ENERGY 0 : The Frst Internatonal Conference on Smart Grds, Green Communcatons and IT Energy-aware Technologes Fruency response from electrc vehcales Janzhng Wu, Janaka Ekanayake, Kamalanath Samarakoon Insttute of Energy Cardff School of Engneerng Cardff, UK e-mal:wuj5@cf.ac.uk, EkanayakeJ@cf.ac.uk, kamalanath@eee.org Abstract The contrbuton of plug-n Electrc Vehcles (EV) for fruency response was nvestgated. 4-hour EV load profles obtaned from a probablstc approach was used. Three tme dependent chargng modes were consdered. A sngle bus model of Great Brtten power system was used for smulatons. Smulatons were carred out for a day wth low demand. The smulaton results shows that sheddng EVs that are chargng can reduce the fruency excurson sgnfcantly. Keywords - Fruency response, Electrcal vehcles, Smart meter, Vrtual power plant, Power system nerta I. INTRODUCTION A number of countres have taken specfc polcy ntatves to encourage renewable power generaton and for ntroducng electrc vehcles as they contrbute to decarbonse ther electrcal energy and transport sectors. For example, n the UK, 5% of all energy s to be suppled by renewable energy by 00. Ths translates nto 30-40% of electrcal energy beng generated from renewable sources. It s antcpated that a large proporton of ths power wll come from wnd power. Perhaps up to 40 GW of wnd turbne generaton on a Great Brtan (GB) system wth a total of around 00 GW of generatng plant. The UK government also has plans to cut emssons from domestc transport by 4% on 008 levels by 00. The uncertanty brought by varablty of renewable energy generaton wll ntroduce a number of concerns over operaton of the power system. A very hgh penetraton of renewable energy sources demands consderable ncrease n fruency response and reserve that the system operator should mantan to ensure fruency performance wthn the control lmts [][]. In ths paper, the flexblty offered by plug-n electrc vehcles (EV) by removng ther chargng load mmedately after a fruency event s nvestgated. It s antcpated that ths wll enable operaton of the future power system wth current level of reserve margns. The paper s organzed as follows: ntally the fruency control n the GB system s dscussed; secondly ntegratng EVs for fruency studes s dscussed; then modelng of the GB system wth EV for fruency studes are dscussed and fnally results are presented. II. FREQUENCY CONTROL IN THE GB SYSTEM Fruency s determned and controlled by balancng system demand and total generaton. The nomnal fruency of the GB system s 50Hz. If the demand s greater than the generaton, the fruency falls below 50Hz. Conversely, f the generaton s greater than the demand, the fruency rses above 50Hz. In practce, the fruency vares around 50Hz by a small amount as the system demand contnuously changes. When there s a sgnfcant power mbalance of the system, the fruency wll show a large devaton. The Electrcty Supply Regulatons rure the system fruency to be mantaned at 50Hz ±% []. The Transmsson Lcense places an oblgaton on the Natonal Grd Company (NGC) to plan and operate the system to ensure complance wth the Electrcty Supply Regulatons []. To meet these oblgatons the system s desgned to accept the largest credble loss of 30MW of generaton (two of the largest generators, 660MW, on the system) and s operated to the followng fruency contanment polces: System fruency under normal operatng condtons wll be mantaned wthn the operatonal lmts of 50±0. Hz (NGC s current practce), For a sudden loss of generaton or demand up to 300MW, the maxmum fruency change wll be lmted to ±0.Hz, For a sudden loss of generaton or demand greater than 300MW and less than or ual to 000MW, the maxmum fruency change wll be lmted to ±0.5Hz, For a sudden loss of generaton greater than 000MW and less than or ual to 30MW, the fruency change wll be lmted to 0.8Hz wth fruency restored to 49.5Hz wthn mnute. Any loss of generaton greater than 30MW wll be treated as an emergency condton as t may cause the system fruency to fall below 49Hz. Automatc low fruency load sheddng arrangements usually commence at 48.8Hz. In the event that the fruency s above 5Hz or below 47Hz, the ndependent protectve actons are permtted to protect generators aganst danger to plant and/or for personnel safety. A typcal fruency transent for a generaton loss of 30MW s shown n Fgure [3][4]. Copyrght (c) IARIA, 0. ISBN: 978--608-36-6 48
