Impact of Energy Management of Electric Vehicles on Transient Voltage Stability of Microgrid

Transcription

1 EVS28 KINTEX, Korea, May 3-6, 2015 Impact of Energy Management of Electrc Vehcles on Transent Voltage Stablty of Mcrogrd Muhammad Shoab Khald 1, Xangnng Ln 2, Yxn Zhuo 3, Ramesh Kumar 4, Muhammad Kashf Rafque School of Electrcal and Electroncs Engneerng, Huazhong Unversty of Scence and Technology, 1037 Luoyu, Road, Wuhan, Chna. emal shoabusman@hust.edu.cn, xangnng.ln@hust.edu.cn 4 School of Electroncs and Computer Engneerng, Hanyang Unversty, South Korea. 5 School of Informaton and Communcaton Engneerng, Sungkyunkwan Unversty, South Korea. Abstract There s cause and effect relatonshp between ncrease n load due to ncreasng penetraton of electrc vehcle load and unbalanced condtons that may lead to further degradaton of power qualty, voltage problems and even damage the equpment f the system s not properly managed. Ths paper presents detaled revew of energy supply and management n conjuncton wth load synchronzaton through EVs for mantanng transent voltage stablty by provdng reactve power support to the power grd n vehcleto-grd (V2G) mode of operatons. The energy management system s consdered at dfferent levels such as, stand-alone PV, stand-alone wnd, stand-alone battery storage, stand-alone EV parkng lot, resdental feeder and commercal buldng feeders. Frst we proposed energy management algorthm, to lmt the peak power drawn by electrc vehcles from dstrbuted energy resources of mcrogrd, such that addtonal electrcal resource wll be transferred to resource constraned devces. The electrc vehcles negotate based on ther demand, prorty and avalable electrcal resource such that durng hgher electrcty prce the hgher prorty vehcles stll requre resource and perform unnterrupted operaton. The transfer of electrcal resource from one load devce to another wll help n reducng peak demand and mprovng the effcency of the system. Secondly we proposed transent voltage stablty margn ndex and test the capablty of EVs n contrbutng storage and supply servces to the grd. The energy management control smulatons are realzed n DIgSILENT Power factory. Keywords: dstrbuton network relablty, energy management system, vehcle-to-grd (V2G), ancllary servces 1 Introducton Electrc vehcles have been developed and marketed snce the end of the 19 th century. Energy scarcty and envronmental polluton has compelled researchers to develop effcent, envronmental-frendly electrc vehcles along wth the advancement n the clean energy technology [1]. Currently the penetraton of electrc vehcles load s less than the expected but n near future t s expected that the use of electrc vehcles wll rse. EVS28 Internatonal Electrc Vehcle Symposum and Exhbton 1

2 Moreover t s expected that there wll be more and more nducton motor loads connected to the power grd. The motor loads are random and have fluctuatons n the characterstcs whch have sgnfcance mpact on grd securty [2]. Hence, wth the ncrease n non-lnear load the power system optmzaton control, power qualty mantenance and dstrbuton network plannng become more dffcult. The grd securty and stablty studes are accomplshed to assess the mpact of nducton motor load growth n terms of grd voltage loss, harmoncs and load mbalances and ts mtgaton usng electrc vehcles as a provson of ancllary servces, such as reactve power compensaton, the load support and the frequency tunng [3-5]. Electrc vehcles can act a specal knd of load and source, whch can be suppled power from the grd for chargng and take back power when necessary [6]. As a moble storage unt, EV can be connected to the grd as moble storage unt and act lke a source to provde reactve power compensaton n order to mprove the grd stablty and securty [7]. Therefore t has mportant theoretcal and practcal sgnfcance to study the transent voltage stablty of dstrbuton network whch may be mproved by vehcle-to-grd (V2G) mode of operaton. Several measures have been taken n the recent decades to provde ancllary servces by electrc servces for better stablty and relablty to the stressed systems. Dstrbuted control can provde local compensaton lke power system stablzers (PSS) are nstalled n order to damp lowfrequency oscllaton. The best qualty response s expected wth local measurements about the dynamcs of dsturbance. But there are adverse nteractons between multple adjacent areas of the power system or between power electroncs and grd, only a global control can provde better control [8, 9]. Smlar to wde-area controllers (WACs) mcrogrd energy supply and management system (ESMS) coordnate the actons of the dstrbuted energy management systems usng IEEE W-MAX [10]. The ESMS receves nformaton/data of dfferent areas n the power system and based on some predefned objectve functon, sends approprate control sgnals to the dstrbuted EMS for enhancng system s dynamc performance [11]. Several researchers have contrbuted n the area of ntellgent systems and autonomous agents, such as [12] uses a general case of usng agent n control systems, [13] motvates the uses of agents at dfferent control levels e.g. feeder and load agents. Whereas, n [14] the multagent mechansm for slandng operaton s depcted. Wde area sgnal based ntellgent control for FACTS have been presented n [15]. Most of the studes mentoned so far have consdered conventonal power system and some of them have focused the renewable sources but 1) the effects of bdrectonal power flows ntroduced n the case of electrc vehcles for the stablty of the power system, 2) takng motor load nto consderaton for transent stablty of the power system, 3) load synchronzaton for steady state response of the system and 4) the role of energy management system n mprovng the stablty of the system are not yet reported. In ths study, 1) the mpact of chargng and dschargng cycles of the electrc vehcle (connected to a CERTS mcrogrd) on the mprovement of transent voltage stablty of the mcrogrd n grd connected and slanded mode s presented. 