Day-ahead tariffs for the alleviation of distribution grid congestion from electric vehicles

Transcription

1 Downloaded from orbt.dtu.dk on: Jul 25, 2018 Day-ahead tarffs for the allevaton of dstrbuton grd congeston from electrc vehcles O'Connell, Namh; Wu, Quwe; Østergaard, Jacob; Nelsen, Arne Hejde; Cha, Seung-Tae; Dng, Y Publshed n: Electrc Power Systems Research Lnk to artcle, DOI: /j.epsr Publcaton date: 2012 Document Verson Early verson, also known as pre-prnt Lnk back to DTU Orbt Ctaton (APA): O'Connell, N., Wu, Q., Østergaard, J., Nelsen, A. H., Cha, S-T., & Dng, Y. (2012). Day-ahead tarffs for the allevaton of dstrbuton grd congeston from electrc vehcles. Electrc Power Systems Research, 92, DOI: /j.epsr General rghts Copyrght and moral rghts for the publcatons made accessble n the publc portal are retaned by the authors and/or other copyrght owners and t s a condton of accessng publcatons that users recognse and abde by the legal requrements assocated wth these rghts. Users may download and prnt one copy of any publcaton from the publc portal for the purpose of prvate study or research. You may not further dstrbute the materal or use t for any proft-makng actvty or commercal gan You may freely dstrbute the URL dentfyng the publcaton n the publc portal If you beleve that ths document breaches copyrght please contact us provdng detals, and we wll remove access to the work mmedately and nvestgate your clam.

2 Day-ahead tarffs for the allevaton of dstrbuton grd congeston from electrc vehcles Namh O Connell, Quwe Wu*, Jacob Østergaard, Arne Hejde Nelsen, Seung Tae Cha, Y Dng Center for Electrc Technology, Department of Electrcal Engneerng, Techncal Unversty of Denmark ABSTRACT An economcally effcent day-ahead tarff (DT) s proposed wth the purpose of preventng the dstrbuton grd congeston resultng from electrc vehcle (EV) chargng scheduled on a dayahead bass. The DT concept developed heren s derved from the locatonal margnal prce (LMP), n partcular the congeston cost component of the LMP. A step-wse congeston management structure has been developed whereby the dstrbuton system operator (DSO) predcts congeston for the comng day and publshes DTs pror to the clearng of the day-ahead market. EV fleet operators (FOs) optmze ther EV chargng schedules wth respect to the predcted day-ahead prces and the publshed DTs, thereby avodng congeston whle stll mnmzng the chargng cost. A Dansh 400V dstrbuton network s used to carry out case studes to llustrate the effectveness of the developed concept for the preventon of dstrbuton grd congeston from EV chargng. The case study results show that the concept s successful n a number of stuatons, most notably a system over-load of 155% can be successfully allevated on the test dstrbuton network. Keywords: Day-ahead tarff, Congeston allevaton, Dstrbuton grd, Locatonal margnal prce, Electrc vehcle chargng management 1 Introducton Wth ncreasng levels of ntermttent renewable generaton penetraton n the power system, n a future when electrc vehcles (EVs) assumed a sgnfcant role, ther use to balance the *Correspondng author. Tel.: ; fax: E mal address: qw@elektro.dtu.dk (Q. Wu).

3 2 power fluctuatons ntroduced by ntermttent generaton would become an attractve opton. Incorporatng the transport sector nto the electrcty sector by replacng conventonal nternal combuston engne (ICE) vehcles wth EVs can help reduce green house gas (GHG) emssons from the transport sector. Denmark presents a unque opportunty for renewable energy utlzaton and EV deployment. The current wnd power penetraton level n Denmark s approxmately 20%. The Dansh government has set a target of 50% penetraton of wnd power by 2025 [1]. The average drvng dstance n Denmark s approxmately 40 km per day [2]. A fully charged 25 kwh battery can provde suffcent energy to meet daly drvng energy requrements n most cases. If addtonal energy s requred, EV batteres can be recharged durng the day, where the approprate chargng nfrastructure s n place. Therefore, from both the system balancng and end-user drvng perspectves, t s an nterestng opton to mplement the ntegraton EVs nto the power system. Nonetheless, the electrfcaton of transportaton wll lead to sgnfcant challenges for the power system. These challenges are expected to manfest at several levels of the power system, from ncreased generaton capacty requrements to congeston on dstrbuton networks. Each wll present unque condtons and solutons, and wll requre substantal efforts on the part of power system engneers and other system partcpants n the comng years to ensure contnued system relablty. Grd congeston results from excessve demand beyond the desgn lmtatons of the system. Congeston from EVs can be observed at the medum voltage (MV) level, as a number of studes demonstrate [3]-[4]. Many studes have been conducted analyzng congeston ssues on the MV network, however they also note that the problems lkely orgnate on the dstrbuton network, and as such, analyss of ths network should be conducted as the prmary stage of congeston studes [3], [5], [6]. The degree of grd congeston,.e. the percentage of overloadng, s dependent on a number of factors ncludng local grd ratng and topology, penetraton and dstrbuton of EVs, and chargng management procedures. Coordnated chargng appears to be an effectve means of allowng ncreased penetraton of EVs wthout volatng grd constrants. There s some

