Electric Vehicle Simulator for Energy Consumption Studies in Electric Mobility Systems

Transcription

1 Eletri Vehile Simultor for Energy Consumption Studies in Eletri Moility Systems Rirdo Mi, Mro Silv, Rui Arújo, nd Urno Nunes Astrt One of the most importnt environmentl prolems in lrge ities is the vehiulr emission. Eletri Vehiles (EVs) re growing lterntive for internl omustion engine (ICE) vehiles. Sine this kind of vehile hs low utonomy yet, it is importnt to optimize energy onsumption, for instne y plnning suitle infrstruture of ttery rehrge nd/or ttery-swith sttions. This pper presents n rhiteture for EV simultion, importnt to nlyze trffi flow, its dynmis nd the performne when there re ostrutions or intense trffi. There re severl tools for trffi simultion, SUMO (Simultion of Urn MOility) is one of them. But none of the existing trffi simultors integrtes models of EV tht llow, for exmple, perform simultion studies regrding energy onsumption. SUMO is portle open soure simultor with multi-modl trffi feture pilities tht permit the simultion of vrious types of vehiles. This work is n extension of the SUMO, two-dimensionl (2D) vehiulr simultion pkge. To llow the simultion of energy onsumption of EV, two extensions were inorported in SUMO: EV models nd modeling of ltitude, trnsforming SUMO into threedimensionl (3D) simultor. The energy model effetiveness nd orretness with 3D pilities hs een vlidted using two driving shedules (Urn Dynmometer Driving Shedule nd Highwy Fuel Eonomy Driving Shedule). This new tool will lso support the study of etter routes hoie in 3D environment with EV iming minimum energy onsumption. I. INTRODUCTION The inresing numer of motor vehiles on the streets nd rods is demnding n tive trnsport poliy. This growth lso results in trffi jms nd pperne of wide diversity of drivers with different driving ehviors, inresing the proility of idents. These re only two resons why developed ountries hve gret interest in Intelligent Trnsporttion Systems (ITS). Nowdys, petrol s high ost nd serious environmentl pollution prolems deriving from fossil urning fuel led utomotive industry to hevily investing in plug-in eletril vehiles (PHEV) s well s fully eletri vehiles (EV). In [1], Spongenerg sttes tht the numer of eletri rs on Europen rods is going to oost next yers due to oil pries, limte hnge onerns nd tough EU environment regultions. The use of EV rises severl prolems tking This work ws prtilly supported y the Portuguese Foundtion for Siene nd Tehnology (FCT) under grnt PTDC/SEN-TRA099413/2008 (EVSIM09 Projet). Rui Arújo nd Urno Nunes re with the DEEC - Deprtment of Eletril nd Computer Engineering, University of Coimr, Portugl. All uthors re with the ISR-Institute of Systems nd Rootis, University of Coimr, Portugl. Rirdo Mi nd Mro Silv lso knowledge PhD fellowships grnted y FCT, respetively SFRH/BD/44644/2008 nd SFRH/BD/38998/2007. Mro Silv is lso with IPC-ISEC - Polytehni Institute of Coimr, Coimr, Portugl. emil: {rmmi, msilv, rui, urno}@isr.u.pt into ount energy issues nd trffi nlysis, suh s route plnning, street onnetivity nd diretions, nd plement of rehrging sttions. Modeling nd simultion methods re essentil elements in design nd opertion of trnsporttion systems. Severl resons justify the simultion tsk. Constrution osts n e minimized with prior simultion; nlysis my e done with minimum risk; dynmi nlysis n e mde without need of prototype onstrution; simultion nlysis n e mde in the design phse of the system t frtion of the ost of onstrution. Trnsporttion systems re the kone onneting the vitl prts of ity / region nd therey the indepth understnding of the trnsporttion system omponents is essentil for the plnning, design nd opertionl nlysis of the ity / region. The EV energy onsumption n e redued y mny wys, nmely y hoosing est routes. Alves et l. [2] uses the nt olony optimiztion lgorithm to improve route hoie. In [3], the routing prolem is hrterized using multi-gent systems. The gent s tions my use internl stte informtion out the vehile itself suh s vehile size, top speed or torque. A rel-time rpooling system using Djisktr s lgorithm with n ojetive funtion omining witing time nd trveling time is