Combined Heat and Power Unit Commitment with Smart Parking Lots of Plug-in Electric Vehicles

Transcription

1 Combined Hea and Power Uni Commimen wih Smar Parking Los of Plug-in Elecric Vehicles Hamidreza Sadeghian, Zhifang Wang Deparmen of Elecrical and Compuer Engineering Virginia Commonwealh Universiy, Richmond, VA, USA {sadeghianh}, Absrac Vehicle-o-grid (V2G) echnology has drawn grea ineres in he recen years and is efficiency depends on scheduling of charging process of plug-in elecric vehicles (PEVs) as small porable power plans in smar parking los. On he oher hand, acive shif from cenralized o decenralized power generaion and environmenal concerns have caused an increase in he uilizaion of combined hea and power (CHP) unis in power sysems. The goal of his sudy is o develop and simulae a novel approach for combined hea and power uni commimen wih PEVs for cos reducion in elecric power sysem. A schedule for charging and discharging processes of PEVs wih respec o load curve variaions is proposed in his paper. A modified es sysem consising of convenional TG (hermal power generaing) unis, CHP unis, and PEVs is employed o invesigae he impacs of PEVs on generaion scheduling. For he modified es sysem, resuls obained are encouraging and indicae boh he feasibiliy of he proposed echnique and is effeciveness on generaion scheduling. Index Terms Uni commimen, Combined hea and power, Smar grid, Plug-in elecric vehicle, Benders decomposiion I. ITRODUCTIO Smar grid incorporaes a communicaion infrasrucure ha enables sysem componens o exchange informaion and commands securely and reliably. By providing such real ime informaion, independen sysem operaors (ISOs) achieve he abiliy o manage generaing resources associaed wih he load demand concurrenly. Online supervision may cause o handle supply demand equilibrium in a real ime fashion and beer inegraion of renewable energies associaed wih elecriciy sorages [1, 2]. PEVs as porable elecriciy sorages are poenially no only environmenally friendly and quie bu also cos-effecive in erms of operaing coss and energy prices compared o convenional TG unis. PEVs can decrease dependencies on small expensive unis by discharging a peak hours hrough paricipaing in vehicle o grid (V2G) service which leads o operaing coss reducion [3]. oreover oher applicaions of PEVs include flaening load curve and imizing loadabiliy of he nework by discharging PEVs in peak ime and charging during off-peak periods [4]. Oher han reducing emission and miigaing ransmission line congesion, PEVs have abiliy for fas load racking due o high availabiliy of parking los and local managemen of generaion subec o placing parking los near he load ceners [5]. In addiion, o reduce dependency on exernal energy supplies and miigae climae change, many counries have adoped policies o increase boh energy conservaion and he share of renewable energy resources [6]. The benefi of renewable energy aggregaion has been demonsraed in a number of sudies [7]. oreover, in some counries and regions e.g. in he EU such policies include increasing he share of combined hea and power (CHP) [8]. CHP or cogeneraion is he producion of elecriciy and hermal energy in a single, inegraed sysem. I is a valuable energy producion echnology ha can yield much higher oal energy efficiency han separae hea and power generaion. The fuel efficiency of CHP producion uni can be as much as 90%. In fac high efficiency of CHP sysems is one of he maor facors ha make hem aracive for invesors [9]. Several sudies are repored on impacs of PEVs on generaion scheduling problem. Auhors in [10] developed a uni commimen model for Texas elecric power sysem considering plug-in hybrid elecric vehicle flee o measure he poenial cos savings. An inelligen uni commimen associaed wih vehicle o grid has been sudied based upon imizing operaing coss