2) An energy management algorthm s developed to balance the load among dfferent types of loads at demand sde n mcrogrd such that addtonal electrcal resource wll be transferred to resource constraned devces. The energy management control s mplemented n DIgSILENT Powerfactory. The results wth and wthout energy management system for moderate dsturbance due to the start of the motor load lke a sudden dschargng of the EV and also for some extreme stuaton lke sudden transton from dschargng to chargng mode or a three phase fault durng chargng or dschargng cycle are presented. The analyss show that the electrc vehcles connected to the power system and the use of energy management system mproves the stablty of the system sgnfcantly. The prmary contrbutons of ths study are; Demonstraton of energy management smulaton model. Illustratng the mpacts of EV operatons on the stablty of the system. The desgn of energy management control for mprovng the stablty of an ntegrated power system. Rest of the paper s organzed as follows: Secton 2 descrbes the smulaton model of CERTS mcrogrd/ power dstrbuton network archtecture. Secton 3 llustrates the load synchronzaton algorthm; Secton 4 ndcates the the cosmulaton platform for energy management system, Secton 5 explans a case study and smulaton analyss and results, and the last Secton 6 shows the concluson and future research drectons. EVS28 Internatonal Electrc Vehcle Symposum and Exhbton 2

3 2 Network archtecture Smulatons are performed for a medum voltage dstrbuton network wth a nomnal voltage of 20 kv. The mcrogrd adopts CERTS mcrogrd archtecture as shown n Fgure 1, whch contans four radal feeders, feeder A, feeder B, feeder C, and feeder D. The CERTS mcrogrd has two crtcal components, the statc swtch and the mrcosource. The statc swtch has the ablty to sland the mcrogrd n case of dsturbance n less than a cycle [16] and reconnect when the trppng event s no longer present. In order to prove the effectveness of the mcrogrd reconfguraton we have added four backup lnes BC1, BC2, CD1 and CD2, these four lnes normally dsconnected durng grd connected mode [17]. Feeder A contans resdental subscrber loads contanng ordnary loads (about 50% nducton motor load), statc loads and chargng ple for chargng of electrc vehcles. Chargng ple s sutable for chargng of electrc vehcles at small scale at slow chargng rate, whle feeder B, feeder C, and feeder D are commercal buldng subscrber loads whch contan less-senstve loads (about 65% nducton motor load) and statc loads, ordnary loads (about 50% nducton motor load) and statc loads, and senstve loads (about 75% nducton motor load), statc loads and chargng staton for the chargng of electrc vehcles. Chargng statons are sutable for chargng of electrc vehcles at large scale, havng the capablty of quck chargng. Swtch power plant s deployed at the medum voltage bus bar whch has the capablty to accommodate bulk of electrc vehcle for quck chargng. The chargng ple, chargng staton and swtch power plant employ power electroncs controllers that nterface the electrc vehcle to the grd. These controllers nclude on board ac to dc converter whch s coupled to the grd wth a sngle or three phase connector. The converter has dual functons as dode brdge rectfer for chargng the battery or a swtch-mode converter whch not only controls the chargng of the battery, but also have the capablty of feedng power from the vehcle to grd (V2G) n regeneraton mode [18]. The electrc vehcle can contrbute to supply storage and supply servces to the grd for transent voltage stablty. Stand-alone PV, Wnd and battery storage are also connected to the medum voltage bus bar. Whereas, stand-alone battery storage s mportant to provde transent stablty n slandng mode, stand-alone PV modules and wnd power to get benefts of renewable energy and synchronous generators necessary for stable operaton of mcrogrd n slanded mode. Batteres and synchronous generators are the two types of sources whch may use the droop control for normal operatons. The senstve loads are suppled through more stable source lke synchronous generator due to very hgh power qualty of senstve loads, whle less-senstve loads may be suppled through both synchronous generators and photovoltac system (PV). The ordnary loads may or may not have the local generaton. Moreover senstve and less senstve loads are controllable loads whle the ordnary loads are non-controllable loads. The non-controllable loads are suppled from the power commands through the load profle of