO as adopted to optmze the s szng to be nstalled n poer transmsson netor. Placement of s done emprcally as the plot study. The PSO and BA technques hch performed on the IEEE 6-bus RTS have ndcated that the proposed methods are found to be crucal n loss mnmzaton scheme. II. MATERIAL AND METHODS Flexble AC Transmsson Systems (FACTS) devces have several types namely: thyrstor controlled statc compensator (TCSC), statc var compensator (), unfed poer flo controller (UPFC), statc compensator (STATCOM), and thyrstor controlled phase shfter transformer (TCPST) [- ]. The s a shunt type FACTS devce hch defned as a shunt connected statc var generator or absorber hose output s adjusted n order to exchange the capactve or nductve current to mantan or control specfc parameters of the poer system, typcally the bus voltage [3]. The can nject or absorb ts reactve poer (Q ) at a chosen bus. It njects reactve poer nto the system Q < 0 and absorbs reactve poer from the system f Q > 0 [4]. The orng range of s beteen +0MVar and +00MVar []. The s modeled as a generator or absorber of reactve poer as shon n fgure. It s modeled as an deal reactve poer njecton at bus, as shon n fgure. The njected poer at bus s: [5-6]. ΔQ s = Q () Fgure Bloc dagram of Fgure Mathematcal model of A. Cost of Installaton The cost of nstallaton of devces has been mathematcally formulated and gven by the follong equaton [, ]: IC = C S 000 Q mn Q Q (). Where IC = the nstallaton cost of devces n [US$], C = the cost of devces n [US$/KVar] Installaton of devce can be calculated usng the cost functon gven by [, 6, 5]. C = S 0.305S+ 7.38[US$ /KVar] (3) S = Q Q (4) here S = operatng range of n [MVar] Q = reactve poer flo through the branch before nstallaton. Q = reactve poer flo through the branch after nstallaton. III. PARTICLE SWARM OPTIMIZATION (PSO) The PSO provdes a populaton-based search procedure n hch ndvduals called partcles and changes ther postons. The poston of each partcle s presented n X-Y plane. Each partcle moves to the ne poston usng velocty accordng to ts on experence as called P best. G best s the overall best value obtaned so far by any partcle n the populaton. By tme to tme, the PSO conssts of velocty changes of each partcle toards ts P best and G best [8-9]. Each partcle tres to modfy ts current poston and velocty accordng to the dstance beteen ts current poston and P best, and the current poston and G best. After fndng the best values the partcle updates ts velocty and poston. Velocty of each partcle can be modfed by equaton (5) [, 3, 0]. here rand v + = v + c rand ( P best s ) + c rand ( G s best ) c and v + = velocty of partcle ι at teratons = eght functon = eght coeffcen t both equal to () 5 and c rand = random number beteen 0 and s = current poston of partcle at teraton Pbest = best poston of partcle - th up to the current teraton Gbest = best overall poston found by the partcle up to the current teraton. Weght functon s gven by (6) [, 3, 4, 0] = mn ter ter (6). 40