4 3 ncongruty on the optmal manner n whch to coordnate chargng, wth a number of dfferent objectves proposed, ncludng maxmzaton of EV penetraton [3], mnmzaton of losses [4], and mnmzaton of customer chargng costs [7]-[8]. The study conducted n [7] shows that the computatonal power requred to handle grd constrants at the dstrbuton level n the lnear programmng optmzaton of EV chargng management s qute sgnfcant. An alternatve method to prevent grd congeston s the ncluson of devces n chargers to detect voltage and halt chargng when the voltage at the pont of connecton drops below a set threshold. Alternatvely, the power factor could be adjusted to rectfy the voltage drop. These methods are mentoned n a number of papers [9]-[11], however, t s unclear how effectve these measures wll be at hgh levels of EV penetraton. There are many approaches to congeston management n transmsson systems, dependng on the electrcty market structure. The congeston management methods can be categorzed nto three groups: the Optmal Power Flow (OPF) based method, the prce area congeston control method and the transacton-based method [12]. The OPF based congeston management method s based on a centralzed optmzaton and s consdered to be the most accurate and effectve congeston management method. A prce-sgnal based congeston management scheme has been proposed to elmnate congeston by generaton reschedulng determned by congeston prces wthn an OPF framework [13]. A one-step congeston management method was proposed n [14] to manage congeston n hybrd markets. The proposed congeston management method ncorporates the support from reactve power manpulaton and the embedded cost of reactve power, and OPF s used to determne the congeston management scenaro. A new transmsson prcng scheme based on pont-to-pont tarffs for pool markets has been proposed based on pay as bd (PAB) to provde prce sgnals to promote the maxmum use of the exstng transmsson network, encourage approprate bddng behavors, and reduce market power and prce spkes [15]. A common theme n the lterature n the area of optmal EV chargng management s that of a commercal entty that oversees the chargng of EVs on behalf of the end-use prvate consumer; the termnology for such an entty vares, but t s henceforth referred to as a fleet operator (FO) n ths paper. FOs have been found to be partcularly useful for the allevaton of grd congeston,

5 4 as the geographcal and temporal dversty of ther portfolos can be exploted to reallocate chargng to locatons and tmes of low demands, whle stll respectng the energy requrements of each EV. In addton, as a commercal entty, ther ncentve for proft maxmzng behavor s larger than that for ndvdual consumers, resultng n a hgher elastcty of demand and effectve response to prce sgnals. In ths paper, a day-ahead grd tarff has been developed to prevent dstrbuton grd congeston from electrc vehcle (EV) chargng from a day-ahead plannng perspectve. The DT concept developed heren s derved from the locatonal margnal prce (LMP), n partcular the congeston cost component of the LMP. A step-wse congeston management scheme has been developed whereby the dstrbuton system operator (DSO) predcts congeston for the comng day and publshes DTs pror to the clearng of the day-ahead market. EV fleet operators (FOs) optmze ther EV chargng schedules wth respect to the predcted day-ahead prces and the publshed DTs, thereby avodng congeston whle stll mnmzng the chargng cost. The concept of dstrbuton grd congeston allevaton usng day-ahead tarffs (DTs) s explaned n detal n Secton 2. The determnaton of DTs s presented n Secton 3. The optmal EV chargng algorthm wth respect to both day-ahead prces and DTs s descrbed n Secton 4. Case studes have been mplemented usng a Dansh 0.4kV grd and are presented n Secton 5 to llustrate the effectveness of the proposed DT concept. Fnally, conclusons are drawn based on the case study results. 2 Dstrbuton grd congeston allevaton usng DTs The central requrement of an optmal day-ahead dynamc tarff concept s essentally to effectvely allevate congeston, whle ensurng economc effcency. The tarff concepts are roughly dvded nto two categores ntegrated and stepwse. An ntegrated model nvolves the mplementaton of a fully nodal prcng system where both system balance and grd congeston are settled n a sngle step, e.g. FOs submt demand bds to the day-ahead market and are then subject to locatonal prces. Ths s consdered to be the most soco-economcally optmal soluton as the electrcty prce reflects ts true cost. Implementng ths on the dstrbuton network however would requre a complete overhaul of the current Nordc