grd, whle the controllable loads are suppled from the power commands through the load profle of DERs n mcogrd usng ESMS [17]. Fgure 1: Network archtecture. Controllable and non-controllable loads are further classfed as shftable and non-shfable loads. Where, shftable loads follow the smart prcng provded by the power grd/mcrogrd ESMS and non-shftable loads cannot follow power grd/mcrogrd ESMS. In the context of shftable and non-shftable loads, loads of dfferent characterstcs are avalable. For example; Smple tmed loads, whch are shftable loads. The user can specfy ther power ratng, run tme and desred wndow tme n whch they run. The tme can be chosen by the user or by the ESMS. Setpont based operatng loads, whch are nonshftable loads. The user or the ESMS can specfy the setpont that the devce must meet. EVS28 Internatonal Electrc Vehcle Symposum and Exhbton 3

4 Dynamc loads are shftable loads. These loads can be actvated multple tmes at dfferent nstances durng a day or multple loads can be actvated together at the same nstance of tme lke electrcal vehcles (EV). In our mcrogrd EV does not stand for a sngle vehcle only, but rather for any number of chargng nstances also. The mportant parameters of EV chargng are gven n Table 1. PWT Where stands for the wnd turbne rated power, vn stands for the cut-n speed, vr stands for the rated wnd speed and vout stands for the cutout wnd speed. Relatonshp between the output power and wnd speed s represented by Fgure 2. Table 1: Parameters of EV chargng Nssan -LEAF (Electrc Vehcle) Battery Capacty E cap 24kWh Fnal SOC E fnal E max Mnmum SOC E mn 30% Maxmum SOC E max 80% EV Plug-n Tme (T Intal ) 11 pm T recharge EV Plug-out Tme (T Fnal ) T charged 7 am Inttal SOC E nttal 40% Charger Ratng P chargwe 3.0kW Vehcle to Grd Ratng P V2G 0 kw Delay Tme Settng Co-effcent to Reduce Peak Power 12 am R t The ESMS s not only connected to the power grd/mcrogrd DERs but also wth a combnaton of fxed WMAX and LAN. The ESMS of each unt has the capablty to communcate wth each other, as well as between the central mcrogrd EMS and other EMS unts wthn n the mcrogrd. In the next sub-sectons we dscuss the mathematcal modelng of wnd turbne generator, photovoltac cell, battery storage, desel generator, nducton motor and electrc vehcle chargng. 2.1 Wnd turbne generator model Power output of wnd turbnes depends on wnd speed. Mechancal energy of the wnd s converted nto electrcal energy, the relatonshp between wnd speed and the wnd turbne output P power wt s expressed n term of the followng functons gven n Equaton 1; 0 v vn v vn PWT vn v vr Pwt () t vr vn PWT vr v vout 0 v vout (1) Fgure 2: Wnd generator actve power output curve 2.2 Photovoltac cell model The output power of photovoltac cell s easly affected by the radaton ntensty, envronment temperature, and factors lke shadows, clouds, etc. In ths paper, the photovoltac cell output only consders the effects of rradaton ntensty and ambent temperature as shown by Equaton 2; P P G [1 k( T T )]/ G (2) V STC C C STC STC GC Where stands for the lght ntensty of work 2 ponts GSTC 1 kw / m ; PV stands for photovoltac output power, PSTC stands for photovoltac power TC rated power, stands for the battery surface temperature TSTC 25C ; k stands for the temperature coeffcent of the power. 2.3 Storage battery model The excessve energy from renewable resources s stored n the battery and when there s nadequate energy generated by the renewable resources battery dscharges and provdes contnuous energy. Battery plays a very mportant role n slanded mode of operatons. Accordng to the energy conservaton law, storage batteres charge and dscharge cycles satsfy the followng Equatons. E ( t) E ( t 1)(1 ) P ( t) BAT BAT BAT _ C c E ( t) E ( t 1)(1 ) P ( t) BAT BAT BAT _ D d (3) PBAT _ C () t PBAT _ D () t Where s the perod t to the outsde of the battery charge and dscharge system power, c d s the charge and dscharge EVS28 Internatonal Electrc Vehcle Symposum and Exhbton 4

5 effcency of the battery, s storage battery dran rate (% / h). 2.4 Desel generator model The perod t n desel generator for fuel consumpton s gven by followng Equaton 4. V ( t) P ( t) fuel fuel engne (4) Vfuel () t Where desel generator fuel consumpton Pengne perod t; () t s the desel generator output fuel power perod t, s desel fuel factor L/kWh. 2.5 Inducton motor load model The power system ntegrated load model s comprsed of nducton motor n conjuncton wth the statc load that can be representaton of actual dynamc load [19]. Ths paper consders slp ndex of the nducton motor for transent voltage stablty margn ndex. The value of slp ndex s based on ether the clearng tme of small dsturbance or by the reactve power support by the avalable electrc vehcles. It s a measure of voltage nduced by the stator to the rotor of nducton motor durng the fault/dsturbance. The slp s of an nducton motor s the dfference between the synchronous speed and the rotor speed, expressed as a per-unt of synchronous speed [20] and the slp-torque characterstcs are gven n [21] Impact of nducton motor on transent voltage stablty margn n dstrbuton network Short term nterference n voltage manly refers to dynamc load stablty analyss. Dynamc load especally nducton motor load has the greatest mpact on the transent voltage stablty. In extreme condtons t may cause a larger voltage collapse. We consder the dynamc role of nducton motor load n perspectve of transent voltage stablty. It s necessary to study the transent behavour of the turbulent nducton motor connected to the power grd n order to determne the power grd transent stablty. The transent voltage stablty margn ndex can be explaned as follows; 1). The slp s can be characterzed as a measure of transent voltage stablty margn ndex [20]. s n n / n (5) s s Where s s the slp, n s s the synchronous speed (r/mn) and n s the rotor speed (r/mn). The slp s practcally zero at no load and s equal to 1 when the rotor s locked. 