3 0 IEEE Symposum on Industral Electroncs and Applcatons (ISIEA0), September 5-8, 0, Langa, Malaysa here = ntal eght equal to 0.9 mn = ntal eght equal to 0.4 ter = mum teraton number,and ter = current teraton number. The ne poston can be modfed (7) s + = v + s (7). The computaton procedure for the developed PSO algorthm taes the follong steps [0]: Step : Generaton of ntal condton of each partcle. Intal searchng pont (s 0 ) and veloctes (v 0 ) of each partcle are usually random thn t range. The current searchng pont s set to P best for each partcle. The best evaluated value of P best s set to G best, and the best value s stored. Step : Evaluaton of searchng pont of each partcle. The objectve functon s evaluated for each partcle. If the value s better than the current P best of the partcle, the P best value ll be replaced by the current value. If the P best value s better than the current G best,g best ll be replaced by the best value and the best value s stored. Step 3: Modfcaton of each search pont. The current searchng pont of each partcle s updated usng (5), (6) and (7). Step 4: Checng the ext condton. The current teraton number reaches the pre-determned mum teraton number as the stoppng crteron. Otherse the process proceeds to step. IV. TEST RESULTS In order to realze the effectveness of the proposed PSO technque, the IEEE 6-bus system as tested n order to fnd the optmal szng of. The lne data and the bus data of the IEEE 6-bus system are gven n []. The parameters of the optmzaton algorthm are lsted n Table [, 3, 4, 0]. TABLE I PARAMETERS OF OPTIMIZATIONS TECHNIQUES Parameters PSO Populaton Sze 5, 0, 5, 0 Inertal Weght, Constant, C Constant, C Number of teraton 50 Rand 0 to Rand 0 to The nstallatons n the transmsson system to reduce the transmsson loss n the system have been conducted at several load condtons subjected to buses 6, 0, and 3. The smulaton as done n MATLAB and several buses ere subjected to load varatons n order to realze the effect of s nstallaton th PSO as the optmzaton approach. Three s ere nstalled at bus 9,, and 4 and assgned as x, x and x 3. These buses are the ea load buses n the system; havng among the loest n mum loadablty condton []. A. Transmsson Loss Reducton th Installatons. Results for transmsson loss reducton hen load.e. buses 6, 0 and 3 are subjected to load varaton are tabulated n Table, 3 and 4. The szng of s to acheve loss reducton at several loadng condtons can be referred to the same table. TABLE TRANSMISSION LOSS REDUCTION LOAD VARIATION AT BUS 6. Loadng Condton Total Poer Loss (MW) thout th szng Cost (US$/KVar) For nstance n Table th loadng condton of 50MVar, the transmsson loss has been reduced from 5.760MW to 4.498MW. In order to acheve ths, the szng of s are MVar, MVar, and 7.903MVar as ndcated n the table. The cost of nstallaton at ths scenaro s thousand US$. From Table t s observed that the value of transmsson losses s decreased rapdly and the cost of nstallaton s ncreased accordngly as the reactve poer loadng s ncreased. It s also shon that, th the nstallaton of the transmsson loss of the bus for all loadng condton has been reduced sgnfcantly. It s pretty obvous that, th the nstallaton of optmzed usng PSO, the transmsson losses has been able reduced at all loadng condtons. IC 0 3 (US$)

4 0 IEEE Symposum on Industral Electroncs and Applcatons (ISIEA0), September 5-8, 0, Langa, Malaysa TABLE 3 TRANSMISSION LOSS REDUCTION LOAD VARIATION AT BUS 0 Loadng Condton Transmsson Loss (MW) thout th szng Cost (US$/KVar) The szng of s to acheve optmal loss reducton at several loadng condtons can also be referred to Table 3. For nstance, at loadng condton of 50MVar the transmsson loss has been reduced from MW to 4.48MW. In order to acheve ths, the szng of s are MVar, MVar and 3.545MVAr as ndcated n the Table 3. The cost of nstallaton at ths scenaro s thousand US$. From Table 3 t