in addiion o environmenal emission exernaliies [11]. In [12], a charging and discharging schedule of PEVs wih respec o load curve variaions wih hermal generaion scheduling is proposed. Lieraure review reveals a gap for sudying he impacs of CHP sysems and PEVs inegraion in convenional TG unis and scheduling. In his paper, a bridge beween PEVs and CHP uni commimen problem has been made and he impacs of parking los peneraion on combined hea and power generaion scheduling in smar grids have been invesigaed. The conribuions of his sudy can be summarized as follows: 1) A new srucure of combined hea and power uni commimen incorporaing PEVs, so-called CHPUC-PEV was proposed. 2) A modified double Benders decomposiion mehod was uilized o solve he proposed opimizaion problem. 3) The performance and effeciveness of he proposed approach are evaluaed wih numerical simulaions. II. A. omenclaure CHPUC-PEV FORULATIO F(P i,, H i,) fuel consumpion funcion of ih uni a ime HD oal sysem hea demand a ime H imum hea oupu of ih generaor (W) i

2 H i H i, i cov hea dsch, ch, dsch dsch imum hea oupu of ih generaor (W) hea level of ih generaor a h hour (W) index for generaor uni index for smar parking lo oal number of smar parking los oal number of all generaing unis oal number of convenional TG unis oal number of hea only unis # of conneced discharging vehicles o he grid a hour a smar parking lo # of conneced charging vehicles o he grid a hour a smar parking lo oal vehicles in he sysem imum # of discharging vehicles a hour imum # of discharging vehicles a hour ch imum # of charging vehicles a hour imum # of charging vehicles a hour PD oal sysem power demand a ime imum power oupu of ih uni (W) ch P i P i P i, π P PEV, pv RD SD i, SH i ST i SU i, T T i down down i, imum power oupu of ih uni (W) power level of ih generaor a h hour (W) V2G cos coefficien for smar parking lo available PEV power for V2G in parking lo capaciy of each vehicle for smar parking lo oal sysem reserve demand a ime shudown coss of hermal uni i a ime shudown cos of uni i($) sarup cos of uni i($) sarup coss of hermal uni i a ime dispach period in hours index for ime imum down-ime of ih generaor uni ime duraion for ih uni ha has been OFF a up T i imum up-ime of ih generaor uni up ime duraion for ih uni ha has been O a i, i. δ η commimen sae of ih uni a h hour sae of charge for smar parking lo oal efficiency for smar parking lo where SU i, and SD i, represen sarup and shudown coss of hermal uni i a ime, respecively, which are deered based on he following inequaliies: SUi, STi i, ( i, i, 1) (2) SDi, SHi i, 1 ( i, 1 i, ) (3) In Eq. (1), second par represens PEVs usage cos as small porable power plans. In his erm, P PEV, defines as available PEV power for V2G in smar parking lo which can be given by he Eq. (2). P, pv.. (4) PEV where pv, η, δ, represen he average capaciy of each baery, inverer efficiency parameer, and deparure sae of charge parameer for smar parking lo, respecively. In his paper we considered hourly scheduling of uni commimen and PEV charging/discharging. Therefore, he complexiies relaed wih fas dynamics are ignored and a simple saisic model is used insead which is sufficien for our sudy. The obecive funcion includes generaion cos of boh hermal unis and CHP unis. The fuel coss are presened as follows: Thermal unis CHP unis,.,., 2 i i i i i i i (5) F P a b P c P 2... F P a b. P c. P d. H i i, i i i, i i, i i, e H f P H i i, i i, i, 2 (6) where a i, b i, c i, d i, e i and f i are posiive fuel cos coefficiens of ih uni. Fig. 1 shows he hea-power feasible operaion region (FOR) of a CHP uni. I should be menioned ha CHP unis assumed o uilize waer vapor or gas urbines and boh of hem are modeled by means of a feasible operaion region designaed by ABCDEF ha links he elecric power generaion and hea producion [13]. Along he boundary