2). If S scm represents non-steady state slp (at mnmum voltage) and S max the maxmum value of nducton motor slp (at clearng tme), then the transent voltage stablty margn ndex can be represented as followng n Equaton 6. s scm s s scm max (6) 2.6 Electrc vehcle chargng Dfferent knd of batteres can be employed wth the electrc vehcles such that lead acd, nckel cadmum and lthum on etc. The lthum on batteres are preferred due ther hgher energy densty and ther capablty of less selfdschargng [22]. Moreover dfferent types of chargers can be employed such as hgh frequency swtchng (HFS) chargers and phase control (PC) chargers. HFS s preferred for chargng at power factor above 0.9; lower harmonc polluton, hgher effcency and fast dynamc response as compared to PC. EV chargng has three modes, modes of slow (L1), routne (L2), and fast (L3) chargng. These modes have fve parts as shown n Table 2. Chargng Modes L1 L Table 2: EV chargng modes Nomnal Voltage (V) Sngle-phase 220 Three-phase 220 Three-phase 380 Nomnal Current (A) Nomnal Actve Power (KW) Three-phase L Development of chargng statons s one of the necessary condtons for the wde spread of electrc vehcles. The electrc vehcle chargng may be dvded nto three categores contanng chargng ple, chargng staton and swtch power plant as shown n Table 3. Although chargng ples belongs to the resdental users but consderng the popularty of the electrc vehcles the mpact of chargng ple cannot be gnored for the transent stablty of the dstrbuton network. The operaton of chargng ple s dfferent EVS28 Internatonal Electrc Vehcle Symposum and Exhbton 5

6 from the other household electrcal applances due to the load aggregaton. The penetraton of electrc vehcles can be defned by the followng relaton gven n Equaton 7. P P ev L (7) Where ρ s the electrc vehcle penetraton rate, P ev s actve power demand for the electrc vehcles and P L s the actve power demand for all other load ncludng the nducton motor load. The reactve power demand of the electrc vehcle follows the relaton gven below; Q ev P PF ev 1 sn(cos ( PF)) (8) Where Q ev s the reactve power avalable by the electrc vehcle and PF s the power factor of the electrc vehcle charger. Type of Charger Chargng Ple Chargng Staton Swtch Power Plant Table 3: Types of EV chargng staton Chargng Tme Sutablty 1-6 Hours Sutable for Resdental Parkng at Small Scale n Dstrbuted Fashon 30 Mnutes (Approx.) 5-10 Mnutes (Approx.) Sutable for Traffc Intensve Areas/ Hghways at Large Scale n Centralzed Manner Sutable for Cty Centres/ Hghways at Large Scale n Centralzed Manner 2.7 Transent voltage stablty and research response In recent years, the nducton motor, power electroncs and other dynamc load has mpact on transent voltage stablty of the power dstrbuton network and n future more and more electrcal vehcles wll be n use and t s expected to buld large scale electrc vehcles chargng statons and they wll rely on the power electroncs controlled devces connected to the power grd and wll be n the grd for provdng the base load and reduce the stablty of the power system. Contrary to fact, wth the use bdrectonal battery chargers the electrc vehcles can be used n V2G mode as spnnng reserve and enhance the stablty of the grd. As a matter of fact, f the electrc vehcles are consdered as load as shown n [23], smaller sze electrc vehcles are consdered as threat to the safety of voltage stablty of dstrbuton network. To mtgate the peak power drawn by electrc vehcles we establshed a load synchronzaton algorthm. The purpose of the algorthm s to transfer the extra resource on resource constraned loads n the mcrogrd and lmtng the peak power drawn by electrc vehcles usng energy management system. The load synchronzaton algorthm s explaned n the next secton. 3 Load synchronzaton algorthm for energy management of EV Consderng the network archtecture shown n Fgure 1 we balance between the shftable (EV) and non shftable loads (other loads) by transferrng the extra electrcal resource to the resource constraned loads usng game theoratc approach. In whch each electrcal laod plays a non-cooperatve game to reach desred optmal pont called Nash Equlbrum. We consder N loads = (1, 2, 3.N), and tme of twenty-four hours j = (0, 1, 2, 3.24) to play a non-corporatve game for balancng of loads each hour. Frst we determne total avalable resource R j and fx reference resource RR j for each load at tme j based on hstory data, to gve an estmate to energy supply and management system of usual hourly requrement of each load. Second we consder an mportant parameter resource share rato RS j, whch s related to predefned percentage of share n R j. The ESMS allocates the resource RA j for every load based on hstorc resource share rato RS j each hour after calculatng Rj. Where resource allocaton s gven by; RA RS j R 1RS j j N j (9) After resource allocaton we determne surplus or defcency SD j of resource to the load usng EVS28 Internatonal Electrc Vehcle Symposum and Exhbton 6

