Multi-objective Transmission Planning Paper

Transcription

1 Downoaded from orbt.dtu.dk on: Nov, 8 Mut-objectve Transmsson Pannng Paper Xu, Zhao; Dong, Zhao Yang; Wong, Kt Po; an, Zhun Pubshed n: APPEEC9 Lnk to artce, DOI:.9/APPEEC Pubcaton date: 9 Document Verson Pubsher's PD, aso known as Verson of record Lnk back to DTU Orbt Ctaton (APA): Xu, Z., Dong, Z. Y., Wong, K. P., & an, Z. (9). Mut-objectve Transmsson Pannng Paper. In APPEEC9 (pp. 88). IEEE. DOI:.9/APPEEC Genera rghts Copyrght and mora rghts for the pubcatons made accessbe n the pubc porta are retaned by the authors and/or other copyrght owners and t s a condton of accessng pubcatons that users recognse and abde by the ega requrements assocated wth these rghts. Users may downoad and prnt one copy of any pubcaton from the pubc porta for the purpose of prvate study or research. You may not further dstrbute the matera or use t for any proft-makng actvty or commerca gan You may freey dstrbute the URL dentfyng the pubcaton n the pubc porta If you beeve that ths document breaches copyrght pease contact us provdng detas, and we w remove access to the work mmedatey and nvestgate your cam.

2 Mut-objectve Transmsson Pannng Zhao Xu Centre for Eectrc Technoogy Department of Eectrca Engneerng Technca Unversty of Denmark Kgs. Lyngby, Denmark Zhao Yang Dong Department of Eectrca Engneerng The Hong Kong Poytechnc Unversty Hung Hom, Kowoon, Hong Kong Kt Po Wong Department of Eectrca Engneerng The Hong Kong Poytechnc Unversty Hung Hom, Kowoon, Hong Kong Zhun an Secton of Engneerng Desgn Department of Management Engneerng Technca Unversty of Denmark Kgs. Lyngby, Denmark Abstract Ths paper descrbes a transmsson expanson pannng method based on mut-objectve optmzaton (MOOP). The method starts wth constructng a canddate poo of feasbe expanson pans, foowed by seecton of the best canddates through MOOP, of whch mutpe objectves are tacked smutaneousy, amng at ntegratng the market operaton and pannng as one unfed process n the market envronment. Subsequenty, reabty assessment s performed to evauate and renforce the resutant expanson pan from MOOP. The proposed method has been tested wth the IEEE 4-bus system and anayses and dscussons have been presented. Keywords- Transmsson pannng, Mut-objectve optmzaton, Eectrcty market. I. INTRODUCTION The eectrcty devery system, ncudng transmsson and dstrbuton, pays an mportant roe n overa eectrc power system. To fuf the customer s ever growng demand, constructng new and renforcng exstng transmsson networks are of hgh mportance. Propery panned expanson w ensure the heathy operaton of the power systems thereby supportng the deveopment of economy and socety. In the past, eectrcty secton was thought naturay a monopoy. Eectrcty generaton, transmsson and dstrbuton were vertcay-ntegrated as one system. Expandng transmsson network was soey the responsbty of the power system operator, eventuay the government. As such, the major concern of transmsson pannng used to be the nvestment cost. The transmsson pannng probem, though dynamc n nature, s often smpfed as a statc optmzaton mode, mnmzng the tota nvestment of network expanson for a snge future scenaro, subject to a number of constrants [-]. In most terature, the statc transmsson network expanson pannng (STNEP) mode s typcay formuated to mnmze the sum of nvestment cost and the oad curtaments caused by ack of transmsson capacty, subject to DC or AC oad ow equaton [-8]. The resut s an expanson pan wth the east nvestment wth strcty mted oad curtament where necessary. The east-cost pannng mode presumes that the power system s vertcay ntegrated, owned, operated and panned by one operator. However, due to the market berazaton, ths s no onger the case n many countres, where the ntegrated power systems have now been spt nto sma, separated and compettve