proposed in [4]. In [5], Sgheier et l. report n rhiteture for dt olletion nd nlysis performne from EV. Relted works to route improvement pply some kind of omputtionl intelligene, like geneti lgorithm nd fuzzy logi [6], [7], [8], [9] ut none of them ddress issues relted with EV, like performne, rnge nd route optimiztion, iming minimum energy onsumption. EV employ regenertive reking tehnology, whih llows the onversion of kineti energy into eletril energy when the vehile is slowing down or is driving downhill. With regenertive reking, the eletri drive motor lso funtions s genertor, supplying energy k to the tteries. To simulte this hrteristi, trffi simultors environment need to e three-dimensionl, e.g., ltitude hs to e known nd must e represented in the environment model. However, known trffi simultors re two-dimensionl, i.e., the mps lie in Crtesin x-y projetion, nd therefore re not pproprite to simulte regenertive reking. This pper desries omponents tht extend the 2D trffi simultor pkge SUMO (Simultion of Urn MOility) [10] in 3D simultion environment for eletril vehiles. The EV model used here follows losely the formultions desried in [11]. It ws developed to provide mens for onduting studies of eletri moility in urn res. The model ws inserted into the r-following model proposed

2 TABLE I: Physil onstnts. onstnts mening g elertion due to grvity ρ ir density µ rr oeffiient of rolling resistne A vehile frontl re C d drg oeffiient α ngle of slope or hill m vehile mss v vehile veloity vehile liner elertion I moment of inerti of rotor of the motor G ger rtio of the system r rdius of the tyre ger system effiieny η g y Kruß [12], [13], ut n e esily dpted for other rfollowing models. The pper is orgnized s follows. Setion II estlishes the EV model with fous on the energy onsumption omponent. Setion III presents the simultion softwre nd its modifitions. Setion IV desries the simultion senrio nd simultion relevnt results. Finlly, in Setion V, some onluding remrks re drwn. II. MODEL DESCRIPTION An EV is omplex system inluding severl susystems, suh s: mehnil, eletril, ontrol, mgneti, pneumti, eletrohemil nd hydruli, et. In this work, most susystems re strted, nd only those needed to provide mehnil nd eletril trtion to vehile will e hrterized. A. Mehnil Trtion The fore needed to provide mehnil trtion to propel the vehile forwrd is the trtive effort. This fore hs to overome the rolling resistne, erodynmi drg, hill liming fore, the fore to elerte the vehile nd the fore to provide ngulr elertion to the trtion motor. Thus, the trtive effort n e expressed s [11]: where: F te = F rr + F d + F h + F l + F w (1) F rr = µ rr mg (rolling resistene fore); (2) F d = 1 2 ρac dv 2 (erodynmi drg); (3) F h = g sin(α) (vehile s weight omponent); (4) F l = m (fore required to give liner elertion); (5) F w = I G2 (fore required to give rottionl η g r2 elertion to the trtion motor); (6) Sine frequently the motor s moment of inerti is not known, in these ses it is resonle to inrese the vehile s mss y 5% in (5) nd ignore F w. The physil quntities ppering in (2) to (6) re physil onstnts, or relted to vehile s physil hrteristis, nd their mening re stted in Tle I. It should e noted tht EV hve regenertive rking feture. This mens tht, if F te is negtive, the trtive fore will not e pplied from the eletril motor to the wheels, ut from the wheels to the motor; nd the urrent will flow into the ttery, hrging it. F rr nd F d re frition fores nd they must e s low s possile to minimize energy onsumption, whih is hieved with good design reduing µ rr, A nd C d. Both F rr nd F d re non-negtive; so, only F h nd F l, together, re le to mke F te negtive. F h will e negtive when the vehile is going downhill (α < 0), nd F l will e negtive when the vehile is slowing down ( < 0). The mehnil energy required to move the vehile is E te = F te v dt (7) so tht the energy tken from ttery to e supplied to the trtion motor to provide E te (7), is E te, driven se, (8) η E mot in = m η g E te η m η g, regenertive se. (9) where η m nd η g re the motor nd the ger system effiienies, respetively. When the vehile is eing driven, it holds (8); ut if the motor is eing used to