int rt CIC b 24 th International Conference on Electricity Distribution Glasgow, June 2017 Paper 1346

Transcription

1 24 th Internatonal Conference on Electrcty Dstruton Glasgow, June 2017 SIMULTAEOUS OPTIMIZATIO OF TIE SWITCHES PLACEMET AD RESERVE CAPACITY MARGI OF SUB-TRASMISSIO SUBATIOS COSIDERIG THE COFLT BETWEE SHORT-TERM AD LOG-TERM PLAIG Mlad HOSEIPOUR Mahmoud-Reza HAGHIFAM Mansoreh ZAGIABADI Tarat Modares Unversty Iran Tarat Modares Unversty Iran ewcastle Unversty - UK m.hosenpour@modares.ac.r haghfam@modares.ac.r Mansoureh.zangaad@newcastle.ac.uk Sayed Mad MIRI LARIMI Tarat Modares Unversty Iran m.mr@modares.ac.r ABRACT Outage management has a crucal role n mprovng relalty level of dstruton networks, and t s profoundly dependent on the reserve acty margn of Su-Transmsson () Sustatons and load transferrng alty etween them. These factors enale DSOs to restore nterrupted load ponts n out of servce areas. In ths paper, the connecton of reserve acty n sustatons and load transferrng alty, made y te swtches, s studed. The mentoned correlaton s addressed va Costumer Interrupton Cost (C) derved from outage management n prmary dstruton networks. Then, a mathematcal model s developed n whch optmal reserve acty of sustatons and optmal placement of te swtches are determned smultaneously. The resulted optmzaton model s formulated as a mxed-nteger non-lnear prolem (MILP), and the genetc algorthm (GA) s used to solve the prolem. Fnally the effectveness of the proposed model s evaluated n a test dstruton network. Keywords: sustaton plannng, reserve acty, outage management, load transferrng alty, te swtch, OMECLATURE 1) Indces: σ n S Index of sustatons. Index of canddate te ponts. Index of network ranches. Index of nterrupted load ponts. Index of network nodes. Index of feeder sectons. 2) Sets: Ω Set of sustatons. ʌ Set of network ranches. Set of network nodes. Set of control varales. dlp, Set of nterrupted load ponts due to fault n ranch. 3) Parameters: n umer of sustatons. T T umer of canddate te ponts. umer of network ranches. Installaton cost of a te swtch [$/swtch]. TCC l r, t nt rt d, d 4) Varales: C, Te lne creaton cost [$/Km]. Length of the te lne n canddate te pont [Km]. Falure rate of ranch [fal/year]. Repar tme of ranch [hour]. Interrupton cost n load pont durng fault n ranch [$/kw]. Study perod [year]. Interest rate. Demand of load pont whch could not e restored due to falure n ranch [kw]. Load demand of man transformer n sustaton [MVA]. Utlzaton factor of sustaton. Presence or asence of te swtch n canddate locaton Total acty of sustaton [MVA].. Capacty of vrtual transformer n sustaton [MVA]. Installaton cost of vrtual transformer n sustaton [$]. MCC Maneuver creaton cost [$]. C Customer nterrupton cost ecause of falure n ranch [$]. Transferred load to sustaton L, ITRODUCTIO ecause of falure n ranch [kw]. Today dstruton companes (dscos) are n a tme of changes and challenges. The man changes come wth dstruton networks leralzaton, modernzaton, automaton, and more effcent utlzaton of network assets. On the other sde, the most mportant challenge s the ncreasngly demand of customers for relalty regard to ncreasngly growth of crtcal loads such as dgtal loads. So dscos should adapt for the changes n CIRED /5