ENERGY 0 : The Frst Internatonal Conference on Smart Grds, Green Communcatons and IT Energy-aware Technologes Fruency (Hz) 50. 49.8 49.5 49. Contnuous servce Even 0 s 30 s 60 s O X Prmary Occasonal Secondary to 30 Fgure. Typcal fruency devaton followng a loss of 30MW generaton [3] Even though EVs are consdered as an deal choce for provdng prmary and secondary response, they have a negatve effect durng phase OX (see Fgure ). As modern wnd turbne generators and EVs are connected to the grd through a power electronc nterface, they wll not contrbute to the system nerta. Ths reducton n the overall system nerta and the ncrease of demand wll lead to rapd change n fruency durng phase OX. III. INTEGRATING EVS FOR FREQUENCY SERVICES An EV could partcpate n low fruency response servces manly n two ways. Easest approach would be to swtch off all EVs that are chargng. Ths wll ntroduce a proportonal reducton n load, thus reducng the fruency excurson. In an event of a hgh fruency event all the plugn EVs that are n stand-by mode wth the state of charge of battery s less than 00% could be charged thus addng an addtonal load to the grd. EVs could also support the grd by actng as an energy store. For example durng a low fruency event, EVs could dscharge ts stored energy thus actng as a dstrbuted energy source. Ths paper concentrates on former aspect that s dsconnectng a chargng fleet of EVs durng a fruency event. The way EVs could partcpate for fruency servces depends on the grd operator. For example, n the UK the fruency response servces such as Frm Fruency Response (FFR) and Fruency Control Demand Management (FCDM) allow demand sde partcpaton n prmary fruency control. The large-scale consumers are contracted n advance to swtch OFF ther loads (more than 0 MW and 3 MW n FFR and FCDM respectvely) durng a fruency excurson. In FFR and FCDM schemes, the contracted consumer should reduce load wthn 30 sec and sec and mantan 0mn and 30 mn respectvely. A. Indvdual EV on a fruency-responsve swtch The demand sde support for fruency reserve by controllng loads was proposed as early as n 980 [5]. The paper proposed a fruency-responsve swtch whch controls sgnfcant energy consumng ndustral or domestc loads. A smlar swtch could be utlzed at each EV to swtch them off when a fruency excurson occurs. Recently, Smart Meter (SM) has drawn wde attenton as a devce whch can help to save energy and to mprove the effcency of a power system. A number of ntatves that Tm deploy smart meters are reported n [6]. The SM has a twoway communcaton between the suppler and also wth domestc applances ncludng EVs connected to home area network (HAN). On the recept of a sgnal from supplers, the SMs could send sgnals to the control unts to shed the EV mmedately. B. EV as a vrtual power plant (VPP) Several studes have dentfed the potental of EVs to partcpate n the electrcty markets [7][8]. As power capabltes of an ndvdual EV s rather small, ther partcpaton n the electrcty markets wll rure a new entty: the EV Suppler/Aggregator (EVS/A). The EVS/A wll serve as an ntermedary between a large number of EVs and market players and/or system operators [9]. The role of the EVS/A s to cluster geographcally dspersed EVs, and manage ther generaton and demand portfolos as a sngle entty. The Vrtual Power Plant (VPP) concept s an aggregaton model whch ams to overcome the challenges of Dstrbuted Energy Resources (DER) ntegraton and enable ther market partcpaton. The VPP concept s consdered as an deal canddate for EVS/A. Fgure shows how the VPP nteracts wth the system operators (DSO and TSO). Upon recognzng a fruency excurson, the system operator could nstruct the VPP to shed some or all of the chargng EVs. Fgure. Integraton between the VPP and grd operators [0] IV. MODELLING OF POWER SYSTEM AND EVS FOR FREQUENCY STUDIES A. Power system representaton Assumng a coherent response of all generators n the system to changes n the load, the power system can be represented by an uvalent generator []. The uvalent generator has an nerta constant H and calculated usng the followng uaton: Copyrght (c) IARIA, 0. ISBN: 978--608-36-6 49