7 Equaton 10. Load havng more RA j reference resource RR j has surplus resource (SD j >=0), otherwse defcency of resource. SD j RA j RR j (10) Before re-dstrbuton of resource we add all the surplus resource to get total surplus sum TSS j for any stage of tme j and n a smlar manner add defcency of resource to get total defcency sum TDS j. Where, TSS j s avalable for aucton. We consder that non-shftable loads have strct resource requrements and shftable loads have mnmum resource requrements. It s mportant to satsfy non-shftable loads before redstrbuton whereas f the resource cannot fulfll the mnmum requrement of shftable loads, they can be preempted to swtch-off. Resource redstrbuton RD j s calculated after satsfyng the non-shftable loads and teratvely dstrbutes extra resource to resource defcent loads one unt a tme, whch s gven by Equaton 11. RD TSS TDS j (11) j j Next we calculate surplus/defcency reserve after re-dstrbuton SRAR j or DRAR j for loads at tme of equlbrum T j as gven n Equaton12 and Equaton 13. SRAR j RA j RD j DRAR j RA j RD j (12) (13) Surplus resource wll be dstrbuted t to the hgh prorty devces PR j. Lke f the requrement of setpont operated devces s not fulflled they wll be consdered now, moreover f the resource cannot fulfll the mnmum requrement of shftable loads, they can be preempted to swtchoff. Now we calculate the satsfacton level SL j of each load by Equaton 14; SRAR N j SL j PR j 1 RR j (14) Based on resource re-dstrbuton, satsfacton level s calculated after complete resource redstrbuton. It defnes how much resource a load has receved as compared to ts reference resource. Strategy that maxmzes the satsfacton and payoff for each player s the equlbrum strategy for ths stage. It s termed as Nash Equlbrum n game theory. The equlbrum or optmal pont s calculated by; EQ j arg SL max (15) j T X j Equaton 15, calculates the expected optmal pont after resource dstrbuton and try to gve the best response by all players n current stage of the game because at ths tme no player has beneft to devate from ths pont. The dea s not to fnd a global optmal but rather a local optmal soluton. 4 Co-Smulaton Platform for Energy Management System There are many real tme smulators such as RTDS, HYPERSIM and ARENE [11-14]. These smulators are capable of testng real tme power network smulatons wth real lfe networks or smulaton that operates n real tme lke QualNet [10]. Real tme smulator has the capablty to run complex models at the same tme as the physcal tme. There are tremendous advantages of usng real tme smulators n terms of development tme and lower costs when they are ntegrated wth the HIL and SITL smulatons. For example t wll advantageous to use HIL smulatons under dfferent condtons and scenaros, such as for normal and transent stuatons otherwse t would be costly or dffcult to test wth a physcal plant. Real tme smulaton s not the goal of our research however the focus s to ntegrate the power and network communcaton so that smulaton tme clocks of both advance seamlessly. It s assumed that all subsystems are synchronzed together wth very hgh precson allowng multple controls under test to be connected to the system smultaneously. There are four man objectves for modellng and smulaton of energy management system; Desgn of control algorthms for energy management and ther correspondng software archtectures. Analyse the requrements of and mpact of communcaton technologes on the performance of control algorthms. Better desgn of the controller. Development of flexble and modular smulaton envronment. These requrements are mportant for nvestgatng n the context of energy management system for the development of smart chargng algorthms that EVS28 Internatonal Electrc Vehcle Symposum and Exhbton 7

8 can reduce peak power; mnmze the energy costs and optmal use of renewable energy sources Communcaton effects n control system Electrc vehcles may be consdered as dynamc load requrng large amount of power from the grd but also capable of feedng power from the vehcle to grd (regeneraton). Properly desgned and controlled network can provde ancllary servces and supply the supply network, such as frequency regulaton by controllng the actve power flow, supply/demand matchng and the reactve power support. Communcaton thus plays an mportant role n managng the dstrbuton of energy. It s mportant for the chargng staton about the state of the grd n order to better estmate the tmes for chargng [24]. Dfferent communcaton methods can be adopted for chargng statons such as Power Lne Carrers (PLC), IEEE (Zgbee) etc. [25]. Integraton of communcaton and power systems/power electroncs can be consdered as Networked Control System (NCS) n whch there s a communcaton between the remote controller and plant. It s mportant to menton here that there are several communcaton factors whch affect the performance of NCS