s observed that the value of transmsson losses s decreased hle the nstallaton cost s ncreased accordngly as the reactve poer loadng s ncreased. The results of s szng to acheve optmal loss reducton at several loadng condtons subjected to bus 3 can be referred to Table 4. For nstance, at loadng condton of 50MVar, the transmsson loss has been reduced from 5.760MW to 4.504MW. In order to acheve ths value the szng of s are been proposed as MVar, MVar, and 8.644MVar as ndcated n the Table 4. The cost of nstallaton at ths scenaro s thousand US$. Smlar phenomenon can be observed n ths case here by; Table 4 t s observed that the value of transmsson losses s decreased and the cost of nstallaton s ncreased accordngly as the reactve poer loadng s ncreased. IC 0 3 (US$) technque can be optmzng the transmsson loss loer than PSO. Hoever the dfferent results can be observed at load condton of 40MVar and 50Mvar. From Table 5, at loadng condton of 50MVar; PSO managed to reduce the transmsson loss from 5.760MW to 4.504MW, hle BA managed to reduce the transmsson loss to 4.599MW. From ths smulaton shos that the result usng by PSO s loer than the BA t hch ndcates the capablty of PSO throughout the process. TABLE 4 TRANSMISSION LOSS REDUCTION LOAD VARIATION AT BUS 3 Loadng Condton Transmsson Loss (MW) thout TABLE 5 TRANSMISSION LOSS REDUCTION LOAD VARIATION AT BUS 6 PERFORMED USING PSO AND BA Loadng th Szng Cost (US$/KVar) Transmsson Loss (MW) Condton thout PSO BA IC 0 3 (US$) B. Comparatve Studes th Other Technque Comparatve studes ere conducted th respect to the results obtaned usng BA []. The results are tabulated n Table 5, 6 and 7 for load subjected to buses 6, 0 and 3. In Table 5 at loadng condton of 0MVar; PSO managed to reduce the transmsson loss from MW to MW, hle BA managed to reduce the transmsson loss to 4.37MW. The same scenaros can be observed as ell th 0MVar and 30MVar. It s shon that, BA Fgure 6: Results for transmsson losses load varaton reducton at bus 6 performed usng PSO and BA 4

5 0 IEEE Symposum on Industral Electroncs and Applcatons (ISIEA0), September 5-8, 0, Langa, Malaysa On the other hand, Fgure 6 llustrates the transmsson loss n the system hen the load at bus 6 s gradually ncreased. Ths s n fact exhblty the profles of transmsson loss varaton performed usng PSO and BA. Both PSO and BA are comparable. In Table 6, at loadng condton of 0MVar; PSO managed to reduce the transmsson loss from MW to MW, hle BA able to reduce the transmsson loss to 4.47MW. The same scenaros can be observed at 0MVar and 30MVar. It s shon that, BA technque can be mnmze the transmsson loss loer than PSO. Hoever the dfferent results can be observed at load condton of 40MVar and 50Mvar. From Table 6, at loadng condton of 50MVar, PSO managed to reduce the transmsson loss from MW to 4.48MW, hle BA managed to reduce the transmsson loss to 4.555MW. It s shon that the result by PSO s loer than BA. Fgure 7 llustrates the transmsson loss n the system hen the load at bus 0 s gradually ncreased. TABLE 6 TRANSMISSION LOSS REDUCTION LOAD VARIATION AT BUS 0 PERFORMED USING PSO AND BA Loadng Transmsson Loss (MW) Condton thout PSO BA Smlar phenomenon observed at bus 3. In Table 7, at loadng condton of 0MVar, PSO able to reduce the transmsson loss from MW to 4.493MW, hle BA able to reduce the transmsson loss to MW. The same scenaros can be observed at 0MVar and 30MVar. Hoever the dfferent results can be observed