enttes. These ndependent companes provde unbunded servces of generaton, transmsson and dstrbuton, and compete aganst each other to pursue maxma nterests n an open envronment. In pannng the transmsson expanson, ther objectves woud become rather market orented, n whch both nvestment and proft prospects woud be concerned. On the other hand, the physca topoogy of transmsson network has natura mpacts on eectrcty market operaton whch shoud be addressed nto the pannng process as we. Obvousy, the tradtona east-cost pannng method does not refect these economc sgnas and s no onger approprate n ths context. Recognzng the nherent mtatons of the tradtona pannng approach, the authors prevousy deveoped a mutobjectve transmsson pannng mode [9]. The deveoped mode ntegrates the objectves of nvestment, transmsson proft, and market dspatchng nto one optmzaton approach, foowed by reabty assessment and optma renforcement to acheve the fna pan. In formuatng the canddate expanson pan, practca engneerng and management concerns have aso been consdered. In ths work, the mode has been further deveoped to enhance the souton methodoogy. A superor mut-objectve technque.e. the constrant method s apped n ths paper due to ts capabtes of fndng pareto-optma soutons even wth non-convex and dscrete objectve space []. II. OVERVIEW O THE PLANNING METHOD gure presents the structure of the proposed method, where preparaton stages for pannng study such as demand forecastng and reabty assessment are not ncuded. The method can be dvded nto three stages, ncudng the nta formuaton of the canddate poo for expanson, foowed by a mut-objectve optmzaton subject to constrants, and the fna securty assessment and necessary renforcement of the expanson pan from MOOP based on the N- crteron. A. ormuaton of canddate poo The nta canddate poo for expanson s formuated based on the nformaton of the gven system, such as the generaton and transmsson capacty, the estmated transmsson tarffs, the pannng scope and correspondng forecasted system demand etc. Human knowedge and emprca rues are apped n ths stage to ensure ratonaty of the canddate ne seecton wth practca engneerng and management concerns, whch may ncude but not mted to, network redundancy; envronmenta factors; fnanca constrants; estmated constructon perods and the pannng horzon etc. The formuaton of canddate poo can start from determnng the pannng horzon and correspondng market /9/$5. 9 IEEE Authorzed censed use mted to: Danmarks Teknske Informatonscenter. Downoaded on Apr, at 4:: UTC from IEEE Xpore. Restrctons appy.

3 forecasts. Subsequenty, market smuaton can be carred to dentfy the transmsson bottenecks. The poo s then formuated to enhance transmsson wth redundancy for system reabty. Next, by examnng nvestments and constructon perods of ndvdua nes, a number of canddates can be dropped off at ths stage to avod excessve workoad n foowng steps. By takng nto account the envronmenta factors, further reducton of the poo can be made, for exampe, to avod a ne constructon through an mportant natura reserve. ncomes happen at the md of the pannng year (). gure ustrates the fnanca fow n the pannng horzon. The objectves of pannng are gven n Equatons (-7). Mnmze, h = /( u ) C T + () = f = [ ( ( a N G P G ) /( + b P + ) + c P )] /( + ) () mze, = f ( f,, r, d = [ f r d ) /( (, j ) + ) ] /( + ) + () Subject to, f - x + x )( - ) = (4) ( j f f - x x ; = (5) x P Mn P P (6) gure. owchart of the hybrd pannng scheme for TNEP - nvestm ent h year Iustraton of fnanca fows n pannng gure. Iustraton of fnanca fows n pannng B. Expanson Pan Seecton by MOOP Ths step w seect the best expanson pan from the formuated canddate poo by MOOP. To mathematcay mode the expanson pan, an array s used to represent the canddate nes n each transmsson corrdor [,7,9]. The proposed pannng mode ncude severa objectves