curve BC, he hea capaciy increases as he power generaion decrease. Alhough in mos cases he fuel cos is a convex funcion, he feasible operaion region of advanced CHP unis is non-convex. From Fig. 1, in CHP sysems power generaion depends on he hea generaion and vice versa, which implies ha he producion of hea and power mus be planned in coordinaion wih each oher. B. Obecive funcion The obecive of CHPUC-PEV problem is imizing he oal operaion cos of unis over a scheduling period. Indeed, he CHPUC-PEV obecive funcion usually includes differen erms such as he sar-up cos, shudown cos and he fuel cos, while, he erm of PEVs energy coss has been also aken ino accoun in his manuscrip., P, H, dsch i1 1 T f F P, H. SU SD i i, i, i, i, i, T 1 1 dsch,. P. PEV, (1) Figure 1. Feasible operaion region for a CHP uni

3 C. Consrains The CHPUC-PEV involves consrains like power and hea balance, spinning reserve requiremen, imum up/down ime of a uni and limied number of PEVs. In addiion, a uni is o generae power wihin a given range. The formulaions of he consrains are given below: Power and hea balance: The power and hea generaed by all he commied unis wih PEVs power a a ime insan mus mee power and hea demand a ha ime insan, respecively. Pi, i, ( pv. dsch, pv. ch, ) PD (7) i1 1 Hi, i, HD (8) i1 Generaion limi: The generaion limi of a CHP uni is specified by he FOR as shown in Secion II.B and Fig.1. The upper and lower limis of he convenional TG unis and he hea only unis can be described by consrains (9) and (10) respecively as follows: P P P x, i 1,, (9) ma i i, i cov H H H (10),, 1,, hea inimum up and down imes: I is considered ha a uni mus be on/off for a imum ime before i can be shu down or resared, respecively: ( T )( ) 0 (11) up up i, i i, 1 i, ( T )( ) 0 (12) down down i, i i, i, 1 Spinning reserve: In order o fas response o compensae he deviaion beween real and predicive demand in power sysem, spinning reserve is required. ahemaically, spinning reserve requiremen a each hour is he oal amoun of imum capaciy of all synchronized unis us he oal generaing oupu in ha hour which can be given by he Eq. (13). Pi, i, ( pv. dsch, pv. ch, ) PD RD i1 1 (13) Charge/discharge limis: In supporing he daily use and have a reliable operaion, a cerain amoun of power should be ineced ino he PEVs baeries. Therefore, i is necessary o limi he accumulaed charging/discharging power. In addiion, limied number of PEVs should charge/discharge a he same ime over a predefined horizon. T dsch, (14) 1 1 dsch dsch, dsch (15) T ch, (16) 1 1 ch ch, ch (17) In his sudy, charging/discharging frequency is assumed once a day. Each vehicle should have a desired deparure sae of charge (SOC) level, while η is defined as inegraed efficiency for charging/discharging plus inverer. III. DOBBLE BEDERS DECOPOSITIO APPROACH As a mixed-ineger nonlinear opimizaion problem wih a large number of coninuous and ineger conrol variables, he CHPUC-PEV problem is non-convex due o CHP feasible operaion region. In his sudy, double Benders decomposiion (DBD) approach is used o solve he proposed CHPUC-PEV problem. In general, Benders decomposiion (BD) approach decomposes he original problem ino one maser problem and several sub-problems. By solving each sub-problem, a se of dual variables are obained and used o generae benders cus for he maser problem. The procedure o implemen he proposed DBD approach for solving CHPUC-PEV problem is shown in Fig. 2, where i is consis of wo BD algorihms, namely he ouer BD and he inner BD. For he ouer BD, he maser problem deeres he ineger variables (on/off sae of each generaing uni) and he sub-problem solves he economic dispach (ED) along wih charge/discharge scheduling problem. Figure 2. Sraegy of he proposed double Benders decomposiion approch As i menioned, for mos CHP unis, he hea producion capaciies depend on he power