at load condton of 40MVar and 50Mvar. From Table 7, at loadng condton of 50MVar, PSO able to reduce the transmsson loss from 5.760MW to 4.498MW, hle BA able to reduce the transmsson loss to 4.58MW. Fgure 8 llustrates the transmsson loss n the system hen the load at bus 3 s gradually ncreased. It s shon that PSO able mnmze the transmsson loss better than BA hen loadng condton s greater than 30MVar. Fgure 7: Results for transmsson losses reducton load varaton at bus 0 performed usng PSO and BA TABLE 7 TRANSMISSION LOSS REDUCTION LOAD VARIATION AT BUS 3 PERFORMED USING PSO AND BA Loadng Transmsson Loss (MW) Condton thout PSO BA Fgure 8: Results for transmsson losses reducton load varaton at bus 3 performed usng PSO and BA V. CONCLUSION An approaches for transmsson loss reducton by usng nstallaton va PSO and BA as the optmzaton technques are presented. Source code of PSO optmzatons technque as developed to determne the optmal szng of n order to mnmze the losses n the transmsson system. Besdes that, the cost of nstallaton s consdered n the system. Tests are performed on the IEEE 6 bus RTS. Result shos that the mplementatons of PSO and BA have reduced the transmsson losses of the system ndcatng t as a feasble technque to perform the losses optmzatons. 43

6 0 IEEE Symposum on Industral Electroncs and Applcatons (ISIEA0), September 5-8, 0, Langa, Malaysa REFERENCES [] G.Hngoran, Poer electroncs n electrcal utltes: role of poer electroncs n future poer systems, n Proc 988 IEEE, Vol. 76 No, 4 Aprl 988, pp [] Saravanan. M, Slochanal. S.M.R, Venatesh. P, Abraham, P.S, Applcatons of PSO Technque for Optmal Locaton of FACTS Devces Consderng System Loadablty and Cost of Installaton, n Proc th Internatonal Poer Engneerng Conference (IPEC). Pp Vol. 005 [3] G.I.Rashed, H.I.Shaheen, S.J.Cheng, Optmal Locatons and Parameters Settngs of Multple TSCSs for Increasng Poer System Loadablty Based on GA and PSO technques, n Proc 007 Thrd IEEE Internatonal Conference on Natural Computaton (ICNC 007), 007. [4] P.Bhasaputra, and W.Ongsaul, Optmal Poer Flo th Multtype of FACTS Devces by Hybrd TS/SA Approach, n Proc. IEEE ICIT 0 Bango Thaland, 00. [5] Idrs. R.M, Kharuddn. A, and Mustafa, M.W, Optmal Choce of FACTS devces for ATC Enhancement Usng Bees Algorthm, n Proc. 009, Poer Engneerng Conference, 009. (AUPEC 009). Australasan Unverstes 009, pp. 6, 009. [6] Phashant Kumar Toar, and Tog Raj Sood, Optmal Locaton of FACTS Devces n Poer System Usng Genetc Algorthm, n Proc. IEEE World Congress on Nature And Bologcally Inspred Computng (NaBIC 009), 009. [7] W. Ongsaul, and P. Jrapong, Optmal allocatons of FACTS devces to enhance total transfer capablty usng evolutonary programmng, n Proc. Internatonal Symposum on Crcuts and Systems, Japan, 3-6 May vol. 5 pp , 005. [8] M.M.E. Metally, A.A. E. Emary, F.M.E.Bendary, and M.I. Mosaad, Optmal allocatons of FACTS devces to enhance total transfer capablty usng evolutonary programmng, n Proc. Poer System Conference, 008, MEPCON 008, th Internatonal Mddle East, pp. -4. [9] S. Panda, and N.P.Padly, Comparson of partcle sarm optmzaton and genetc algorthm for FACTS-based controller desgn, Appled Soft Computng, vol 8, Issue 4, pp. 5-59, Mar [0] S. Chansareettaya, and P, Jrapong, Poer Transfer Cabablty Enhancement th Multtype FACTS Controller Usng Partcle Sarm Optmzaton, n Proc IEEE TENCON 00. [] Ismal Musrn, Nur Danah Mohd Radz, Muhammad Murtadha Othman, Mohamad Khayat