6 5 day-ahead market (DAM) and would be dffcult to ntroduce, partcularly from the perspectve of complexty and market partcpant acceptance. In a step-wse tarff scheme, the system balance and grd congeston are handled separately. Generally, ths requres predcton from one or more market partcpants. DSOs can predct demand and determne tarffs based on ther predctons, or FOs can predct tarffs and optmze ther demand so as to avod the addtonal tarffs, thereby also avodng congested perods and locatons. It s consdered dffcult to acheve soco-economc optmalty n a step-wse tarff concept, however t s very smple to ntroduce nto the Nordc DAM, as t currently exsts. The DT scheme developed here s based on the concept of a step-wse process where grd congeston s managed pror to the determnaton of system balance on the day-ahead market. The proposed scheme s ntended to ft drectly nto the Nordc electrcty market as t currently stands. Ths wll ensure ease of acceptance from all market partcpants. A key concern wth a step-wse process s the dffculty n attanng a soco-economcally optmal soluton. In ths scheme, t s proposed that the tarffs wll be based on the locatonal margnal prce of electrcty, thus they wll reflect the true cost of generaton, transmsson and dstrbuton of electrcty. For ths purpose, a nodal prcng market s smulated by the DSO, however the prce s mplemented n the format of tarffs for flexble demand (n ths case EVs) thus avodng alteratons n the structure of the current DAM and unfar penaltes for non-flexble demand. The flowchart n Fgure 1 explans the proposed step-wse dstrbuton grd EV congeston allevaton scheme usng DTs. The steps of mplementng the proposed method are lsted below. A. DSO predcts day-ahead prces of electrcty for the comng day and EV drvng patterns the day-ahead prce predcton wll be nfluenced by the wnd forecast, partcularly as the penetraton of low margnal cost generators such as wnd ncreases. The drvng pattern predcton wll be based on hstorcal drvng data; the hstorcal drvng data could be from the transport authorty or FOs. B. Based on the predcted day-ahead prces and EV drvng pattern, the DSO forecasts the EV chargng schedules of all EVs on the dstrbuton grd wth the most cost effcent EV chargng scenaro. C. The DSO runs an optmal load flow calculaton based on predcted conventonal and EV chargng demand and determnes locatonal margnal prces (LMPs) for each node n the dstrbuton grd.

7 6 D. DTs are determned based on the congeston cost component of the LMP and these wll then be made avalable to the FOs pror to the market clearng of the dayahead market. E. FOs bd nto the day-ahead market based on the publshed tarffs and predcted day-ahead prces. The concept of the proposed step-wse dstrbuton grd congeston allevaton from EVs usng DTs s shown n Fgure 2. In the proposed concept, there s an assumpton that all FOs are senstve to the prce sgnals and wll try to come out the most economcally effcent EV chargng schedules based on both spot prces and varyng grd tarffs. The dversty of FOs has not be consdered n the current work and t wll be nvestgated n the future work. The proposed concept s qute dependent on the fact that the EV demands wth DTs can adjust the EV demands n such a way that the predcted congestons n electrc dstrbuton networks can be allevated. In realty, there could be msmatch between the predcted EV demands wthout DTs and the real EV demands. Ths msmatch could have a dstortve effect on the effcency of the proposed DT concept. Ths knd of msmatch could be handled by more real tme market mechansm or drect control methods. 3 Determnaton of DTs Locatonal margnal prcng s a prcng structure that reflects the margnal cost of electrcty supply at each ndvdual bus. The prce reflects the margnal cost of generaton as well as the costs assocated wth the physcal constrants of the system. The LMP exposes producers and consumers to the true cost of energy delvery throughout the system. It can be decomposed nto three consttuent components; the margnal cost of generaton, margnal cost of losses and the margnal cost of congeston [16]. AC optmal power flow (ACOPF) s typcally used for LMP calculaton. However, DC optmal power flow (DCOPF) s also wdely used for LMP calculaton due to ts effcency and acceptable accuracy. The DCOPF s consdered to be suffcent n many cases [17]. In ths case, for the calculaton of LMPs, the DCOPF s a sutable opton, partcularly consderng the margnal congeston cost used for determnng DTs and the large number of nodes on the dstrbuton system. In ndustry, DCOPF has been employed by several software tools for chronologcal