generaion and vice versa, which implies ha he producion of hea and power mus be planned in coordinaion wih each oher. Therefore, i presens a naural decomposiion scheme for he BD algorihm. For mos CHP sysems feasible operaion region is non-convex and i increases he complexiy o ge he global opimal soluion. However, decomposing hea and power variables wih BD approach resuls o a simple convexificaion procedure and hen we solve convex problems in boh maser problem and subproblem which i leads o find global opimum soluion [14, 15]. In he proposed approach in his sudy, for solving he ED problem (ouer sub-problem) anoher BD algorihm (inner BD) is used. For his BD algorihm, he variables represening he hea producion are solved for in he maser problem while he paricipaing PEVs in addiion o he ones represening he power producion are kep in he sub problem. ore deails of he proposed DBD approach can be found in [14]. The obecive funcion is decomposed o he obecive funcions of he maser problem and sub-problem. The maser problem in he proposed soluion approach has mixed ineger linear programg model wih he following obecive funcion T i. a i, SDi, SUi, 1i1 (18) where γ denoes he benders cus obained from dual variables and included consrains for he ouer maser problem are he

4 inequaliies given by Eqs. (2) and (3) as well as Eqs. (11) and (12). In he ouer sub-problem, he dispach decision variables, P i,, H i,, and dsch, are accouned. An ED allocaes power and hea generaion among commied unis based on a BD algorihm. The obecive funcion of inner sub-problem is as follows P, dsch subec o RF(, H, P, ) dsch, : x ( ) ( ) H H : h ( v) ( v) (19) where H (v) represens he value of hea producion vecor a v h ieraion of inner BD, (κ) is he value of unis saus vecor a κ h ieraion of ouer BD and λ is prefix indicaing dual variables. The addiional consrains of he inner sub-problem include he power demand inequaliy (Eq. (7)), power generaion limi (Eqs. (18)), imum spinning reserve consrain (Eq. (13)), and charge/discharging limis (Eqs. (14-17)). The oupu of his sub-problem deeres he values of paricipaing PEV, P (v) and he dual variable vecor associaed wih hose consrains ha fix he complicaing variables, H. The obecive funcion of inner maser problem solely as a funcion of he complicaing variables (H) for ime sep is given by subec o (20), ( ) ( ) ( ) ( ) ( ), RF(, P, H, dsch, ) h i, Hi, H i, i1 (21) where σ indicaes ieraion wih he highes value of normal Benders cu. Assumed consrains of he inner maser problem include he hea demand balance (Eq. (8)), and hea generaion limis (Eq. (10)). In his sudy, srong Benders cu is used which can enhance he convergence of he BD approach proposed [16]. If a sub-problem becomes infeasible, he srong Benders cu is added o he maser problem of he nex ieraion as well, however, despie of he previous case (feasible sub-problem), comparison among ieraions does no include he curren ieraion. Solving ED problem for each hour provides Benders cus (γ ) for ouer maser problem, ( ) ( ) ( ) ( ) v v v (, ( ) ( ) RF, P H, dsch, ) x i, i, i, i1 IV. (22) SIULATIO AD RESULTS The modified 11 uni sysem used in he simulaion of PEVs impac on generaion scheduling is based on daa presened in [14]. A oal number of PEVs aggregaed from muliple smar parking los in he sysem are considered he simulaions. Spinning reserve requiremen is assumed o be 10% of he hourly load demand in 24 hour scheduling ime period. Following parameers are assumed for PEVs: imum baery capaciy = 25 kwh, average baery capaciy = 15 kwh, imum baery capaciy = 10 kwh, charging/discharging frequency = 1 per day, oal efficiency (η) = 85%, and sae of charge (δ) = 50%. In his paper, hree differen scenarios are invesigaed (Table I). Firs scenario consiss of a ypical cos based uni commimen problem wih obligaion of hea and power demand. In he second