A Two Stage Methodology for Siting and Sizing of DG for Minimum Loss in Radial Distribution System using RCGA

Size: px

Start display at page:

Download "A Two Stage Methodology for Siting and Sizing of DG for Minimum Loss in Radial Distribution System using RCGA"

Geoffrey West
6 years ago
Views:

1 teratoal Joural of Computer Applcatos ( ) Volume 5 No., July 0 A Two Stage Methodology for Stg ad Szg of for Mmum Loss adal Dstrbuto System usg CGA Dr.M.adma ltha EEE Member rofessor & HOD Dept. of E.E.E Uversty, A..T.S, ajampet, A.. Dr.V.C. Veera eddy rofessor & HOD Dept. of E.E.E S.V.U.C.E,S.V.Uversty, Trupath, A. N.Svaram eddy rofessor & HOD Dept. of Mech.Egg., A..T.S, ajampet, A.. V.Usha edddy Assstat rofessor Dept. of E.E.E S.V.U.C.E,S.V. Trupath, A.. ABSTACT Ths paper presets a ew methodology usg eal Coded Geetc Algorthm (CGA) for the placemet of Dstrbuted Geerators() the radal dstrbuto systems to reduce the real power losses ad to mprove the voltage profle. A two-stage methodology s used for the optmal placemet. the frst stage, sgle placemet algorthm s used to fd the optmal locatos ad the secod stage, eal Coded Geetc Algorthm s used to fd the sze of the s correspodg to maxmum loss reducto. The proposed method s tested o stadard EEE 33 bus test system ad the results are preseted. Keywords: placemet, eal Coded Geetc Algorthm, loss reducto, radal dstrbuto system.. NTODUCTON Dstrbuted or dspersed geerato () or embedded geerato (EG) s small-scale power geerato that s usually coected to or embedded the dstrbuto system. The term also mples the use of ay modular techology that s sted throughout a utlty s servce area (tercoected to the dstrbuto or sub-trasmsso system) to lower the cost of servce []. The beefts of are umerous [, 3] ad the reasos for mplemetg s are a eergy effcecy or ratoal use of eergy, deregulato or competto polcy, dversfcato of eergy sources, avalablty of modular geeratg plat, ease of fdg stes for smaller geerators, shorter costructo tmes ad lower captal costs of smaller plats ad proxmty of the geerato plat to heavy loads, whch reduces trasmsso costs. Also t s accepted by may coutres that the reducto gaseous emssos (maly CO ) offered by s s major legal drver for mplemetato [4]. The dstrbuto plag problem s to detfy a combato of expaso projects that satsfy load growth costrats wthout volatg ay system costrats such as equpmet overloadg [5]. Dstrbuto etwork plag s to detfy the least cost etwork vestmet that satsfes load growth requremets wthout volatg ay system ad operatoal costrats. Due to ther hgh effcecy, small sze, low vestmet cost, modularty ad ablty to explot reewable eergy sources, are creasgly becomg a attractve alteratve to etwork reforcemet ad expaso..numerous studes used dfferet approaches to evaluate the beefts from s to a etwork the form of loss reducto, loadg level reducto [6-8]. Naresh Acharya et al suggested a heurstc method [9] to select approprate locato ad to calculate sze for mmum real power losses. Though the method s effectve selectg locato, t requres more computatoal efforts. The optmal value of for mmum system losses s calculated at each bus. lacg the calculated sze for the buses oe by oe, correspodg system losses are calculated ad compared to decde the approprate locato. Ths method s used to calculate sze based o approxmate loss formula may lead to a approprate soluto. the lterature, geetc algorthm has bee appled to placemet [0-]. all these works both szg ad locato of s are determed by GA. ths paper, the optmal locatos of dstrbuted geerators are detfed based o the sgle placemet algorthm ad a GA based techque whch takes the umber ad locato of s as put has bee developed to determe the optmal sze(s) of to mmze real power losses dstrbuto systems. The advatages of relevg GA from determato of locatos of s are mproved covergece characterstcs ad less computato tme. Voltage ad thermal costrats are cosdered. The effectveess of the proposed algorthm was valdated usg 33-Bus Dstrbuto System [3].To test the effectveess of proposed method, results are compared wth the results of a aalytcal method reported [4]. t s observed that the proposed method yeld more savgs as compared to aalytcal method.. THEOETCAL BACKGOUND The total loss ( L ) a dstrbuto system havg umber of braches s gve by: Lt = = Here s the magtude of the brach curret ad s the resstace of the th brach respectvely. The brach curret ca be obtaed from the load flow soluto. The brach curret has two compoets, actve compoet ( a ) ad reactve compoet ( r ). The loss assocated wth the actve ad reactve compoets of brach currets ca be wrtte as: () 0