2 24 th Internatonal Conference on Electrcty Dstruton Glasgow, June 2017 such a way that s n lne wth challenges and manage ther assets nclned to ther oectve, n top of them network relalty. But there s a conflct etween mprovng the servce relalty and controllng of costs. In ths emergng envronment of changes and challenges, asset management plays a key role. Asset management sgnfes strategc plannng, mantenance, utlzaton, and operaton of a physcal resource. In addton, possle tme scopes of t contans real tme, short term, mdterm and long term. From a practcal prospectve, there are nteractons etween aforementoned tme scopes. Furthermore, t has known that the coordnaton of asset management strateges n dfferent tme scopes has a sgnfcant mpact n decson makng process of a dsco. A overall vew nto the promnent nteractons n asset management strateges of dstruton networks s gven n Fg1. Long term asset management Automato DG Determnng Reserve acty margn of sustatons Swtchng devces placement Md-term asset management Optmal mantenance strategy Optmal outage others Short term asset management DERs schedulng and market prce smulaton Control of Automatc Remote Real Tme asset management System montorng and contngency analyss to mnmze possle outages Onlne Outage Management and load Fg1. The promnent nteractons n asset management strateges of the dstruton networks Hgh proporton of customer s nterruptons are caused y falures n prmary dstruton networks [1]. Therefore, t s ratonal that specal attenton s pad to mprove the relalty of dstruton networks n asset management strateges [2]. One of the effcent methods for relevng customer s nterrupton duraton, and also mprovng relalty, s to provde reserve acty n sustatons and feeders [3, 4]. Wth ths addtonal acty n sustatons and y usng the alty of the normally closed and opened swtches n prmary dstruton networks, the DSO can restore out of servce areas from other routes [5]. The alty of te swtches for transferrng nterrupted load ponts to adacent sustatons s hghly dependent on avalale reserve acty. However, n the lteratures hgh share of works concern aout placement of te swtches and swtchng devces n relalty optmzaton prolems regardless of the correlaton of te swtches and the reserve acty of sustatons. In [6], to mprove the system relalty wth dstruted generaton under fault condtons, te-lnes placement schemes through graph-ased algorthms have een proposed. A mcro genetc algorthm, n conuncton wth the fuzzy logc, for modellng expert knowledge has een proposed as a soluton for optmal placement of sectonalzers and te swtches n [7]. An ant colony optmzaton method for placement of sectonalzng swtches n presence of dstruted generaton sources y usng a fuzzy mult oectve approach has een llustrated n [8]. Partcle swarm optmzaton algorthm was developed to determne optmum numer and locatons of sectonalzers and reakers [9]. In addton, optmum numer of swtches n a dstruton network was otaned y cost analyss on a system, and optmum locatons of swtches were found y smulated annealng algorthm [10]. As mentoned earler, studes whch relates to reserve acty margn of sustatons are rare regardless of vast studes n swtchng devces and load transferrng alty. For example, the acty of a sustaton was emprcally analysed ased on Sngle-Contngency Polcy (SCP), whch assumes that adacent sustatons should support one another when one of them s unavalale [11-13]. In addton to SCP for the determnaton of reserve acty n sustatons, the acty/load rato s a popular and useful ndex n sustatons acty plannng [14-15]. Despte low effcency and lack of accurate mathematcal valdaton, ths approach s practcal and has een wdely used y engneers. The aforementoned studes do not gve the optmal soluton ecause the supply alty of dstruton networks could not e effcently utlzed. Wth regard to growng tendency and satsfyng consequences derved from mplementaton of asset management actvtes n dscos, researchers mplement more effcent and accurate approaches for usng the supply altes of dstruton networks n whch multple sustatons are nterconnected wth normally open swtches [16]. Plus, n the prevous studes, real tme, short-term, mdterm, and long-term asset management actvtes were performed ndependently whereas coordnated plannng n dstruton networks reduces overall cost of dscos [3]. In ths regard, hghlghted oxes n Error! Reference source not found. ndcate the nteractons that are consdered n the present study. Due to these hghlghted oxes, the man contruton of ths study s the coordnaton etween long-term decsons, whch are the determnaton of sustatons reserve acty and te swtches placement, and real-tme decsons, whch s outage management usng te swtches. CIRED /5