Idrs, Tt Khaa Abdul Rahman, Voltage Profle Improvement Usng Unfed Poer Flo Controller va Artcal Immune System, WSEAS Transacton on Poer Systems, Issue 4, Volume 3, Aprl 008. [] N.G.Hngoran, and L.Gygy, Understandng FACTS: Concept and Technology of Flexble AC Transmsson Systems, n Proc. IEEE Press, 000, p [3] S. Aucharyamet, and S.Srsmrannuul, Optmal Reactve Poer Plannng th FACTS Devces by Partcle Sarm Optmzaton, In Proc. 8 th Internatonal conference on Advances n Poer System Control, Operaton and Management (APSCOM 009), 009, pp [4] L.J. Ca, I.Erlch, and G.Stamtss, Optmal Choce and Allocaton of FACTS Devces n Deregulated Electrcty Maret usng Genetc Algorthm, n Proc. Poer Systems Conference and Exposton 004. (IEEE PES 004) 004, pp vol [5] Nor Rul Hasma Abdullah, Ismal Musrn, Muhammad Murtadha Othaman, Statc VAR Compensator for Mnmsng Transmsson Loss and Installaton Cost Calculaton, Australan Journal of Basc and Appled Scences, 4(4): , 00. [6] R. D. Zmmermann, and D. Gan, "Matpoer a Matlab poer system smulaton pacage, User s Manual, Verson.0, Dec [7] J. Kennedy, and R. Eberhart, Partcle Sarm Optmzaton, n Proc. 995 IEEE Internatonal Conf. on Neural Netor, vol 4, pp [8] Chansareettaya. S, and Jrapong, P, Poer Transfer Capablty Enhancement th Multtype FACTS Controllers Usng Partcle Sarm Optmzaton, n Proc. 00 IEEE TENCON Conference 00, pp 4-47E. 00. [9] Hashemzadeh. H, and Hossen. S.H., Locatng Seres FACTS Devces Usng Lne Outage Senstvty Factors and Partcle Sarms Optmzaton for Congeston Management, Poer & Energy Socety General Meetng, 009 (PES '09). pp -6, 009. [0] Sundaresaran. K, Harharan. B, Parasser. F.P, Antony. D.S, and Subar, B, Optmal Placement of Statc VAr Compensators ( s) Usng Partcle Sarm Optmzaton, n Proc 00, Internatonal Conference on Poer, Control and Embedded Systems (ICPCES), 00, pp. 4, 00. [] N. Othman, I. Musrn, M.A.Rahm, and Z.Othman, Bees Algorthm Technque for Loss Mnmzaton n Poer Transmsson Netor Usng Statc Var Compensator, n Proc The 4 th Internatonal Poer Engnnerng and Optmzaton Conference (PEOCO 00), 3-4 June 00. BIOGRAPHIES St Amely Jumaat graduated from the Insttut Tun Hussen Onn (ITTHO-UTM) th honours degree n BSc. Electrcal Eng. n 00 and MEng. (Poer), UTM n 003. She s currently s pursung a PhD n poer system at Unverst Tenolog MARA, Malaysa. Her research nterests nclude poer system stablty, facts devces and Artfcal Intellgent technques. Assocate Professor Dr. Ismal Musrn obtaned Dploma of Electrcal Poer Engneerng n 987, Bachelor of Electrcal Engneerng (Hons) n 990; both from Unverst Tenolog Malaysa, MSc n Pulsed Poer Technology n 99 from Unversty of Strathclyde, Unted Kngdom and PhD n Electrcal Engneerng from Unverst Tenolog MARA, Malaysa n 004. Hs research nterest ncludes poer system stablty, optmzaton technques, dstrbuted generator and artfcal ntellgence. Dr. Muhammad Murtadha bn Othman receved hs B.Eng. (Hons) from Staffordshre Unversty, England n 998; M.Sc from Unverst Putra Malaysa n 000 and PhD from Unverst Kebangsaan Malaysa n 006. He currently lectures at the UnverstTenolog MARA, Malaysa. Hs area of research nterests are artfcal ntellgence, transfer capablty assessment and relablty studes n a deregulated poer system. Dr. Hazle Bn Mohls receved M.Sc. degree from Malaya Unversty and Ph.D degree n 009 from Manchester Unversty, Unted Kngdom. Hs research nterest s Electrcal poer system (Transmsson and Dstrbuton) Poer qualty, stablty, EMC & Islandng 44