2 teratoal Joural of Computer Applcatos ( ) Volume 5 No., July 0 Lr = = = = a r Note that for a gve cofgurato of a sgle-source radal etwork, the loss assocated wth the actve compoet of brach currets caot be mmzed because all actve power must be suppled by the source at the root bus. However by placg s, the actve compoet of brach currets are compesated ad losses due to actve compoet of brach curret s reduced. Ths paper presets a method that mmzes the loss due to the actve compoet of the brach curret by optmally placg the s ad thereby reduces the total loss the dstrbuto system. A two stage methodology s appled here. the frst stage optmum locato of the s are determed by usg a aalytcal approach ad the secod stage CGA s used to determe szes of the s for maxmum real loss reducto. 3. DENTFCATON OF OTMAL LOCATONS STAGE- 3.. Sgle lacemet Algorthm Ths algorthm determes the optmal sze ad locato of uts that should be placed the system where maxmum loss savg occurs. Frst optmum szes of uts for all odes are determed for base case ad best oe s chose based o the maxmum loss savg. f sgle placemet s requred ths process s stopped here. Ths process s repeated f multple locatos are requred by modfyg the base system by sertg a ut to the system oe-by-oe. 3.. Sgle lacemet Algorthm Assume that a sgle-source radal dstrbuto system wth braches ad a s to be placed at bus m ad α be a set of braches coected betwee the source ad bus m. As referred to Fg., assume that a s placed at bus 3, the set of α cossts of braches,,3,4,5, 6,7,8,9,0, ad. The produces actve curret, ad for a radal etwork t chages oly the actve compoet of curret of brach set α. The curret of other braches ( =α) are uaffected by the. Thus the ew actve curret ew a of the th brach s gve by () (3) ew a = a + D (4) where D = ; f brach α = 0; otherwse The loss com assocated wth the actve compoet of brach currets the compesated currets the compesated system (whe the s coected) ca be wrtte as com = ( a + D ) (5) = The loss savg S s the dfferece betwee equato ad 5 ad s gve by S = com (6) = (D a + D c ) = The curret that provdes the maxmum loss savg ca be obtaed from S = ( D a + D ) = 0 (7) = Thus the curret for the maxmum loss savg s = = = D a D = a α α The correspodg sze s = V m (9) V m s the voltage magtude of the bus m. The optmum sze of for each bus s determed usg eq (9). The possble loss savg for each s determed by usg eq (6).The wth hghest loss savg s detfed as caddate locato for sgle Dg placemet. Whe the caddate bus s detfed ad s placed, the above techque ca also be used to detfy the ext ad subsequet bus to be compesated for loss reducto Algorthm Step : Coduct load flow aalyss for the orgal system. Step : Calculate the currets ( ) ad sze usg equatos 8 & 9 from = for all buses except source bus. Step 3: Determe loss savg (S) usg equato 6, from = for all buses except source bus. Step 4: (8) detfy the maxmum savg ad the correspodg sze. Step5: Modfy the actve load o the etwork ad coduct the load flow aga. Step 6: Check whether the savg obta s more tha kw. f the savg s sgfcat, go to step. Otherwse, go to ext step. Step 7: Ths bus s a caddate bus where s placed. epeat the procedure utl the loss s sgfcat (5 kw). Here the effect of placemet o real power loss oly s cosdered. The effect of o eactve power loss, voltage profle ad system capacty rse s eglected. Sce the s are added to the system oe by oe, the szes obtaed usg sgle placemet algorthm are local optma ot global optmum soluto. The global optmal soluto s obtaed f multple s are smultaeously placed the system by usg geetc algorthm. Ths method s explaed ext secto.