to maxmze the beneft of expanson (BOE),.e. the transmsson servce ncome, whe mnmzng the overa nvestment, subject to power fow and securty constrants etc. urthermore, mnmzaton of the generaton cost s ncuded snce the eectrcty market shoud be dspatched at the east generaton cost gve a System Margna Prce (SMP) aucton based market []. To account for the tme vaue of money, ndvdua objectve s converted to ts Net Present Vaue (NPV), assumng the nvestment happens at the begnnng of the pannng year (h) and new crcuts are mmedatey ready to serve []. It aso assumes the transmsson and generaton x = r ; (, j) (7) where and are the tota nvestment of expanson and the generaton cost of the system respectvey, and s the expected proft from the expanded transmsson nes, based on the MW-Me method, n whch the transmsson cost s computed by mutpyng the actua power fow, the dstance and the pre-determned unt cost of each transmsson ne []. C s constructon cost vector of new transmsson nes; s an nteger representng the actua new crcut addton n transmsson route -j (bus -j) whe s ts upper bound; s the dscount rate and h s the pannng year; f, d and r are the rea power fow, ength and pre-determned transmsson tarff per unt capacty n $/MW-Me respectvey for the transmsson route -j; r s the crcut susceptance of added nes; x and x are the nta and tota new susceptance n branch j; P s rea power outputs of generator I, and a, b and c are the coeffcents of the cost functon; s the set of canddate transmsson nes for expanson seected at nta stage of pannng; The MW-Me prcng method aocates the costs based on actua system usage as cosey as possbe, whch s superor to other methods, such as post-stamp and contract path methods []. Constrants (4)-(6) are smar to those n DC power fow, whe Equaton (7) smpy updates the crcut susceptance due to expanson. Compared to the mode n [9], the use of DC fow constrants effectvey reduces the computatona tme. Authorzed censed use mted to: Danmarks Teknske Informatonscenter. Downoaded on Apr, at 4:: UTC from IEEE Xpore. Restrctons appy.

4 The transmsson proft n the pannng mode can be consdered as the dfference between transmsson customer s payment and servce provder s cost n future transactons, whch measures the soca beneft from the transmsson deas. In the marketpace, maxmzaton of soca benefts s the frst prorty and shoud be refected nto the pannng stage. By dong so, market operaton and pannng can be ntegrated nto one unfed process. The nvestment of expanson s another part of the costs whch may not be fuy recovered by the transmsson proft. Thus the pannng objectve s formuated to maxmze the beneft of expanson, namey the proft, whe mnmzng the pannng costs. Moreover, t woud be reasonabe to assess transmsson proft based on reastc market dspatchng, and therefore a east cost dspatchng objectve s ncuded n the pannng mode. The proposed pannng method actuay seeks to maxmze the overa soca beneft from expanson by ncorporatng varous factors of nterests n horzon nto one optmzaton approach and tackng mutpe objectves smutaneousy. It shoud be noted that the proposed approach s one step further to current practce n transmsson pannng n countres where compettve market has been estabshed. or exampe, n Austraa Natona Eectrcty Market, transmsson network servce provders (TNSPs) need to meet reabty or market beneft crtera set by the Natona Eectrcty Rue n order to be funded for system augmentaton. Bascay, optons anayss to compare the NPV of cost and beneft of dfferent expanson optons need to be performed n order to seect the most preferred expanson opton, and then subject to (mosty) reabty test and/or market beneft test, [5]. C. Mut-Objectve Optmzaton The proposed pannng mode actuay formuates a mutobjectve optmzaton probem wth ncompatbe objectves, subject to a st of constrants. One possbe souton s to