slow the vehile down, the effiieny works in the opposite sense, supplying energy to the ttery, nd (9) tkes ple. Finlly, it must e onsidered ll other vehile s eletril systems (lights, heting, ooling, inditors, rdio, et.), E, whih shll e dded to the motor energy. Thus, the totl energy required from the ttery is B. Eletril Trtion E t = E mot in + E (10) When in driven se, the power required from the motor to mke the vehile run t ertin speed is supplied from the ttery. On the other hnd, in regenertive se, the urrent flows into the ttery. The urrent tht flows from/into the ttery is expressed y: I = V o V 2 o 4R in P t 2R in, driven se, (11) V o + V 2 o + 4R in P t 2R in, regenertive se. (12) where V o is the open iruit voltge from the ttery, R in is its internl resistne, nd P t is the power produed y the urrent. As the motor drins urrent from ttery, wht is relly needed to e known is how ttery dishrges while

3 Fig. 2: Intertion etween SUMO s trffi module nd energy module. Fig. 1: Simultion proess with SUMO. the EV is moving. The depth of dishrge (DoD) is given y DoD = CR C p (13) where C p denotes the Peukert Cpity nd CR is the hrge removed. If in regenertive se CR = I dt (14) otherwise, in driven se CR = where k denotes the Peukert Coeffiient. III. SIMULATION PACKAGE I k dt (15) The onsumption model desried in the previous setion ws implemented s module of SUMO. Here, we desrie few fetures of SUMO nd the modifitions mde to inorporte new fetures. The min fetures of SUMO trffi simultor re: support different vehile types, ple of hndling lrge rod networks, hndle i-dimensionl networks, omputtionlly fst. Its vehiles flow s model (r-following model [12]) is sed on mirosopi routes, where eh vehile is treted individully with its own route. The model is lso ontinuous in spe nd disrete in time. The pkge is open soure, liensed under the GPL (Generl Puli Liense) nd highly portle. A. Modifitions To simulte energy onsumption of EVs, SUMO s vehile lss hs een modified to reeive ttriutes nd methods to implement the funtionlities explined in Setion II. Moreover, the energy n e regenerted when the vehile slows down or goes downhill. Thus, the i-dimensionl rod network ws hnged to reeive the z oordinte, relted to the elevtion of the network nodes. Fig. 1 shows shemti of the min proesses nd dt files involved in the simultion proess using SUMO [14]. A new file, XML ltitudes, ws dded in order to llow the speifition of rod elevtion. This wy, the NETCONVERT tool, gets the extr z ttriutes from XML ltitudes file nd outputs the rod network (XML network) whih inorportes the elevtion informtion. EV physil prmeters nd routes informtion re provided to the SUMO simultor through XML onsumption nd XML route files, respetively. Algorithm 1 Energy Model Pseudo-ode. 1: Reeive from CFM vlues of next veloity v i+1, lst vlue of veloity v i nd elertion ; 2: Clulte torque T needed to pply those elertion nd veloity; 3: if (T > T mx ) then 4: Clulte new v i+1 nd vlues to reh T mx ; 5: end if 6: Clulte P mot ; 7: P t = P mot + P ; 8: Clulte ttery urrent I to provide P t, using (11) or (12); 9: Clulte DoD i+1, the DoD in the next time step; 10: if (DoD i+1 > DoD limit ) then 11: Re-lulte I nd P t vlues; 12: Re-lulte v i+1 nd vlues; 13: Re-lulte DoD i+1 ; 14: end if 15: Return v i+1 ; The intertion etween SUMO s r-following module (CFM) nd the new EV energy module is illustrted in

4 TABLE II: Bttery Pk Prmeters. Mnufturer: Ovoni Energy Produts Type: Nykel Metl Hydride Numer of Modules: 26 Weight of Module: 18.3 kg Weight of Pk(s): 481 kg Nominl Module Voltge: 13.2 V Nominl System Voltge: 343 V Nominl Cpity: 77 AH Stored Energy: 26.4 kwh Fig. 2. In eh time step of the SUMO simultion proess, the desired vehile veloity is determined depending on the veloity limit of the rod, nd the trffi demnd represented y the distne to the vehile hed nd its veloity. For the r-following module to provide the desired veloity, the energy module verifies eh time step if there is enough energy in ttery to produe the required power. For this propose, the torque is lulted y T = F te r (16) G where F te is given y (1), r is the rdius of the tyre, nd G is the ger rtio. If (16) is