scenario, an inegraion of 50,000 PEVs charged by renewable sources as he CHPUC-PEV problem is considered. I is assumed ha 50,000 PEVs are aggregaed in parking los managed by an aggregaor organizaion and supplied by renewable energies such as wind urbines and solar power, near he parking lo o avoid ransmission losses. The hird scenario sudies he inegraion of 50,000 PEVs charged/discharged via power grid wihou renewable sources in he CHPUC-PEV problem. I is worhy o menion ha he charged sae is when he PEVs are charged via he elecrical grid and he discharged sae is when baeries of PEVs are depleed o deliver heir sored energy o he elecrical grid. aximum number of charging/discharging vehicles a each hour for scenarios 2 and 3 is 10% and 20% of oal vehicles, respecively. TABLE I. SUARY OF SCEARIOS# 1-3. Scenarios Scenario definiion Scenario#1 Thermal and CHP unis wihou PEVs Scenario#2 Thermal and CHP unis wih PEVs and renewable energy Scenario#3 Thermal and CHP unis wih PEVs and wihou renewable energy A. Scenario 1 In his scenario, a modified 11 uni sysem consiss of eigh convenional TG unis, wo CHP unis and one hea-only uni is uilized. For 24 hours of a day, he oal cos obained $914, as shown in Table II. The bes resuls for he hree scenarios are compared wih each oher in his able. oe ha, he large difference beween he bes cos of he hird scenario and he second scenario is he resul of insalling addiional renewable generaion unis in in he second scenario. TABLE II. RESULTS OF PROPOSED ETHOD FOR DIFFERET SCEARIOS Scenarios Bes resul ($) Improvemen ($) Improvemen (%) Scenario#1 914, Scenario#2 904, , Scenario#3 911, , B. Scenario 2 In his case, we considered ha 10-unis (Thermal and CHP unis) would be employed o supply he grid elecriciy and in addiion, CHP unis plus one hea-only uni will mee he hea load demands for he es sysem. In addiion, PEVs are oally supplied by renewable sources. In his regard, he bes operaing cos equals o $904, By discharging PEVs during paricular hours and supplying adequae power, operaion hours of commied unis are decreased and consequenly oal operaion cos is improved. oreover, imum number of discharging PEVs for each hour is assumed as shown in Table III. The bes resuls of he second scenario is presened in Table II o enable comparison wih oher wo scenarios. I is worhwhile o noe ha in his sudy one smar parking lo is

5 considered, however, by considering muliple peneraion of PEVs parking los, higher improvemens will be achievable. PEVs smar parking los wih differen seings of PEVs operaions. In addiion, he inegraion of uncerain renewable energy generaion is also worh invesigaion. TABLE III. dsch IIU UBER OF DISCHARGIG PEVS. dsch dsch C. Scenario 3 dsch Figure. 3 illusraes he hea and power load curve of he es sysem, where i is divided ino hree differen periods based on he power load, i.e. valley period, off-peak period, and peak period. To reduce he operaion coss, PEVs are charged during off-peak periods and discharged during peak periods. The resul of uni commimen problem in presence of PEVs are presened in Tables IV and V in he appendix. Table IV shows he bes resul of he second scenario and he bes resuls of he hird scenario is shown in Table V. As i can be seen from he resuls, coss of all scenarios ha include PEVs in uni commimen problem are lower han convenional CHP uni commimen which shows he effeciveness of incorporaing PEVs in oal operaion cos reducion. I is worhwhile o noe ha, he main challenge of uni commimen is o properly schedule small expensive unis, as large cheap unis are always on. Operaors expec ha large cheap unis will mainly saisfy base load and oher small expensive unis will fulfill he peak loads. Using PEVs as small porable power plans reduce dependencies on small expensive unis. Figure 3. Hea and power load curve of he es sysem [12] V. COCLUSIO This