drecty combne a objectves nto one by usng the weghted sum method [], whch suffers from the dffcuty n fndng the Pareto-optma souton,.e. the genune optma souton for a objectves, though wth great smpcty. In our prevous work, the goa programmng method s used n sovng the mode. In ths paper, further deveopment of the pannng method has been made to enhance the souton methodoogy. The pannng mode s to be soved usng a superor MOOP technque,.e. the -constrant method, whch dstngushes tsef by ts capabtes of fndng pareto-optma soutons even wth non-convex and dscrete objectve space. The method retans one of the probem objectves, whe restrctng the rest wthn user-defned vaues. In ths appcaton, objectve s retaned and the pannng mode resuts n a typca nonnear snge-objectve optmzaton probem, whch can then be expressed as foows []: mze ( x) (8) Subject to ( x) ε (9) ( x) ε () G ( x) = () H mn H x) ( H () where Eqn. (8) s the same objectve defned earer; and x s the souton vector of the optmzaton mode; objectve and max become soft constrants bounded by pre-defned ε and ε ; vector G denotes the equaty constrants (oad fow equatons) defned by Eqn. (); and vector H s the set of nequates defned by Eqn. () (). ε vectors are user-defned and represent upper bounds of the objectves and, and are usuay seected n the range of the mnmum and maxmum vaues of the objectves. The soutons obtaned from the ε- constrant method are very much dependabe on the vaues chosen for the converted soft constrans. Inapproprate ε vector may produce naccurate or erroneous resuts []. The Branch and Bound (BNB) method s used to sove the MOOP mode due to ts capabty of handng mxed nteger nonnear optmzaton probem, though BNB s mted n deang wth probems of hgh dmenson []. D. Securty Assessment The expanson pan obtaned from the MOOP stage may contan nsecure confguratons. To ensure the system reabty, the expanson pan s assessed by a st of credbe snge ne outages,.e. the N- crtera, usng a base case AC oad fow. Shoud overoad happen n any transmsson ne, best ndvduas from the rest of the canddate poo are seected to renforce the network. The securty assessment s repeated unt no overoadng happens n the system and the optma expanson pan s fnazed. Three steps of the securty assessment nvove contngency screenng, N- checkng, and seecton of extra ne for renforcement. Contngency screenng s based on the Performance Index (PI) []. To avod maskng effect [4], the seecton s not stopped unt no overoad happens for two contnuous outages [, 4]. N- check s conducted by removng snge ne accordng to the contngency st and computng the overoad each tme. The sum of ne overoads n a outages s thus determned. nay, the best nes to enhance reabty are seected from the rest of the canddate poo based on a cost-beneft anayss, namey the overoad reeve per unt ne constructon cost as defned beow, where P Oc = () C P = OVL - OVL (4) B and OVL s overoad n transmsson nes (subscrpts B and A stands for before and after); Oc s overoad reeve per unt ne constructon cost. Combnatory effect of addng more than one ne s consdered n the equaton above. Obvousy, the nes wth arger Oc have the prortes for renforcement. When OVL B, tota overoad before new ne addton, s zero, the expanded system s consdered to be N- reabe and the securty assessment s termnated wthout renforcements. III. CASE STUDY The IEEE 5-machne, 4-bus system s used n our case study. The tota demand for the base year s the 59 MW whe the forecasted demand for the pannng year (h=) s ncreased by tmes, of whch AC optma power fow proved to dverge, due to overoads happenng n transmsson nes. gure presents the demand s uneveny dstrbuted among the buses for the base and pannng year. Orgnay, the system A Authorzed censed use mted to: Danmarks Teknske Informatonscenter. Downoaded on Apr, at 4:: UTC from IEEE Xpore. Restrctons appy.