8 7 LMP smulaton and forecastng, such as ABB GrdVew TM, Semens Promod, GE MAPS TM and PowerWorld [18]. In the proposed DT scheme, the tarffs are determned by the margnal cost of congeston; the margnal cost of losses s not a core element for the determnaton of the DTs. Therefore, the lossless DCOPF s used to calculate the LMPs. The DCOPF wthout losses can be represented by the followng optmzaton formulaton. Objectve functon N Mn c G 1 (1) subject to the followng equalty and nequalty constrants: N G N 1 1 D (2) N PTDFk ( G D) Lmtk k 1,2,, M (3) 1 mn G G G max, 1,2,..., N (4) where N s the bus number, M s the lne number, G s the generaton dspatch at bus (MWh/h), generaton output at bus, lne k from bus, D s the demand at bus, s the generaton cost at bus (DKK/MWh), max G and Lmt k s the transmsson lmt of lne k (MW). mn G are maxmum and mnmum PTDFk s the generaton shft factor to In the above DCOPF formulaton, the power transfer dstrbuton factors (PTDFs) are the senstvtes of branch flows to changes n nodal real power njectons [19]. In general, the PTDFs depend on the operatng pont and topology of an electrc power system. However, t has been demonstrated n [20] that the PTDFs are relatvely nsenstve to the operatng pont of an electrc power system f the topology s fxed and there s c

9 8 suffcent reactve power compensaton to have voltages constant at all buses. Therefore, t s feasble to keep the PTDF matrx constant for the DCOPF calculaton. The Lagrangan functon of the DCOPF s descrbed by Eq. (5). N N N M N (5) L ( c G ) ( G D) u ( PTDF ( G D) Lmt ) k k k k 1 1 where s the Lagrangan multpler of the equalty constrant of Eq. (2), s the Lagrangan multpler of nequalty constrant of Eq. (3). The LMP at bus can be calculated usng Eq. (6). M L LMP u PTDF D (6) k k k 1 where LMP s the LMP at bus, s the locatonal margnal energy prce and k M k PTDFk s the locatonal margnal congeston prce. k 1 The locatonal margnal congeston prce reflects the real congeston cost and s used to determne the DTs. The DTs can be calculated by Eq. (7). DT M t t t k PTDFk k 1 (7) t where DT s the DT at bus of tme step t, M t t k PTDFk s the margnal congeston prce k 1 t at bus for tme step t. As k s the Langrangan multpler for Eq. 3, t s only non-zero when the constrant s bndng, thus durng uncongested perods (when the component loadng s below the defned lmt) the LMP reverts to the margnal cost of generaton and no tarff s appled. A number of power systems across the world operate on a LMP based market. The remarkable example s the PJM n the east of USA. A few other examples are the New England system, the New York System, the Electrc Relablty Councl of Texas (ERCOT) system n the USA, the New Zealand system and the Alberta system n Canada [21]. The LMP reflects the margnal cost of electrcty supply at each system bus In order to determne the approprate tarff

10 9 magntude over a suffcent perod, the LMP calculaton has been altered whereby the electrc dstrbuton system s consdered to contan only two generators, whose relatve prce dfference corresponds to the maxmum prce dfference durng the optmsaton perod. Under ths framework, durng uncongested perods only one generator supples power, and the system prce corresponds to the predcted day-ahead prce. Durng congested perods, where system loadng exceeds a gven threshold, the second generator supples power n order to redrect power flow accordng to the congeston experenced. In ths case the DT, or congeston component of the LMP, s non-zero and ts magntude s related to the system prce profle; the approprate DT magntude s therefore determned. 4 Optmal EV chargng management The DT concept for the allevaton of dstrbuton grd congeston s dependent on the assumpton that FOs wll optmze ther fleet chargng behavor n response to prce sgnals from both predcted day-ahead prces and publshed dynamc tarffs. The tarffs are calculated based on the congeston cost component of the LMP, thus they are only actve durng perods of congestons. Ths has the effect of shftng chargng demands from congested locatons and tmes. The objectve functon shown below s a lnear expresson, whch nherently assumes that the FO s a prce taker and cannot exert any market power [22]. For ths study t s assumed that there are suffcent dstnct FOs to ensure a compettve market where market power cannot be exercsed. Ths wll requre a degree of market regulaton to ensure, however ths wll lkely only become an ssue once hgh EV penetraton levels are realzed, at whch pont the market volume wll ensure the economc vablty of multple FOs. If the FOs are allowed to exert market power, an alternatve objectve functon s requred. A quadratc programmng optmzaton can account for alteratons n the predcted day-ahead prce due to gamng behavor from FOs [22]. The mathematcal formulaton of the optmzaton for FOs s shown below. Objectve Functon