paper presens a new approach o solve combined hea and power uni commimen problem in presence of plug-in elecric vehicles as small porable generaion unis in addiion o ypical generaion consrains in uni commimen problem. The numerical resuls illusrae he effeciveness of his approach on he sysem operaion cos reducion. In fac, PEVs no only eliae he need for small expensive unis in he power sysems, bu also provide addiional reserve capaciy and reliabiliy for he exising power sysems. The resuls of generaion scheduling wih he opimal charging/discharging scheme of PEVs suggess ha he oal operaion cos decreases noiceably wihou any change in he oal energy consumpion. However, fuure work will incorporae muliple peneraion of REFERECES [1] H. Lund and W. Kempon, Inegraion of renewable energy ino he ranspor and elecriciy secors hrough V2G, Energy Policy, vol. 36, no. 9, pp , [2] wasilu, Francis, e al. "Elecric vehicles and smar grid ineracion: A review on vehicle o grid and renewable energy sources inegraion." Renewable and Susainable Energy Reviews 34 (2014): [3] T. Karašnik, Energy conversion phenomena in plug-in hybridelecric vehicles, Energy Convers. anag., vol. 52, no. 7, pp , [4] Y. Cao, S. Tang, C. Li, P. Zhang, Y. Tan, Z. Zhang An Opimized EV Charging odel Considering TOU Price and SOC Curve, IEEE Trans.Smar Grid, vol. 3, no.1, pp ,ar [5] C. Guille and G. Gross, A concepual framework for he vehicleo-grid (V2G) implemenaion, Energy Policy, vol. 37, no. 11, pp , [6] Ahari,. H., and.. Ardehali. "Operaional performance of energy sorage as funcion of elecriciy prices for on-grid hybrid renewable energy sysem by opimized fuzzy logic conroller." Renewable Energy 85 (2016): [7] Zhao, Y., & Khazaei, H. (2016, July). An incenive compaible profi allocaion mechanism for renewable energy aggregaion. In Power and Energy Sociey General eeing (PESG), 2016 (pp. 1-5). IEEE. [8] oradi, H., e al. "Opporuniies o improve energy efficiency and reduce greenhouse gas emissions for a cogeneraion plan." Energy Conference and Exhibiion (EnergyCon), 2010 IEEE Inernaional. IEEE, [9] oradi, H., A. Abahi, and. Esfahanian. "Opimal energy managemen of a smar residenial combined hea, cooling and power." In. J. Tech. Phys. Probl. Eng 8 (2016): [10] R. Sioshansi and P. Denholm, The value of plug-in hybrid elecric vehicles as grid resources, Energy J., [11] A. Y. Saber and G. K. Venayagamoorhy, Inelligen uni commimen wih vehicle-o-grid A cos-emission opimizaion, J. Power Sources, vol. 195, no. 3, pp , [12] E. Talebizadeh,. Rashidinead, and A. Abdollahi, Evaluaion of plug-in elecric vehicles impac on cos-based uni commimen, J. Power Sources, vol. 248, pp , [13] Aghaei, Jamshid, e al. "Opimal robus uni commimen of CHP plans in elecriciy markes using informaion gap decision heory." IEEE Transacions on Smar Grid (2016). [14] H. R. Sadeghian and.. Ardehali, A novel approach for opimal economic dispach scheduling of inegraed combined hea and power sysems for imum economic profi and imum environmenal emissions based on Benders decomposiion, Energy, vol. 102, pp , [15] Bersekas, D. P. "Convexificaion procedures and decomposiion mehods for nonconvex opimizaion problems." Journal of Opimizaion Theory and Applicaions 29.2 (1979): [16]. Amady and. Reza Ansari, Hydrohermal uni commimen wih AC consrains by a new soluion mehod based on benders decomposiion, Energy Convers. anag., vol. 65, pp ,

6 APPEDI TABLE IV. SCHEDULE AD DISPATCH OF GEERATIG UITS FOR SCEARIO 2 (TOTAL OPERATIO COST = $904,819.22) Power (W) Hea (Wh) U1 U2 U3 U4 U5 U6 U7 U8 U9 U10 U9 U10 Boiler dschv2g Pv2g (W) ly cos ($) , , , , , , , , , , , , , , , , , , , , , , , ,926 TABLE V. SCHEDULE AD DISPATCH OF GEERATIG UITS FOR SCEARIO 3 (TOTAL OPERATIO COST = $911,488.60) Power (W) Hea (Wh) U1 U2 U3 U4 U5 U6 U7 U8 U9 U10 U9 U10 Boiler dsch Pv2g (W) ch Pg2v (W) ly cos ($)