higher thn the mximum torque, T mx, whih the EV motor n provide then new veloity nd elertion vlues re lulted suh tht F te produes T mx. Next, E t is lulted y (10) nd it is verified if there is enough energy in the ttery to supply to the motor. If not, new veloity nd elertion vlues re lulted sed on the informtion of the existing residul energy in the ttery. Algorithm 1 summrizes the Energy Model s proessing nd omputtions. A. The vehile IV. SIMULATION SCENARIO The simultion experiments the vehile model uses the prmeters of the EV1 eletri vehile produed y Generl Motors from 1996 to This model ws hosen sine it hs een widely used in previous studies. The EV1 prmeters vlues re: A = 1.89 m 2, C v = 0.19, µ rr = 0.005, length = 4.31 m, mximum elertion = 3.08 m/s 2, mximum deelertion = 1.0 m/s 2 nd mximum speed of 129 km/h. The onsidered density of the ir ws ρ = 1.25 kg/m 3. The motor nd the ger system effiienies, η m nd η g respetively, were tken from [15]. The internl resistne R in nd open iruit voltge V o of the ttery pk were modeled ording to [16], [17] nd [18]. The ttery pk min prmeters te e shown in Tle II. B. The Rod Network Two types of tests were performed to verify the rnge of the simulted vehile: driving yles tests [11] nd onstnt speed tests. Two driving yles speified y the U.S. Environmentl Protetion Ageny [16] were used: the Urn Dynmometer Driving Shedule (UDDS), representing ity driving onditions; nd the Highwy Fuel Eonomy Driving Shedule (HWFET) representing highwy driving onditions TABLE III: Rnge with NiMH Bttery. Drive CEPA Shedule report SUMO vrition UDDS km km 0.14% HWFET km km 3.68% Depth of Dishrge (%) Bttery Voltge (V) Fig. 3: UDDS: DoD evolution (19 yles) nd ttery pk voltge. under 96 km/h. Tle III shows the rnges of EV1 reported y CEPA [18] nd rnges otined y the simultions performed in SUMO extended with the EV nd elevtions models here proposed. These results show lose mthing etween the developed models implemented in SUMO nd the CEPA dt. The DoD (13) nd the voltge drop during the 19 omplete UDDS dishrge yles is plotted in Fig. 3. In Fig. 4, depits the power supplied y the ttery pk for the first UDDS yle. The mximum drined power nd the mximum regenerted power vlue is 34 kw nd 21 kw, respetively. These vlues re within the motor/inverter pilities. The dishrged yles were pplied to 95% DoD to not exeed the ut-off voltge. The 3D iruit shown in Fig. 5 ws used in the onstnt speed tests. The iruit onnets three stright-line rod segments. The first rod segment hs positive slope of 3.24%. The seond rod segment onnets points nd with 0% slope. Finlly, rod segment of negtive slope of 6.68% links points nd. After the

5 Delivered / Regenerted Power (kw) m (6.48%) m (0%) m (3.24%) () Fig. 4: First UDDS yle power request. TABLE IV: Constnt speed rnge senrio. Speed (slope / no slope) Rnge 60km/h (slope) 60km/h (no slope) 80km/h (slope) 80km/h (no slope) 260km 430km 196km 302km Stte of Chrge () trjetory the vehile is driven k to point ut following the opposite wy. This sequene ws repeted until the empty hrge ondition ws ttined (DoD limit = 0.95). Figures 5-5 represent the evolution of the Stte of Chrge (SoC = 1 DoD) in onstnt speed profile through the 3D rod test network. Though the EV hs onstnt speed, the urrent is not onstnt, s n e seen in Fig. 5. In ft, s the yles re repeted, the SoC dereses due to the energy drined from the ttery pk. The ttery pk voltge lso dereses (Fig. 3), whih explins the ft tht, to hieve the sme required power to mintin onstnt speed vlue, n inrese is required in the onsumed urrent. This hppens even in zero slope rod segment. In the first pss of the vehile in the diretion the urrent mesured t point is 57A, nd in the orresponding point t the lst pss the urrent is 59A s highlighted in Fig. 5. Tle IV summrizes results for two situtions onsidering the route defined in Fig. 5: () with slopes s speified in the figure, nd () with ll three segments eing horizontl. In oth ses it ws onsidered tht the vehile rried pssenger with 75kg weight. In zero slope nd low speed tests, the two mjor influene influening rnge re the rolling resistne nd the erodynmi drg. The EV1 rnge t onstnt speed of 72 km/h