11 10 mn T I t Ent, Ct DT t 1 1 n (8) subject to the followng EV chargng constrants 0 E w E t, n (9) nt, nt, max n n n, d mn nt nt, nt, dnt,, nt, max (10) t 1 t 1 SOC SOC E w E SOC t, n where wnt, nt, 1 t, n (11) w nt, nt,, 0,1 t, n (12) C t s the predcted electrcty day-ahead prce at tme t [DKK/kWh], bus for tme t [DKK/kWh], t DT s the DT at s the drvng Energy requrement for EV n durng perod t [kwh], E s the maxmum chargng energy durng perod t [kwh], SOC s the mnmum max E dnt,, mn battery state of charge (SOC), SOC s the maxmum battery SOC, s the ntal max SOC nt battery SOC, n s the set of tme at whch vehcle becomes unavalable for chargng,, s the set of duratons for whch vehcle s unavalable for chargng, v nt, s the bnary parameter sgnfyng EVs drvng status - 1=drvng and 0=avalable to charge, s the bus ndex, n s the EV ndex, E nt, s the chargng energy for EV n durng perod t [kwh], w nt, s the bnary control varable sgnfyng EVs chargng status - 1=chargng and 0=not chargng. The objectve functon ensures the total cost of chargng s mnmzed consderng both predcted day-ahead prces and publshed DTs. The constrants guarantee that the battery SOC s at least suffcent to facltate the drvng energy requrement durng the next drvng perod, whle also allowng addtonal chargng to satsfy future drvng requrements, for example to take advantage of low cost perods. In addton to these drvng constrants, the battery SOC s restrcted to between SOC mn and SOC max so as not to reduce the battery lfetme. E max determnes the maxmum chargng power as t s assumed that chargng power s kept constant durng each tme perod. The chargng power s restrcted by the grd connecton. nd

12 11 5 Case studes Case studes are employed to determne the effectveness of the proposed DT concept for the allevaton of dstrbuton grd congeston from EV chargng. A Dansh 0.4kV dstrbuton grd has been used to mplement case studes comparng the stuaton wth and wthout the applcaton of DTs. The optmal chargng behavor of a sngle cohort of EVs s modeled wth respect to a varety of day-ahead prce profles to nvestgate the resultng congeston and the effectveness of the DTs under varous crcumstances. The case studes have been mplemented usng DIgSILENT s PowerFactory and Mathworks Matlab. The selected dstrbuton grd s modeled n PowerFactory and the OPF functon of PowerFactory s used to calculate the LMPs on the network. Day-ahead prces from Nord Pool ElSpot are used to represent the predcted prce of electrcty. The optmal EV chargng management wth and wthout DTs s determned usng the lnear programmng functon n MATLAB. 5.1 Case Study Inputs Dstrbuton Grd The dstrbuton grd used n all case studes s shown n Fgure 3. It s comprsed of two 10/0.4 kv transformers, 33 cables, 33 buses and 72 household customers. One generator s located on the MV sde of the 10/0.4kV transformer at the top of fgure 3 to represent the supply of power to the dstrbuton grd. The margnal cost of power from ths generator s gven as a tme seres of day-ahead prces from Nord Pool ElSpot; ths facltates the calculaton of LMPs at each bus. Another generator s located on the lower rght-hand corner of Fgure 3; ths s used to provde reverse power to allevate congeston, t also contrbutes to the calculaton of the LMP. The case studes that follow resulted n dstrbuton level congeston on the feeders hghlghted n red n Fgure 3. Ths manfested tself n the form of hgh transformer and lne loadng levels. EV data A non-homogenous EV fleet was assumed for the EV chargng management studes. The EV battery sze vared accordng to ndvdual EV drvng requrements. It s assumed that the maxmum chargng power s 2.3kW (based on a 10A, 230V connecton). A typcal value of 150