ENERGY 0 : The Frst Internatonal Conference on Smart Grds, Green Communcatons and IT Energy-aware Technologes S = () H H * = coal, gas,... S sys where H and S are the nerta constant and MVA ratng of the ndvdual power plant. A smple model representng the nerta and dampng of the GB system wthout the contrbuton due to governor acton of synchronous generaton s shown n Fgure 3. In the model Pm refers to change n mechancal power of all the generators on the GB system and PL s any change n total load. The dampng provdng by rotatng loads s lumped nto a sngle dampng constant D. speed. Ths changes the governor valve (steam turbne) or gate (hydro turbne) poston. The typcal governor actuator tme constant, T G, s 0. second. For a stable performance of the speed control, a transent-droop-compensaton, whch s a lead-lag transfer functon wth tme constants T and T, s ntroduced between governor and turbne. The turbne relates the response of mechancal power output followng the governor acton and s characterzed by a tme constant T T whch vares between 0.3 and 0.5 second. R P m + D + H s f + st G + st + st Y + st T Pm H s + D ω r = f P L P L Fgure 3. The system uvalent for fruency control analyss B. Turbne-governor model The composte power/fruency characterstc of the power system depends on the combned effect of the droops ( R, R, ---, R n ) of all generator speed governors. It also depends on the fruency characterstcs of all the loads n the system. For a system wth the n generators and a composte load-dampng constant of D, the steady-state fruency devaton fss followng a load change PL s gven by uaton (). PL PL f ss = = () ( / R + / R + + / Rn ) + D / R + D where, the composte governor speed droop can be wrtten as uaton (3): R / R + / R + = (3) / R The typcal speed droop settng for both thermal and hydro generator governors s around 5% n per unt value. Thus, a system (as above) wth a number of machnes, each wth a droop of 5%, wll have a total system speed droop R of 5%. However, the actual speed droop may range from % to %, dependng on the dfferent types of unt []. Takng account of the characterstcs of steam and hydro turbnes n the system, a system turbne-governor model shown n Fgure 4 can be derved. The speed control of the turbne s provded by a droop governor wth an uvalent gan value, R. It operates on an nput of the speed devaton formed between the reference speed and the actual n Fgure 4. Block dagram of a system turbne-governor model The parameter values of the sngle generator model shown n Fgure 3, were obtaned through parameter dentfcaton and model valdaton. A severe fruency event shown n Fgure 5 whch occurred n the UK on 7th May 008 was used for valdaton. Fgure 5. Fruency response on 7th May 008 (http://www.natonalgrd.com/nr/rdonlyres/d680c70a-f73d-4484- BA54-95656534B5D/697/PublcReportIssue.pdf) C. EV representaton To nvestgate the mpact of the EV partcpaton, a vtal step s how to obtan realstc 4-hour EV load profles. Dfferent from conventonal load forecastng, there s no hstorcal EV use data avalable for reference. A feasble way s to generate the profles va reasonable predctons and assumptons of EV market penetraton, techncal specfcatons, and use patterns (especally chargng patterns), etc. Copyrght (c) IARIA, 0. ISBN: 978--608-36-6 50
ENERGY 0 : The Frst Internatonal Conference on Smart Grds, Green Communcatons and IT Energy-aware Technologes To determne regular EV chargng profle a probablstc approach was used. More detals of ths approach could be found n []. In ths study three tme dependent chargng modes, namely, after-work chargng, on-work and after-work chargng and delayed nght chargng were consdered. The EV power demand profles for 00 were obtaned for three chargng modes and shown n Fgure 6. The nformaton used n ths study n manly from reference [6]. usng a fruency senstve swtch. Ths model was mplemented n MATLAB/Smulnk. 00 000 After-work charge mode Delayed nght charge mode On-work plus after-work charge mode EV Power Demand (MW) 800 600 400 00 0 0 4 6 8 0 4 6 8 0 4 t(hour) Fgure 6. EV power demand profle for three chargng modes consdered D. The GB system model The model shown n Fgure 4 was used to nvestgate the contrbuton of EV for the fruency response. H was calculated for 00 system assumng the generaton schedule and nerta constants shown n Table. TABLE I. Generator type OVERALL INERTIA CONSTANT OF THE GB SYSTEM Assumed Capacty GW H H New Coal.4 6.0 0.3 Coal 9.30 6.0 0.88 Gas 5.0 9.0.3 Nuclear 6.00 4.5 0.43 Interconnector 3.30 0.0 0.00 Other 4.76 6.0 0.45 Onshore wnd 5.7 0.0 0.00 Offshore wnd 3.68 0.0 0.00 Other renewables 3.36 6.0 0.3 Total 63.54 4.44 Fgure 7 shows the model used to nvestgate the EV fruency response for the 00 GB system. It was assumed that mmedately after a fruency event s detected, EVs whch are chargng on the power system are dsconnected Fgure 7. GB system model to nvestgate EV fruency