5 has a tota of transmsson crcuts. These crcuts have dfferent characterstcs such as dstance and proft rate of transmsson. Snce those parameters are not wth the orgna 4 bus system, we have carefuy desgned the data for our case study, whch does not necessary resembe the actua system. The pannng program s mpemented n MATLAB. It starts wth seecton of the canddate crcuts. The vectors n the MOOP mode are found by performng mnmzatons for ndvdua objectves n advance. rom the MOOP procedure, the nta expanson pans are generated n Tabe I, where the canddate routes and are seected. Subsequenty, the securty assessment (SA) s carred out and two extra nes are added to enhance the reabty, and the resutant nvestment s ncreased from 5.4 to 8.75 M$, as a compromse between economy and securty. (absoute vaue) n transmsson routes before and after expanson. The capacty mt of each route s aso presented for before and after the expanson. It s shown that power fow dstrbuton has been argey changed due to network expanson, whch benefts the system wth mproved securty margn of transmsson capacty. Partcuary, the fow n route has been argey ncreased over the orgna mtaton after the expanson ndcatng the necessty of ts expanson. IV. CONCLUSION Due to market berazaton, pannng power system expanson has to take nto account mutpe objectves to dea wth new concerns and uncertantes that have appeared. Based on our prevous work, ths paper has further deveoped the mut-objectve pannng mode by appyng a superor constraned MOOP technque and a smpfed DC oad fow constrants. To resove the mtaton of BNB method and mprove computaton speed, future research w focus on further smpfcaton of the deveoped pannng mode, as we as appcaton of advanced computatona ntegence n sovng the mode. REERENCES Power MW gure. gure Dstrbuton of demand among buses for base and pannng years TABLE I. SIMULATION RESULTS IN CASE STUDY Route expanson Before SA After SA Tota nvestment(m$) Tota capacty added (MW) Base na Branch gure 4. Power fows of transmsson routes To further evauate the expanson resut, optma power fows have been run for the network before and after expanson, of whch the dspatch shoud generate smar resuts as a SMP market. gure 4 compares the resutant rea power fows [] Ahmed H. E-Abad edted, Power systems anayss and pannng, Proceedngs of the Arab Schoo on Scence and Technoogy, Wnter Sesson hed n Kuwat, January -9, 98. [] X.. Wang, undamentas of Power System Pannng, Chna Eectrc Power Press, 99. [] H. A. and E. L. da Sva, A reabe approach for sovng the transmsson network expanson pannng probem usng genetc agorthm, Eectrc Power System Research ,. [4] M. V.. Perera and L. M. V. G. Pnto, Appcaton of senstvty anayss of oad suppyng capacty to nteractve transmsson expanson pannng, IEEE Trans. Power App. Syst., vo. PAS-4, pp. 8 89, eb [5] R. Romero and A. Montce, A herarchca decomposton approach for transmsson network expanson pannng, IEEE Trans. Power Syst., vo. 9, pp. 7 8, eb [6] R. Romero, R. A. Gaego and A. Montce, Transmsson system expanson pannng by smuated anneang, IEEE Trans. Power Syst.,vo., pp , eb [7] E. L. Sva, H. A. and J. M. Areza, Transmsson network expanson pannng under an mproved genetc agorthm, IEEE Trans. Power Syst., vo. 5, pp , Aug.. [8] R. A. Gaego, R. Romero and A. J. Montce, Tabu search agorthm for network synthess, IEEE Trans. Power Syst., vo. 5, pp , May. [9] Z. Xu, Z. Y. Dong and K. P. Wong, A hybrd Pannng Method for Transmsson Network n a Dereguated Envronment, IEEE Transactons on Power Systems, vo:, ssue:, 6. [] K. Deb, Mut-Objectve Optmzaton usng Evoutonary Agorthm, John Wey & Sons, Ltd,. [] K. Bhattacharya, M. H. J. Boen and J. E. Daader, Operaton of Reconstructed Power System, Kuwer Academc Pubshers,. [] C.A. oudas, Nonnear and Mxed Integer Optmzaton: undamentas and Appcatons, Oxford Unversty Press, 995. [] T. A. Mkoynnas and B.. Woenberg, An advanced contngency seecton agorthm, IEEE Trans. on Power App. and Syst., vo PAS-, No. 98. [4] K. Schafer and J. Verstege, Adaptve procedure for maskng effect compensaton n contngency seecton agorthms, IEEE Trans. on Power Systems, vo. 5, ss., May 99, pp [5] Austraan Energy Market Commsson (AEMC), Natona Eectrcty Rues, verson, avaabe on ne at: Authorzed censed use mted to: Danmarks Teknske Informatonscenter. Downoaded on Apr, at 4:: UTC from IEEE Xpore. Restrctons appy.