lke, latency, packet losses, and bandwdth and packet sze ncludng packet dsorderng. Hence NCS can be consdered as stable when the effects of the network are not taken nto account but unstable n real lfe condtons Nature of delays In [26] t s shown that delay can be decomposed nto two parts as shown n Fgure 3: Controller to actuator τ ca and sensor to actuator τ sc. In order to model the system correctly t s mportant to classfy the nature of delays lke, random, tme varyng, or constant and bounded or unbounded, dependng on the Medum Access Control (MAC) protocol [27]. There are varetes of medum access protocol for wred and wreless networks ncludng Carrer Sense Multple Access (CSMA), Tme Dvson Multple Access (TDMA), Token Bus, Token Rng and Token Passng. Where, a CSMA protocol offers uncertan or random delays e.g. Controller Area Network (CAN) or Ethernet whle TDMA or Token type s offers determnstc type of latency. Fgure 3: Networked controlled system NCS modellng There are dfferent types of control methods for modellng NCS, n whch some methods requre hstory of the delays. Stochastc methods are employed for random type of latency lke Posson and Markov [28]. [29] has adopted dscrete approach for the determnstc delays n the NCS: x k 1 h xkh 0 kukh 1 ku k 1 h (16) Where, h s the samplng perod and k ca sc Ah e h k 0 0 h 1 h k As e Bds As e Bds Furthermore the system can be represented n the form; 1 x k A x k Bu k (17) Where, 1 A 0 0, B 0 1, xk xk 1 uk If the perodc delay τ s less than the samplng perod, the system wll stll be consdered tme nvarant and the stablty of the system can be calculated by a regulator such as; u k Kx k, by computng the egenvalues of the matrx A BK 0 [30]. Ths method wll be adopted for future studes n smart grd and EV applcatons. Wth the platform dscussed n Secton 2 t s possble to smulate the model developed. EVS28 Internatonal Electrc Vehcle Symposum and Exhbton 8

9 5 Case study In ths secton, a case study of a modern dstrbuted system s presented. The purposes of ths model are to test the effects of dstrbuted generators that create reverse power flow n the dstrbuton network and b-drectonal power flow whch affect the qualty of power supply and voltage levels. There are ncreased fault current due to the dstrbuted generatons, malfunctonng of protecton system and phase mbalance. Electrc vehcles are consdered as actve loads. They may affect the transent voltage stablty of the dstrbuted network both n chargng and dschargng (regeneraton) mode of operaton. The mpact s expected to be sgnfcant due the ncreasng penetraton of EV and ncrease n use of nducton motor load. The resultant effects are analysed for the desgn of better EV nterface devces and future power networks. It s possble to desgn the EV nterface to elmnate the effects of EV on the network fault level and the protecton system. However, ther effects on the network such as loadng, voltage profle, and phase mbalance and power qualty could be sgnfcant and need to be properly addressed. 5.1 Model Descrpton Consder radal network shown n Fgure 1. Each feeder contans dfferent portons of statc load n parallel to the nducton motor load such as, Feeder A contans resdental subscrber loads contanng ordnary loads (about 50% nducton motor load), statc loads and chargng ple for chargng of electrc vehcles. whle feeder B, feeder C, and feeder D are commercal buldng subscrber loads whch contan less-senstve loads (about 65% nducton motor load) and statc load, ordnary loads (about 50% nducton motor load) and statc loads, and senstve loads (about 75% nducton motor load) and statc loads, respectvely. The chargng staton deployed at feeder D s consdered to provde transent voltage stablty for ths research. The battery SOC can be reached by 80%-90% wthn 30 mnutes. EVs are normally n constant current chargng and ts actve power s far more than the constant voltage chargng. For ths research the charge current and dscharge current s of worth mportance. As gven n Table 2 the routne chargng for our case has DC bus voltage of 380V, the chargng current s 63A, and the actve power s 24kW. The dstrbuton network has two modes of operaton grd connected and slanded. Durng grd connected mode the battery storage nverter s delegated to keep the dc voltage stable and goal of the dc-dc converter s to control the current level of the battery both n ether chargng or dschargng mode. Whle durng slandng mode the battery storage nverter wll change ts mode to keep constant grd voltage and the converter wll mantan the DC bus voltage by changng ts mode to voltage control mode. When there s three phase fault happens at the grd, a sgnal s sent to the mcrogrd ESMS to synchronze the load. If there s voltage nstablty ndcated by any bus the chargng staton s requested for the reactve power support. The measurements sent through the communcaton lnk contan latency, bandwdth lmtaton and packet losses as ntroduced. As the electrc vehcles are added to the system or by the ncrease n motor load a drop on the ac bus voltage sgnfes that statc as well as dynamc load has to be served. In our smulaton frst we consdered the mpact of ncreasng EV load (chargng mode of EVs). Later we consdered the mpact of ncreasng penetraton of EV (dschargng mode of EVs) n order to conform the transent voltage stablty margn ndex explaned n secton Model dstrbuton The dstrbuted network model s manly dstrbuted nto power system and communcaton system nterfaces. DIgSILENT Powerfactory and OPNET can be used for the smulaton of contnuous tme and dscrete event system smulatons. 