3 teratoal Joural of Computer Applcatos ( ) Volume 5 No., July 0 4. DENTFCATON OF OTMAL SZES BY EAL CODED GENETC ALGOTHM 4.. troducto to Geetc Algorthm Geetc algorthms are practcal, robust optmzato ad search methods. Geetc algorthms were veted by Hollad to mmc some of the processes of atural evoluto ad selecto. These algorthms are dfferet from most of the tradtoal optmzato methods ad these algorthms eed desg space to be coverted to geetc space. A more strkg dfferece betwee geetc algorthms ad most of the tradtoal optmzato methods s that GA uses a populato of pots at oe tme, cotrast to the sgle pot approach by tradtoal optmzato methods. The most terestg aspect of GA s that they do ot requre ay pror kowledge of the fucto to be optmzed ad they exhbt very good performace o the majorty of the problems appled. The geetc algorthm repeatedly modfes a populato of dvdual solutos. At each step, the geetc algorthm selects dvduals at radom from the curret populato to be parets ad uses them to produce the chldre for the ext geerato. Over successve geeratos, the populato evolves towards a optmal soluto. The geetc algorthms ca be used to solve a varety of optmzato problems that are ot well suted for stadard optmzato algorthms. There are a umber of dfferet algorthms that ca be appled to may domas, from data aalyss to autoomous avgato [4]. These mmue algorthms were spred by works o theoretcal mmuology ad several processes that occur wth the S. The ASs lead to the developmet of dfferet techques, each oe mappg a dfferet mechasm of the system. For examples, the Artfcal mmue Networks as proposed by Farmer et al. [5], the Cloal Selecto Algorthm proposed by de Castro ad Vo Zube [6], ad the Negatve Selecto Algorthm troduced by Forrest et al. [7]. mmue etwork models are sutable to deal wth dyamc evromets ad optmzato problems, algorthms based upo the cloal selecto prcple are adequate to solve optmzato ad schedulg problems, ad the egatve selecto strateges are successfully appled to aomaly detecto. 4.. Applcato of Geetc Algorthm to determe ut szes After detfyg the umber of caddate locatos usg fuzzy approach, the szes all these caddate locatos are obtaed by usg the eal Coded Geetc Algorthm (CGA). ths proposed method, the real umber ecodg has bee used to determe the szes of umber of s the caddate locatos. Step : tal populato of [op x ] umber of real umbers s geerated radomly wth the lmts, where op s the tal populato sze ad s the umber of s. Each row represets oe possble soluto to the optmal -szg problem. terato cout s set to oe. Step : By placg all the s of each chromosome at the respectve caddate locatos ad load flow performed to fd the total real power loss L. The same procedure s repeated for the op umber of chromosomes to fd the total real power losses. Check voltage ad thermal costrats. f costrats are ot volated evaluate Ftess value. Step 3: The populato s arraged the descedg order accordg to ther ftess values. Maxmum ftess ad average ftess values are calculated. Error = (maxmum ftess - average ftess) (5) Error s calculated usg the equato (5). f ths error s less tha a specfed tolerace the go to step 9. Step 4: The best chromosomes are drectly coped to the ext geerato populato to perform the eltsm wth a probablty of e. Step 5: arets are selected pars by usg the roulette wheel selecto techque based o ther ftess values. Step 6: Crossover s performed usg the two crossover operators. These two crossover operators are the arthmetc crossover ad the heurstc crossover. A radom umber r s geerated betwee zero ad oe. f the radom umber r s less tha 0.5 the arthmetc crossover operator s used to produce the offsprg, otherwse heurstc crossover operator s used to produce the offsprg. Step 7: The terato cout s cremeted ad whether ths terato cout s greater tha terato maxmum or ot s checked. f t s greater tha terato cout the go to step 9. Step 8: After performg the eltsm ad crossover operators, the ew populato s geerated from the old populato. ths preset work mutato operator s elmated. Go to step to repeat the same procedure. Step 9: Stop the procedure ad prt the results. 5. ESULTS AND DSCUSSON Frst load flow s coducted for EEE 33 bus test system[7]. The power loss due to actve compoet of curret s kw ad power loss due to reactve compoet of the curret s kw. A program s wrtte MATLAB to calculate the loss savg, sze ad locato for maxmum loss savg. For the frst terato the maxmum loss savg s occurrg at bus 6. The caddate locato for s bus 6 wth a loss savg of 9.75 kw. The optmum sze of at bus 6 s.4886 MW. By assumg.4886 MW s coected at bus 6 of base system ad s cosdered as base case. Now the caddate locato s bus 5wth MW sze ad the loss savg s.4385 KW. Ths process s repeated tll the loss savg s sgfcat. The results are show Table.