3 24 th Internatonal Conference on Electrcty Dstruton Glasgow, June 2017 The rest of ths paper s organzed as follows: Secton 2 presents the proposed model of the prolem. Secton 3 dscusses formulatons and outage management process. Secton 4 provdes the codfcaton of the prolem. Secton 5 nvestgates the results of smulaton. Fnally n secton 6, the concluson of ths study s presented. DESCRIPTIO OF THE PROPOSED MODEL As mentoned earler, we propose a model that consders the nteracton etween real-tme and long-term asset management strateges of a dsco. In ths work, the reserve acty s modelled as a small ndvdual vrtual transformer n a sustaton apart from the acty of the man transformer whch was otaned from the forecasted demand. As dscussed n the prevous secton, the reserve acty of sustatons depends on the numer and locaton of te swtches. The correlaton etween the reserve acty and te swtches s nvestgated n fault condton through outage management strateges, and fnally the optmal decsons aout the reserve acty and te swtches placement are made. The conceptual framework of the proposed model s llustrated for a dstruton network wth two feeders and m canddate te swtches n Fg 2. Fg 2. Conceptual framework of the proposed model PROBLEM FORMULATIO The decson makng varales n ths paper are consdered as an :1 ( n ) n and array, whch T represent the numer of the sustatons and canddates for te swtches placement respectvely. The set of decson makng varales of ths paper ncludes contnuous varales :,1 n and nary varales :,( n 1) (n ). denotes the presence or asence (1: presence and 0: asence) of te swtch n canddate locaton. Also represents the utlzaton factor of sustaton. Wth respect to the utlzaton factor, the total acty of sustaton, whch conssts of the forecasted load demand of sustaton and the acty of vrtual transformer n t, can e derved as elow: d C,, 1 n (1) Hence the reserve acty or the acty of vrtual transformer can e computed ased upon Equatons (2): 1 d, 1 n (2) The mathematcal model resulted from these assumptons makes a Mxed Integer onlnear Programmng (MILP) prolem. In what presented n followng, formulatons are descred n detal. Investment cost The nvestment cost consst of two parts, the frst part relates to the nstallaton cost of the vrtual transformer wth acty of and can e formulated as:,1 f,,1,2,2 f,,2,3....,m 1 f,, m1, m The second part of the nvestment cost relates to Maneuver Creaton Cost (MCC) whch can e formulated as: MCC ( T l TCC) 1 Relalty cost For computng Customer Interrupton Cost (C) after fault creaton n each ranch, nterrupted load ponts are founded, then load restoraton s done. In load restoraton wth consderng nterrupted load ponts and the set of avalale te ponts all network confguratons whch can restore these load ponts are examned. If sum of the nterrupted load ponts connected to sustaton s hgher than, a load pont s dsconnected from end of the nterrupted feeder. Ths procedure s repeated untl non-delvered power s equal or lower than the acty of vrtual transformer and then forward-ackward sweep load flow s mplemented to check the voltage and current constrants n the new topology. If there s no constrants volaton n the new topology, the loads are (3) (4) CIRED /5