5.3 Smulaton Analyss and Results Consderng that the demand of chargng the electrc vehcle reaches maxmum and the Feeder D has bgger proporton of nducton motor, t s expected that ths feeder can lose the stablty wthn no tme on a 3 phase fault at the power grd. It s noteworthy that ncreasng load of EV as t reaches about 40% the system wll become unstable.e. wth the ncreasng penetraton of EV load the transent voltage stablty margn becomes lower due to decrease n the node voltage as shown by Fgure 4. The chargng staton merely supports completely n ths stuaton and the system wll be gong to be unstable n ths stuaton. In order to mprove the transent voltage stablty margn ndex the electrc EVS28 Internatonal Electrc Vehcle Symposum and Exhbton 9

10 vehcles penetraton may be ncreased n V2G mode. By ncreasng the penetraton n V2G mode the transent voltage margn ndex follows the EV dschargng n an event of fault and the state of dstrbuton network wll transfer from and unstable state to the stable state. The mpact of electrc vehcle penetraton to mantan the voltage stablty of the node s shown n Fgure 5. Hence, t s proposed that confguraton of electrc vehcles has to be changed from chargng mode to dschargng (V2G) mode. The crtcal clearng tme as depcted by the transent voltage stablty margn ndex s very much mportant and reconfguraton of electrc vehcles must fast enough to mantan the transent voltage stablty margn ndex. In our study we consdered separate transent voltage stablty margn ndces for statc load, complex loads and the EV loads and found that wth the ncreasng EV load the transent voltage stablty margn ndex become lower, there s an ndrect relatonshp between ncrease n EV load and the transent voltage stablty margn ndex. Moreover there s drect relatonshp between EV penetraton and the transent voltage stablty margn ndex. 6 Conclusons and future works Ths paper dscusses about the mpact of energy management of electrc vehcles on transent voltage stablty of mcrogrd. In the presence of fault of n the power grd, the mcrogrd s slanded. Energy management s necessary to balance the demand and supply. Electrc vehcles can provde the base load for short term and can act as spnnng reserve to enhance the grd stablty. Transent voltage stablty of the dstrbuton network especally mcrogrd can be acheved through the penetraton of electrc vehcles. Furthermore, the use of nducton motor load s also ncreasng and nducton motor s a threat for grd stablty ether durng start-up/ dsturbance. We are keen to normalze the transent behavour of nducton motor load through the electrc vehcles. In ths paper we have proposed a transent voltage stablty margn ndex as measure of stablty of the dstrbuton network/mcrogrd. Moreover we proposed an energy management algorthm for the dstrbuted vehcles n the mcrogrd for the purpose of load synchronzaton. The energy management algorthm and transent voltage stablty margn ndex may be used n conjuncton to lmt the number vehcles n the network and managng confguraton of the vehcles ether as load or source. In order to fully explot the dea the co-smulaton of the energy management system may be mplemented n future. Acknowledgments Fgure 4: Impact of EV load on voltage stablty Fgure 5: Impact of EV penetraton on voltage stablty Ths work was supported n part by Natonal Natural Scence Foundaton of Chna ( ), n part by the Natonal Key Basc Research Program of Chna (973 Program) (2012CB215100), n part by the Research Fund for the Doctoral Program of Hgher Educaton of Chna ( ), n part by Natural Scence Foundaton of Hube Provnce (2012FFA075) and n part by Doctoral Program of Hgher Educaton Commsson of Pakstan (HRDI/UESTP). References [1] A. G. Boulanger, A. C. Chu, S. Maxx, and D. L. Waltz, "Vehcle electrfcaton: Status and ssues," Proceedngs of the IEEE, vol. 99, pp , [2] K. Schneder, C. Gerkensmeyer, M. Kntner- Meyer, and R. Fletcher, "Impact assessment of plug-n hybrd vehcles on pacfc northwest dstrbuton systems," n Power and Energy EVS28 Internatonal Electrc Vehcle Symposum and Exhbton 10

11 Socety General Meetng-Converson and Delvery of Electrcal Energy n the 21st Century, 2008 IEEE, 2008, pp [3] J. Tomć and W. Kempton, "Usng fleets of electrc-drve vehcles for grd support," Journal of Power Sources, vol. 168, pp , [4] Y. Ota, H. Tanguch, T. Nakajma, K. M. Lyanage, J. Baba, and A. Yokoyama, "Autonomous dstrbuted V2G (vehcle-to-grd) satsfyng scheduled chargng," Smart Grd, IEEE Transactons on, vol. 3, pp , [5] W. Kempton and J. Tomć, "Vehcle-to-grd power fundamentals: Calculatng capacty and net revenue," Journal of power sources, vol. 144, pp , [6] W. Kempton and J. Tomć, "Vehcle-to-grd power mplementaton: From stablzng the grd to supportng large-scale renewable energy," Journal of Power Sources, vol. 144, pp , [7] K. M. Rogers, R. Klump, H. Khurana, A. A. Aquno-Lugo, and T. J. Overbye, "An authentcated control framework for dstrbuted voltage support on the smart grd," Smart Grd, IEEE Transactons on, vol. 1, pp , [8] H. N, G. T. Heydt, and L. Ml, "Power system stablty agents usng robust wde area control," Power Systems, IEEE Transactons on, vol. 17, pp , [9] C. W. Taylor, D. C. Erckson, K. E. Martn, R. E. Wlson, and V. Venkatasubramanan, "WACS-wde-area stablty and voltage control system: R&D and onlne demonstraton," Proceedngs of the IEEE, vol. 93, pp , [10] M. Islam and H.