4 teratoal Joural of Computer Applcatos ( ) Volume 5 No., July 0 terato No. Table -Sgle placemet results Bus No. Sze (MW) Savg (KW) The soluto obtaed above s local optmum soluto but ot global optmum soluto. The szes correspodg to global optmum soluto are determed usg CGA. The caddate locatos for placemet are take from sgle placemet algorthm.e. 6,5,5,3.Wth these locatos, szes of s are determed by usg eal Coded Geetc Algorthm descrbed secto 4. The szes of s are depedet o the umber of locatos. Geerally t s ot possble to stall may s a gve radal system. Here 4 cases are cosdered. case oly oe stallato s assumed. case two s, case three S ad the last case four s are assumed to be stalled. szes the four optmal locatos, total real power losses before ad after stallato for four cases are gve Table. Case bus locatos szes(mw) Total Sze(MW) V Table : esults of EEE 33 bus system. losses before stallato (Kw) loss after stallato (Kw) savg(kw ) savg/ sze The last colum Table represets the savg Kw for MW stallato. The case wth greater rato s ecoomcal. The case V s ecoomcally best tha other cases sce t has hghest rato ad the savg s also maxmum. As the umber of s stalled s creasg the savg s also creasg. case4 maxmum savg s acheved but the umber of s s four. Though the rato of sze to savg s mmum of all cases whch represet optmum soluto but the umber of s volved s four so t s ot ecoomcal by cosderg the cost of stallato of 4 s. But vew of relablty, qualty ad future expaso of the system t s the best soluto. Table 3 shows the mmum voltage ad % mprovemet mmum voltage compared to base case for all the four cases. all the cases voltage profle s mproved ad the mprovemet s very sgfcat case 4. The voltage profle for all cases s show Fg.. Table 3: Voltage mprovemet wth placemet case M Bus No. No. Voltage % chage Base case case case case case Table 4 shows % mprovemets power loss due to actve compoet of brach curret, reactve compoet of brach curret ad total actve power loss of the system the four cases cosdered. 3

teratoal Joural of Computer Applcatos (0975 8887) Volume 5 No., July 0 The loss due to actve compoet of brach curret s reduced by more tha 54% least ad early 80% at best.

5 teratoal Joural of Computer Applcatos ( ) Volume 5 No., July 0 The loss due to actve compoet of brach curret s reduced by more tha 54% least ad early 80% at best. Though the am s reducg the loss, the Lr loss s also reducg due to mprovemet voltage profle. From Table V t s observed that the total real power loss s reduced by 38% case ad 56% case 4. The covergece characterstcs of the soluto of Geetc algorthm for all the four cases are show fgure. Table 5 shows the mmum, average, maxmum values of total real power loss from 00 trals of Geetc algorthm. The average umber of teratos ad average CU tme are also show. Fg.:Voltage profle wth ad wthout placemet for all Cases case No. Table 4 :Loss reducto by placemet (kw) % Savg Lr (kw) % Savg Lt (kw) % Savg Base case case case case case Case 300 Case Best Ftess Best Ftess teratos teratos 00 Case 80 Case V 60 Best Ftess Best Ftess teratos teratos Fgure : Covergece characterstc of Geetc algorthm for 33 bus test system 4