4 24 th Internatonal Conference on Electrcty Dstruton Glasgow, June 2017 restored, otherwse a load pont s dsconnected from end of nterrupted feeder and added to C. Customer Interrupton Cost durng fault n ranch of the network can e otaned from (5) [3].It should e noted that the swtchng tme s neglgle. dlp, C ( d )(1 nt ),, rt 1 Oectve functon The oectve functon of the proposed MILP prolem s to mnmze the total cost from the vewpont of a dstruton company, therefore the resultng optmzaton prolem s n mn ( ) (6) C T l TCC suect to A. Utlzaton factor 0 1, t1 (5) (7) UMERAL AALYSIS In ths secton, the proposed model for coordnate determnng of the optmal reserve acty of sustatons and the placement of te swtches s nvestgated. The proposed model s appled to a typcal 20-kV uran dstruton network (Fg 4). There are fve 63/20 kv sustatons and 82 MV/LV dstruton sustatons n the network. Servce area of each sustaton s separated y automatc normally open swtches. In addton, there are automatc normally close swtches n each secton etween two dstruton sustatons, and n order to avod complcaton of the network dagram, these swtches are not depcted n Fg 4. In ths test case, the desgner engneer consders 6 canddate te ponts. Also, the nput parameters are shown n Tale 1, and the nterrupton cost for all load ponts ncludng resdental, commercal, and ndustral are calculated ased on Customer Damage Functon (CDF) n [17]. B. Load transferrng constrant durng load restoraton ecause of fault n ranch : (8) L,, C. Voltage and current constrants after load restoraton: V V V n (9), mn n max I I max, (10) CODIFATIO The proposed optmzaton prolem s a MILP prolem whch s solved through GA. The codng strategy of ths paper s shown n Error! Reference source not found., n whch the proposed chromosome s composed of two strngs. The frst and second strngs represent the utlzaton factor of each sustaton and te ponts allocaton, respectvely. In the strng whch represents the utlzaton factors of sustatons, each gene contans a value etween 0 and 1, and n the strng representng te ponts allocaton each gene contans a nary value. Ths nary codng characterzes the canddate locatons for te swtches nstallaton n the network. The values 1 and 0 n gene represent exstng and not exstng of a te pont n the canddate locaton. Fg 3. Implemented codfcaton scheme Fg 4. Test case Tale 1.Input parameters TECHAL AD ECOOMAL IFORMATIO Parameter Unt Value t year $/MVA T $ / swtch $/km Case 1 s a ase case for assessng the overall network relalty status and how t s crucal to estalsh te ponts n the network. As mentoned, n Case 1 no te pont s avalale n the network for load restoraton. So, as we expected, the utlzaton factors of the sustatons derved from the model s equally 1. It means that there would e no reserve acty n the network to restore the nterrupted load ponts n emergency condtons. Case2 s dedcated to the man dea of ths paper whch s aout coordnated optmzaton of te ponts placement and the reserve margn of sustatons. In Tale 2 the optmal te ponts for the proposed network, wth regard to the nteracton of the reserve margn and te ponts, are found, whch are T1, T3, T4, and T5. Among the sustatons, sustaton 4 has the maxmum numer of TCC CIRED /5

5 24 th Internatonal Conference on Electrcty Dstruton Glasgow, June 2017 te ponts and the maxmum reserve acty nterpretng the sgnfcant C of feeders connected to the sustaton 4 through the te ponts. As shown n Error! Reference source not found., no reserve acty s allocated to sustaton 2, however ths sustaton s connected through T1 to the sustaton 1. Ths oservaton shows that f any fault occurs n feeder 1, the nterrupted load ponts could not transfer to sustaton 2, and the nterrupted load ponts are restored through T4 from the sustaton 4. Then, the motvaton of the exstence of T1 s only for restoraton of the nterrupted load ponts n feeder 2 under fault condton; n other word, sustaton 1 s a ackup for sustaton 2, whle the sustaton 2 s not a ackup for t. Tale 2. Results of case 2 A comparson s made etween the case 2 and case1 (Fg 5), whch s assumed