13 12 Wh/km s used to calculate the energy consumpton whle drvng [2]. The mnmum and maxmum EV battery SOCs are set as 20% and 85%, respectvely. The ntal EV SOC vares by vehcle, and s set such that ndvdual chargng and drvng requrements can be met; ths s n accordance wth the non-homogenous nature of the vehcle cohort. The ntal SOC s assumed for the purposes of the case study; n realty the FOs would operate on the bass of contnual optmzaton perods and the ntal SOC would be determned from the fnal SOC of the prevous optmzaton perod. A summary of the EV data s lsted n Table 1. Drvng data In addton to the EV battery and chargng power data, the EV drvng pattern data are also very mportant. Drvng data was provded from the Dansh Natonal Travel Survey [2]. Ths data s hghly detaled and provdes sgnfcant nsght nto the drvng habts of Dansh resdents. For ths study, the relevant data were as follows: drvng stop and start tmes, dstance drven durng drvng perods, and day of week of survey data collecton. A cohort of 60 EVs was selected arbtrarly from the travel survey, wth the only requrement beng that the drvng requrements could be met by an EV wth battery capacty not exceedng 25 kwh. The EV avalablty for chargng s defned as the perods durng whch the EV s at the home locaton. As ths study s prmarly concerned wth congeston management on the dstrbuton network ths s an acceptable assumpton; other chargng behavor, such as on-street chargng, wll requre alternatve management, more lkely on the real-tme scale. A realstc drvng profle has been obtaned by selectng vehcles from the same day; n ths case a Monday has been selected. The avalablty of cohort for chargng s llustrated n Fgure 4. Each horzontal secton represents a sngle EV, wth the brown colour sgnfyng avalablty to charge (at resdence), and blue sgnfyng that the EV s away from home, and therefore s unavalable to charge. Tme resoluton for dynamc tarffs and EV chargng management A tme resoluton of 15 mnutes was selected as t provdes ncreased resoluton from the tradtonal hourly dspatch. Ths allows for the hgh-resoluton nature of EV drvng behavor wthout requrng excessve computatonal power or tme. Tarffs are therefore also set at 15- mnute resoluton, however the day-ahead market remans at hourly resoluton. Ths s

14 13 acceptable, as the Nord Pool day-ahead market bds are placed n terms of volume [MWh/h]. Therefore, the bddng system does not requre alteraton for ths concept. Prce Profles The optmal EV chargng profle s dependent on the prce profle; thus for a gven cohort of vehcles on the test network, the level of congeston exhbted wll also vary by prce profle. For ths reason, a number of prce profles have been smulated to explore the capabltes of the proposed LMP based DT for the preventon of congeston. The detals of the smulated prce profles are suppled n Table 2. All case studes are conducted on a sngle cohort of EVs. The LMP calculaton s dependent on the defnton of congeston; typcally ths s gven as system component loadng exceedng 100%. However, for ths applcaton ths defnton rarely results n complete congeston allevaton; hgh levels of system loadng usually occur over extended perods, of whch only a proporton wll have system loadng exceedng 100%. For ths reason, a system loadng lmt (congeston defnton for the purpose of LMP calculaton) below 100% s often benefcal, as the resultng DTs wll be over such a tme extent as to prevent the formaton of secondary congeston peaks followng the applcaton of DTs. However, f the system loadng lmt s too low, the resultng tarffs wll be excessve and economcally neffcent. For ths reason, each of the prce profles examned n the case studes that follow have been tested at a range of system loadng lmts from 50% to 100% n ncrements of 10%. The most successful results are presented, and conclusons can then be drawn on the crtcal factors that determne the approprate system loadng level defnton. 5.2 Case study results Four case studes have carred out n order to llustrate the effcacy of the proposed dayahead tarffs for allevatng congeston from EVs and show the needs of havng a dynamc system loadng lmt to determne the effcent day-ahead tarffs. In the four case studes, the EV number and the drvng pattern are same. The four dfferent prce profles specfed n Table 2 have been used to mplement the case studes. The case study results are presented n fgures 5-8 below. In each case the system loadng curves before and after the applcaton of DTs are presented. Day-ahead preventon of