6 teratoal Joural of Computer Applcatos ( ) Volume 5 No., July 0 Table 5: erformace of Geetc algorthm for EEE 33 Bus System Total real power loss (kw) Case Case Case Case V M Average Max No. of Atbodes Avg. No. of teratos Average Tme (Sec.) Comparso erformace To demostrate the valdty of the proposed method the results of proposed method are compared wth a exstg Aalytcal method. The comparso s show Table 6. From the table t s clear that the savgs by CGA algorthm are a lttle hgher tha the exstg aalytcal method. The reaso for ths s the creased sze of uts. Table 7 shows comparso of voltage profle mprovemet by the two methods. The mmum voltage ad % mprovemet mmum voltage compared to base case for all the four cases, for the two methods dscussed, are show ths Table. For all the four cases the mprovemet s better for the secod method. From the above tables t s clear that the results obtaed by the GA method ad the other method are smlar whch shows the effcacy of the GA method. Ad the added advatages of proposed method cludes easy corporato of real tme costras o the system lke tme varyg loads, dfferet type of uts etc. to effectvely apply for real tme operato of a system. Table 6: Comparso of results of EEE 33-bus system by proposed method ad other exstg method Case Bus locatos szes(mw) Total Sze(Mw) savg(kw) M AM M AM M AM Table 7: Comparso of Voltage mprovemet by proposed method ad other exstg method Case No. M Voltage % mprovemet M AM M AM Base case case case case case CONCLUSONS ths paper, a two-stage methodology of fdg the optmal locatos ad szes of s for maxmum loss reducto of radal dstrbuto systems s preseted. A sgle placemet algorthm s proposed to fd the optmal locatos ad a CGA method s proposed to fd the optmal szes. Voltage ad le loadg costrats are cluded the algorthm. Ths methodology s tested o EEE 33 bus system. By stallg s at all the potetal locatos, the total power loss of the system has bee reduced drastcally ad the voltage profle of the system s also mproved. cluso of the real tme costras 5