-H. Lee, "IEC61850 based operaton, control and management of utlty connected mcrogrd usng wreless technology," n Intellgent Computng Theores and Applcatons, ed: Sprnger, 2012, pp [11] S. Ray and G. K. Venayagamoorthy, "Realtme mplementaton of a measurement-based adaptve wde-area control system consderng communcaton delays," IET generaton, transmsson & dstrbuton, vol. 2, pp , [12] N. R. Jennngs, "An agent-based approach for buldng complex software systems," Communcatons of the ACM, vol. 44, pp , [13] T. Nagata, Y. Tao, H. Sasak, and H. Fujta, "A multagent approach to dstrbuton system restoraton," n Power Engneerng Socety General Meetng, 2003, IEEE, [14] J. M. Solank and N. N. Schulz, "Mult-agent system for slanded operaton of dstrbuton systems," n Power Systems Conference and Exposton, PSCE' IEEE PES, 2006, pp [15] S. Mohaghegh, G. K. Venayagamoorthy, and R. G. Harley, "Optmal wde area controller and state predctor for a power system," Power Systems, IEEE Transactons on, vol. 22, pp , [16] H. Zang, M. Chandorkar, and G. Venkataramanan, "Development of statc swtchgear for utlty nterconnecton n a mcrogrd," Power and Energy Systems PES, [17] M. S. Khald, X. N. Ln, J. W. Sun, Y. X. Zhuo, N. Tong, F. T. Zeng, et al., "System Modelng and Smulaton of Intentonally Islanded Reconfgurable Mcrogrd," n Advanced Materals Research, 2015, pp [18] G. Putrus, P. Suwanapngkarl, D. Johnston, E. Bentley, and M. Narayana, "Impact of electrc vehcles on power dstrbuton networks," n Vehcle Power and Propulson Conference, VPPC'09. IEEE, 2009, pp [19] H. Renmu, M. Jn, and D. J. Hll, "Composte load modelng va measurement approach," Power systems, IEEE Transactons on, vol. 21, pp , [20] W. Theodore, Electrcal machnes, drves and power systems, 6/E: Pearson Educaton Inda, [21] P. Kundur, N. J. Balu, and M. G. Lauby, Power system stablty and control vol. 7: McGraw-hll New York, [22] Y. SONG, Y. YANG, and Z. Hu, "Present Status and Development Trend of Batteres for Electrc Vehcles [J]," Power System Technology, vol. 4, p. 002, [23] P. Rchardson, D. Flynn, and A. Keane, "Impact assessment of varyng penetratons of electrc vehcles on low voltage dstrbuton systems," n 2010 IEEE Power and Energy Socety General Meetng [proceedngs], [24] M. D. Galus and G. Andersson, "Demand management of grd connected plug-n hybrd electrc vehcles (PHEV)," n Energy 2030 Conference, ENERGY IEEE, 2008, pp [25] T. Markel, M. Kuss, and P. Denholm, "Communcaton and control of electrc drve vehcles supportng renewables," n Vehcle Power and Propulson Conference, VPPC'09. IEEE, 2009, pp [26] L. Herrera, R. Murawsk, F. Guo, E. Inoa, E. Ekc, and J. Wang, "PHEVs chargng statons, communcatons, and control smulaton n real tme," n Vehcle Power and Propulson Conference (VPPC), 2011 IEEE, 2011, pp [27] W. Zhang, M. S. Brancky, and S. M. Phllps, "Stablty of networked control systems," Control Systems, IEEE, vol. 21, pp , [28] S. Shakkotta, A. Kumar, A. Karnk, and A. Anvekar, "TCP performance over end-to-end rate control and stochastc avalable capacty," IEEE/ACM Transactons on Networkng, vol. 9, pp , [29] M. S. Brancky, S. M. Phllps, and W. Zhang, "Stablty of networked control systems: Explct analyss of delay," n Amercan Control EVS28 Internatonal Electrc Vehcle Symposum and Exhbton 11

12 Conference, Proceedngs of the 2000, 2000, pp [30] J. R. Hartman, M. S. Brancky, and V. Lberatore, "Tme-dependent dynamcs n networked sensng and control," n Amercan Control Conference, Proceedngs of the 2005, 2005, pp Authors Muhammad Shoab Khald has receved Master s degree from Hanyang Unversty, South Korea n He s presently a PhD canddate n the School of Electrcal and Electroncs Engneerng at Huazhong Unversty of Scence and Technology, Chna. Hs research nterests nclude energy management of mcrogrds and power system optmzaton and control. Xangnng Ln receved a Master s and PhD degree from School of Electrcal and Electroncs Engneerng at Huazhong Unversty of Scence and Technology, Chna. Hs research nterests are modern sngle processng and ts applcatons n the power systems, power system protectve relayng and control. Yxn Zhuo s pursung Master s leadng to PhD degree from School of Electrcal and Electroncs Engneerng at Huazhong Unversty of Scence and Technology, Chna. Hs research nterests are related to wnd energy. Ramesh Kumar has receved Master s degree from Hanyang Unversty, South Korea n He s presently a PhD canddate n the School of Electroncs and Computer Engneerng at Hanyang Unversty, South Korea. Hs research nterests nclude protocol development of smart grds. Muhammad Kashf Rafque receved has receved Master s degree from UET Taxla, Pakstan n He s presently a PhD canddate n the School of Informaton and Communcaton Engneerng at Sungkyunkwan Unversty, South Korea. Hs research nterests nclude chargng optmzaton of electrc vehcles. EVS28 Internatonal Electrc Vehcle Symposum and Exhbton 12