15 14 congeston s acheved n all cases, though at dfferent system loadng lmts. Most notably, congeston correspondng to a system loadng of 155% s successfully prevented n both case 3 and 4. System loadng of 130% s prevented n case 1. It can be observed from case 2 that sharp peaks n system congeston do not requre the applcaton of DTs over an extended perod as the adjacent system loadng s suffcently lows that even wth the shftng of chargng, secondary congeston peaks do not occur. In cases where the prce profle exhbts frequent fluctuatons, such as n cases 3 and 4, a system loadng lmt of 90% s suffcent to prevent congeston, as there are multple low cost hours to whch EV chargng can be shfted accordng to the dversty n avalablty to charge. However, a lower system loadng lmt s requred n cases where prce dps occur durng peak chargng perods such as the early evenng, when EVs become avalable to charge after returnng home. Ths s evdent from case 1 where a system loadng lmt of 80% s requred to fully prevent congeston. The case study results llustrate the requrement for a dynamc system-loadng lmt. Settng a low lmt can lead to excessve cost burdens as tarffs are appled where they are not necessary. Ths s also economcally neffcent. Settng too hgh a lmt can prevent the effectve dspersal of chargng, resultng n falure to completely allevate congeston. 6 Conclusons The study llustrates the effcacy of an economcally effcent day-ahead DT concept for the preventon of dstrbuton grd congeston from EVs, whch can be mplemented n the current Nordc market setup. The concept s found to be successful n a number of stuatons. It s also found that the system loadng lmt s a determnng factor for successful preventon of dstrbuton level congeston from EVs. The case studes have hghlghted that the predcted prce profle s a determnng factor for the approprate system loadng lmt that results n DTs that are both economcally effcent and techncally successful. Further work s requred to develop the concept for ts applcaton n a broader settng. Addtonal nvestgaton wll be requred to verfy the results takng nto account the uncertanty of

16 15 predctons. Prelmnary analyss suggests that uncertanty wll ncrease dversty of demand, thereby reducng congeston and mprovng the effcacy of the DTs. Acknowledgement The authors are grateful to the fnancal support from the Electrc vehcles n a Dstrbuted and Integrated market usng Sustanable energy and Open Networks (EDISON) project whch s funded by the ForskEl program (ForskEL Project Number ). References [1] Z. Xu, M. Gordon, M. Lnd and J. Østergaard, Towards a Dansh Power System wth 50% Wnd Smart Grds Actvtes n Denmark, n: Proceedngs of IEEE Power & Energy Socety General Meetng, Calgary, July 26-30, [2] Q. Wu, A.H. Nelsen, J. Østergaard, S.T. Cha, F. Marra, Y. Chen, C. Traeholt, Drvng Pattern Analyss for Electrc Vehcle (EV) Grd Integraton Study, n: Proceedngs of IEEE PES Conference on Innovatve Smart Grd Technologes Europe, Gothenburg, October 11-13, [3] J.A. Pecas Lopes, F.J. Soares, P.M. Rocha Almeda, Identfyng Management Procedures to Deal wth Connecton of Electrc Vehcles n the Grd, n: Proceedngs of IEEE PowerTech Conference, Bucharest, 28 June 2 July, [4] K. Clement, E. Haesen, J. Dresen, Coordnated Chargng of Multple Plug In Hybrd Electrc Vehcles n Resdental Dstrbuton Grds, n: Proceedngs of 2009 Power Systems Conference & Exposton, Seattle, March 15-18, [5] M. Arndam, S.K. Kyung, J. Taylor, A. Gumento, Grd Impacts of Plug-In Electrc Vehcles on Hydro Quebec's Dstrbuton System, n: Proceedngs of 2010 IEEE PES Transmsson and Dstrbuton Conference and Exposton, New Orleans, Aprl 19-22, [6] J. Taylor, M. Matra, D. Alexander, M. Duvall, Evaluaton of the Impact of Plug-In Electrc Vehcle Loadng on Dstrbuton System Operatons, n: Proceedngs of IEEE Power & Energy Socety General Meetng, Calgary, July 26-30, [7] O. Sundstrom, C. Bndng, Plannng Electrc-Drve Vehcle Chargng under Constraned Grd Condtons, n: Proceedngs of 2010 Internatonal Conference on Power System Technology, Hangzhou, October 24-28, 2010.