7 teratoal Joural of Computer Applcatos ( ) Volume 5 No., July 0 such as tme varyg loads ad dfferet types of uts ad dscrete ut szes to the proposed algorthm s the future scope of ths work. 7. EFEENCES [] G. Cell ad F. lo, Optmal dstrbuted geerato allocato MV dstrbuto etworks, ower dustry Computer Applcatos, 00. ca 00. ovatve Computg For ower - Electrc Eergy Meets The Market. d EEE ower Egeerg Socety teratoal Coferece, May 00, pp [].A. Daly, J. Morrso, Uderstadg the potetal beefts of dstrbuted geerato o power delvery systems, ural Electrc ower Coferece, May 00, pp. A 3. [3]. Chradeja,. amakumar, A approach to quatfy the techcal beefts of dstrbuted geerato EEE Tras Eergy Coverso, vol. 9, o. 4, pp , 004. [4] Kyoto rotocol to the Uted Natos Framework Coveto o clmate chage. [5].E. Brow, J. a, X. Feg, ad K. Koutlev, Stg dstrbuted geerato to defer T&D expaso, roc. EE. Ge, Tras ad Dst, vol., pp. 5-59, 997. [6] E. Daz-Dorado, J. Cdras, E. Mguez, Applcato of evolutoary algorthms for the plag of urba dstrbuto etworks of medum voltage, EEE Tras. ower Systems, vol. 7, o. 3, pp , Aug 00. [7] M. Mardaeh, G. B. Gharehpeta, Stg ad szg of uts usg GA ad OF based techque, TENCON. EEE ego 0 Coferece, vol. 3, pp , -4, Nov.004. [8] Slvestr A.Berzz, S. Buoao, Dstrbuted geerato plag usg geetc algorthms Electrc ower Egeerg, ower Tech Budapest 99, ter. Coferece, pp.57, 999. [9] Naresh Acharya, ukar Mahat, N. Mthulaatha, A aalytcal approach for allocato prmary dstrbuto etwork, Electrc ower ad Eergy Systems, vol. 8, pp , 006 [0 G.Cell,E.Gha,S.Mocc ad F.lo, A mult objectve evolutoary algorthm for the szg ad sttg of dstrbuted geerato,eee Trasactos o power systems,vol.0,o.,pp ,may 005. [] G.Carpell,G.Cell, S.Mocc ad F.lo, Optmzato of embedded szg ad sttg by usg a double trade-off method, EE proceedg o geerato, trasmsso ad dstrbuto, vol.5,o.4, pp , 005. [] C.L.T.Borges ad D.M.Falcao, Optmal dstrbuted geerato allocato for relablty,losses ad voltage mprovemet, teratoal joural of power ad eergy systems,vol.8.o.6,pp.43-40,july 006. [3]M. E. Bara ad F. F. Wu, Network recofgurato dstrbuto systems for loss reducto ad load balacg, EEE Trasactos o ower Delvery, Vol. 4, No Apr 989, pp ,. [4]M.adma ltha,v.c.veera eddy, N.Usha Optmal placemet for maxmum loss reducto radal dstrbuto system teratoal joural of emergg techologes ad applcatos egeerg, techology ad sceces,july 009. [5]M.adma ltha,v.c.veera eddy, N.Usha placemet usg Fuzzy for maxmum loss reducto radal dstrbuto system teratoal joural of computers applcatos egeerg, techology ad sceces, vol.pp 50-56,Oct009, M. adma ltha. s awarded h.d for her work Soft Computg Techques for optmal placemet adal Dstrbuto System from S.V.U. the year 0 ad s a graduate from JNTU, aathapur Electrcal & Electrocs Egeerg the year 994. Obtaed ost graduate degree SOC from S.V.U, Trupath the year 00. Havg 4 years of experece teachg graduate ad post graduate level. Has teratoal joural publcatos to her credt. resetly workg as rofessor ad HOD of EEE departmet ATS, ajampet. Areas of terest clude radal dstrbuto systems, soft computg power systems, ANN. V.C.Veera eddy. resetly workg as HOD & professor EEE departmet of S.V.U. College of Egeerg, Trupath. He was awarded h.d for hs work Modelg & Cotrol of Load frequecy usg New Optmal Cotrol Strategy from S.V.U. the year 999. Have 8 years of experece teachg ad worked varous levels. Guded a umber of research scholars.atteded a umber of atoal & teratoal cofereces. More tha 30 joural publcatos are there for hs credt. Areas of terest clude Dstrbuto systems, Geetc Algorthms, Fuzzy Systems, ANN. N.Svaram eddy. resetly workg as a professor & HOD ME departmet of ATS, ajampet. He s pursug h.d. Have 8 years of experece teachg ad worked varous levels. Atteded a umber of atoal & teratoal cofereces. Areas of terest clude flexble maufacturg systems, Meta heurstc search methods, global optmzato techques. V.Usha eddy. s pursug h.d ad s a graduate from SVU, Trupath Electrcal & Electrocs Egeerg ad obtaed ost graduate degree SOC from S.V.U, Trupath. resetly workg as assstat professor EEE departmet of S.V.U. College of Egeerg, Trupath. Areas of terest clude Dstrbuto systems, Geetc Algorthms, Fuzzy Systems, ANN ad artfcal tellgece power systems, 6

OPTIMAL DG PLACEMENT FOR MAXIMUM LOSS REDUCTION IN RADIAL DISTRIBUTION SYSTEM USING ABC ALGORITHM

OPTIMAL DG PLACEMENT FOR MAXIMUM LOSS REDUCTION IN RADIAL DISTRIBUTION SYSTEM USING ABC ALGORITHM teratoal Joural of Revews Computg 009-00 JRC & LLS. All rghts reserved. JRC SSN: 076-338 www.jrc.org E-SSN: 076-3336 OPTMAL PLACEMENT FOR MAXMUM LOSS REDUCTON N RADAL DSTRBUTON SYSTEM USNG ABC ALGORTHM