response V. RESULTS Based on the obtaned EV power demand profles and the developed GB system model, the contrbuton from EVs for fruency response was studed. Fgure 8 shows the EV contrbuton to the prmary fruency response under dfferent EV loads and low total demand condton. It s assumed that demand n 00 s as same as n 008 due to electrcty network effcency mprovement actvtes n GB. The demand used n the smulatons s the mnmum summer GB demand n 008 plus the three EV load profles shown n Fgure 6. It s assumed that the EV loads wll be dsconnected as soon as fruency starts to drop. VI. CONCLUSIONS The mportance of fruency response from EV n a regme where there s a hgh penetraton of renewable energy generaton s demonstrated n ths paper usng computer smulatons. A sngle bus GB system was used for smulatons. The system nerta was determned to reflect the hgh penetraton of renewable energy sources. EVs were modeled usng ts demand curve over a day. Three chargng modes namely, after-work, on-work and after-work, and delayed-nght chargng were used to construct the EV demand curves. Durng a low fruency event, the chargng load was shed to provde fruency support. Smulatons show that EV load has a sgnfcant contrbuton to reduce the fruency excurson. Hghest effect was obtaned wth delayed ngh chargng mode, f a fruency event occurs at md nght. At ths tme fruency drop can be reduced from.% to 0.%. ACKNOWLEDGMENT Authors wsh to thank EU project MERGE for provdng fnancal support for ths work. Authors also wsh to thanks Dr. X. Chu for provdng EV load profles for ths study. J.B.E thanks the Low Carbon Research Insttute for ther support. Copyrght (c) IARIA, 0. ISBN: 978--608-36-6 5
ENERGY 0 : The Frst Internatonal Conference on Smart Grds, Green Communcatons and IT Energy-aware Technologes REFERENCES [] Strbac, G., Shakoor, A., Black, M., Pudjanto, D., and Boppc, T., Impact of wnd generaton on the operaton and development of the UK electrcty systems, Electrc Power Systems Research Vol. 77, 007, pp. 4-7. [] Pearmne, R., Song, Y.H., and Chebbo, A., Influence of wnd turbne behavour on the prmary fruency control, IET Renewable Power Generaton, 007, Vol., Issue, pp. 4-50. [3] Ernmez, I.A., Bckers, D.O., Wood, G.F., and Hung, W.W., NGC Experence wth Fruency Control n England and Wales- Provson of Fruency Response by Generators, IEEE PES Wnter Meetng, 3 January 4 February, 999, New York, USA. [4] NGC, Techncal and Operatonal Characterstcs of the Transmsson System, Aprl 000; [5] Fred C. Schweppe, Rchard D. Tabores, James L. Krtley, Hugh R. Outhred, Federck H. Pckle, and Alan J. Cox Homeostatc Utlty Control, IEEE Transactons on Power Apparatus and Systems, vol. PAS-99, no. 3, May/June 980. [6] K. Samarakoon, J. Ekanayake, and N. Jenkns Development of Techncal Concepts of DSP nto smart meters, Centre for Dstrbuted Generaton and Sustanable Electrcal Energy, Cardff Unversty, December 008. [7] Kempton, W., and Tomc, J., Vehcle-to-grd power fundamentals: Calculatng capacty and net revenue, Journal of Power Sources, 005. [8] Brooks, A., and Gage, T., Integraton of electrc drve vehcles wth the electrc power grd a new value stream, 8th Internatonal Electrc Vehcle Symposum and Exhbton, Germany, 00. [9] Morera, C. L., Rua, D., Karfopoulos, E., Zountourdou, E., Soares, F., Bourth, I., Grau, I., Peças Lopes J. A., Cpcgan L.M., Seca, L., Moschaks M., Rocha Almeda P. M., Mouts P., Papadopoulos P., Re R. J., Bessa R. J., and Skarvels-Kazakos S. Extend Concepts of MG by Identfyng Several EV Smart Control Approaches to be Embedded n The Smartgrd Concept to Manage EV Indvdually or n Clusters, Moble Energy Resources n Grds Of Electrcty, Delverable D., September 00. [0] Delverable D.3, Identfyng EV smart control approaches under a VPP management, EU project on MERGE, http://www.evmerge.eu/mages/stores/uploads/merge_wp_d.3_fnal.pdf, Access on March 0. [] Kundur P., Power System Stablty and Control, McGraw-Hll, Inc., 994 [] Draft of Delverable D., Functonal specfcaton for tools to assess steady state and dynamc behavour mpacts, mpact on electrcty markets and mpact of hgh penetraton of EV on the reserve levels EU project on MERGE, http://www.evmerge.eu/ndex.php?opton=com_content&vew=category&layout =blog&d=64&itemd=53, Access on March 0. (a) Fruency drops at 00.00 hrs (b) Maxmum fruency drops at every 0 mn durng a day Fgure 8. The graphs when 30 MW generaton s lost n 00 GB system (wth mnmum demand). Copyrght (c) IARIA, 0. ISBN: 978--608-36-6 5