there s no te pont n the network. In case 2 leadng to optmal soluton, a remarkale plunge s oserved n C and therefore n the total cost, ecause of the te ponts whch are placed wth regardng to the connecton of load transferrng alty and the reserve acty margn of sustaton. Fg 5.Comparson etween case 1 and 2 COCLUSIO Ths paper focuses on the correlaton of the reserve acty margn of sustatons and te ponts placement, and ased on ths correlaton t proposes a model for coordnated optmzaton of the reserve acty n sustatons and network te ponts. The hgh value of C n the frst case reveals the adverse consequence of the lack of te ponts n the network. It s shown that a small nvestment n te ponts creaton and the reserve acty makes a decsve plunge n C, and the optmal soluton s otaned y trade-off etween C and nvestment cost. REFERECES [1]. Brown, R.E., Electrc Power Dstruton Relalty, (CRC press, 2008) [2]. Cell, G. and Plo, F., 'Optmal Sectonalzng Swtches Allocaton n Dstruton etworks', Power Delvery, IEEE Transactons on, 1999, 14, (3), pp [3]. Saleh, J. and Haghfam, M.R., 'Determnng the Optmal Reserve Capacty Margn of Su Transmsson (St) Sustatons Usng Genetc Algorthm', Internatonal Transactons on Electrcal Energy Systems, 2014, 24, (4), pp [4]. Lakerv, E. and Holmes, E.J., Electrcty Dstruton etwork Desgn, (IET, 1995) [5]. Haghfam, M.R., 'Optmal Allocaton of Te Ponts n Radal Dstruton Systems Usng a Genetc Algorthm', European transactons on electrcal power, 2004, 14, (2), pp [6]. Mao, Y. and Mu, K.., 'Swtch Placement to Improve System Relalty for Radal Dstruton Systems wth Dstruted Generaton', Power Systems, IEEE Transactons on, 2003, 18, (4), pp [7]. do ascmento Alves, H., 'A Hyrd Applcaton Based on Expert Knowledge for Solvng the Swtches Placement Prolem n Dstruton etworks', Journal of Control, Automaton and Electrcal Systems, 2014, 25, (2), pp [8]. Morad, A. and Fotuh-Fruzaad, M., 'Optmal Swtch Placement n Dstruton Systems Usng Trnary Partcle Swarm Optmzaton Algorthm', Power Delvery, IEEE Transactons on, 2008, 23, (1), pp [9]. Falagh, H., Haghfam, M.-R., and Sngh, C., 'Ant Colony Optmzaton-Based Method for Placement of Sectonalzng Swtches n Dstruton etworks Usng a Fuzzy Multoectve Approach', Power Delvery, IEEE Transactons on, 2009, 24, (1), pp [10]. Bllnton, R. and Jonnavthula, S., 'Optmal Swtchng Devce Placement n Radal Dstruton Systems', Power Delvery, IEEE Transactons on, 1996, 11, (3), pp [11]. Luo, F., Wang, C., Xao, J., and Ge, S., 'Rapd Evaluaton Method for Power Supply Capalty of Uran Dstruton System Based on 1 Contngency Analyss of Man- Transformers', Internatonal Journal of Electrcal Power & Energy Systems, 2010, 32, (10), pp [12]. Wang, J., Carvalho, P., and Krtley, J., 'Emergency Reconfguraton and Dstruton System Plannng under the Sngle-Contngency Polcy', n, Innovatve Smart Grd Technologes (ISGT), 2012 IEEE PES, (IEEE, 2012) [13]. Xao, J., L, X., Gu, W., L, F., and Wang, C., 'Model of Dstruton System Total Supply Capalty Consderng Feeder and Sustaton Transformer Contngences', Internatonal Journal of Electrcal Power & Energy Systems, 2015, 65, pp [14]. Peng, W. and Pan, J., 'Rural Electrfcaton n Chna: Hstory and Insttuton', Chna & World Economy, 2006, 14, (1), pp [15]. Lu, S.-g., Cu, Y.-y., and Zhang, Z.-p., 'Research on Ratonal Capacty-Load Rato of Uran Electrc etwork', n, Electrcty Dstruton (CED), 2010 Chna Internatonal Conference on, (IEEE, 2010) [16]. Xao, J., L, F., Gu, W., Wang, C., and Zhang, P., 'Total Supply Capalty and Its Extended Indces for Dstruton Systems: Defnton, Model Calculaton and Applcatons', IET generaton, transmsson & dstruton, 2011, 5, (8), pp Systems, IEEE Transactons on, 2009, 24, (2), pp [17]. Chowdhury, A. and Koval, D., Power Dstruton System Relalty: Practcal Methods and Applcatons, (John Wley & Sons, 2011) CIRED /5