17 16 [8] N. Roterng, M. Ilc, Optmal Charge Control of Plug-In Hybrd Electrc Vehcles n Deregulated Electrcty Markets, IEEE Transactons on Power Systems, Vol. PP, No. 99, (2010) 1 9. [9] S. Babaem, D. Steen, L.A. Tuan, O. Carlson, L. Bertlng, Effects of Plug-n Electrc Vehcles on dstrbuton systems: A real case of Gothenburg, n: Proceedngs of IEEE PES Conference on Innovatve Smart Grd Technologes Europe, Gothenburg, October 11-13, [10] K.J. Dyke, N. Schofeld, M. Barnes, The Impact of Transport Electrfcaton on Electrcal Networks, IEEE Transactons on Industral Electroncs, Vol. 57, No. 12, (December (12)) (2010) [11] J.A. Pecas Lopes, F.J. Soares, P.M. Rocha Almeda, Integraton of Electrc Vehcles n the Electrc Power System, Proceedngs of the IEEE,Vol. 99, No. 1 (January (1)) (2011) [12] R.D. Chrste, B.F. Wollenberg, I. Wangensteen, Transmsson management n the deregulate envronment, Proceedngs of the IEEE, Vol. 88, No. 2, (February (2)) (2000) [13] H. Glavtsch, F. Alvarado, Management of multple congested congestons n unbounded operaton of a power system, IEEE Transactons on Power Systems, Vol. 13, No. 3, (August (3)) (1998) [14] Y. Xao, P. Wang, L. Goel, Congeston management n hybrd power markets, Electrc Power Systems Research, Vol. 79, No. 10, (October(10)) (2009), [15] Q. Chen, Q. Xa, C. Kang, Novel transmsson prcng scheme based on pont-to-pont tarff and transacton par matchng for pool market, Electrc Power Systems Research, Vol. 80, No. 4, (Aprl(4)) (2010), [16] S. Stoft, Power System Economcs - Desgnng Markets for Electrcty. New York: IEEE Press & Wley Interscence, [17] F. L, R. Bo, DCOPF-Based LMP Smulaton: Algorthm, Comparson wth ACOPF, and Senstvty, IEEE Transactons on Power Systems, Vol. 22, NO. 4, (November (4)) (2007)

18 17 [18] J. Yang, F. L, L. A. A. Freeman, A market smulaton program for the standard market desgn and generaton/transmsson plannng, n: Proceedngs of IEEE Power & Energy Socety General Meetng, Toronto, July 13-17, [19] A. J. Wood and B. F. Wollenberg, Power Generaton Operaton and Control, New York: John Wley & Sons, [20] R. Baldck, Varaton of Dstrbuton Factors Wth Loadng, IEEE Transactons on Power Systems, Vol. 18, No. 4, (November (4)) (2003) [21] E. Hausman, R. Fagan, D. Whte, K. Takahash, A. Napoleon, LMP Electrcty Markets: Market Operatons, Market Power and Value for Consumers. Report prepared for the Amercan Power Assocaton by Synapse Energy Economcs, Feb. 5 th [22] T. K. Krstoffersen, K. Capon, P. Mebom, Optmal Chargng of Electrc Drve Vehcles n a Market Envronment, Appled Energy, vol. 88, p. 9, 3 December 2010.

19 18 Fgure 1 Flowchart of dstrbuton grd congeston allevaton from electrc vehcles usng DTs Fgure 2 Concept of dstrbuton grd congeston allevaton from electrc vehcles usng DTs Fgure 3 Dstrbuton grd from the Island of Bornholm for case studes Fgure 4 EV chargng avalablty for Monday Fgure 5 Case Study 1 wth Nght-tme Peak Prce Profle Fgure 6 Case Study 2 wth Evenng Peak Prce Profle Fgure 7 Case Study 3 wth Snusodal Varaton Prce Profle Fgure 8 Case Study 4 wth Hgher Frequency Snusodal Varaton Prce Profle

20 19 Table 1 EV data summary for EV chargng management study EV Parameter EV battery sze Chargng power Energy consumpton from drvng EV Parameter Value 25 kwh 230 W 150 Wh/km Mnmum SOC 20% Maxmum SOC 85% Table 2 Prce Profle Detals Prce Profle Characterstcs 1 Nght-tme Peak 2 Evenng Peak 3 Snusodal Varatons 4 Hgher Frequency Snusodal Varatons

21 20 Fgure 1 Fgure 2

22 21 Fgure 3 Fgure 4

23 22 Fgure 5 Fgure 6

24 23 Fgure 7 Fgure 8