reported in [16] is 355 km whih is in the rnge of otined vlues in our zero-slope tests (430 km for 60 km/h nd 302 km for 89 km/h). The regenertive reking effet is lerly oserved in the Fig. 5 in the rod segments with negtive slope. Finlly, the model ws pplied on simultes sunetwork of the Coimr ity. The trget re, iruit of pproximtely 8 km length, is illustrted in Fig. 6. The Bttery urrent (A) () Fig. 5: () 3D rod test network, () evolution of the SoC long the omplete test, nd () zoomed nd overlid urrents t the first (red) nd lst (lue) psses of the vehile over the segment. strting/ending point of the iruit is represented y red dot on the mp. The mximum differene of ltitude in the overed iruit is 54.4 m. As shown in Fig. 7, the iruit is diversified in terms of ltitude llowing explortion of the regenertive reking hrteristi of the vehile. For this study, one EV ws injeted t the strting lotion, nd run long the route speified y the rrows. There ws no other trffi long the iruit. To prove the effet of ltitudes over energy onsumption, two different tests were performed: 1) Using originl mp, with the true ltitudes; 2) Using plnified mp, with no ltitude differenes. Both simultion tests onsisted of one omplete turn long the iruit. The orresponding energy results re shown in

6 wy the ttery dishrges over time (with hrging periods ourring in negtive slope segments or vehile deelertion periods). The EV onsumption model ws vlidted with two types of well-known driving yles nd in onstnt speed mode. This model hs een lso pplied on su-network re of Coimr ity. With the enhnements reported in the pper, SUMO frmework ws endowed with suitle tools tht llow lrge sle simultion of eletri moility systems. REFERENCES Fig. 6: Rod iruit simulted (soure y Google). Altitude (m) iruit eginning Altitude profile iruit end Position (m) Fig. 7: Altitude profile long the iruit. TABLE V: Rnge in urn senrio, one yle one pssenger (75kg) top speed DoD 60km/h (3D) 3.27% 60km/h (2D) 2.15% 80km/h (3D) 4.21% 80km/h (2D) 3.12% Tle V. As n e seen, the ltitude signifintly influenes onsumption. The energy is drined differently depending on the ltitude profile of the iruit: with ltitudes the onsumption is higher then with the flt mp. V. CONCLUSION A simultion frmework for eletri vehiles in terms of energy onsumption hs een presented. SUMO trffi simultor ws enhned with omponents tht llow 3D simultion of EV energy onsumption. EV performne depends on the terrin slope, with diret impt on the [1] H. Spongenerg, Euoserver / eu sttes plug in to eletri rs, http: ////euoserver.om/882/26594, August 2008, retrieved [2] D. Alves, J. vn Ast, Z. Cong, B. D. Shutter, nd R. Bušk, Ant olony optimiztion for trffi dispersion routing, 13th Interntionl IEEE Conferene on Intelligent Trnsporttion Systems (ITSC), pp , Sep [3] S. Boskovih, K. Borioonsomsin, nd M. Brth, A developmentl frmework towrds dynmi inident rerouting using vehile-tovehile ommunition nd multi-gent systems, 13th Interntionl IEEE Conferene on Intelligent Trnsporttion Systems (ITSC), pp , Sep [4] V. Suresh, G. Hill, P. T. Blythe, nd M. Bell, Smrt infrstruture for ron foot print nlysis of eletri vehiles, 13th Interntionl IEEE Conferene on Intelligent Trnsporttion Systems (ITSC), pp , Sep [5] M. Sghier, H. Zgy, S. Hmmdi, nd C. Thon, A distriuted dijkstr s lgorithm for the implementtion of rel time rpooling servie with n optimized spet on silings, 13th Interntionl IEEE Conferene on Intelligent Trnsporttion Systems (ITSC), pp , Sep [6] Y. Chen, M. G. H. Bell, nd K. Bogenerger, Relile pretrip multipth plnning nd dynmi dpttion for entrlized rod nvigtion system, IEEE Trnstions on Intelligent Trnsporttion Systems, vol. 8, Issue: 1, pp , [7] H. T. Msy Yoshikw, Cr nvigtion system sed on hyrid geneti lgorithm, World Congress on Computer Siene nd Informtion Engineering, [8] B. Chkrorty nd R. C. Chen, Fuzzy-geneti pproh for inorportion of driver s requirement for route seletion in r nvigtion system, IEEE Interntionl Conferene on Fuzzy Systems, pp , [9] A. J. S. Kumr, J. Arundevi, nd V. Mohn, Intelligent trnsport route plnning using geneti lgorithms in pth omputtion lgorithms, Europen Journl of Sientifi Reserh, vol. 25 (3), pp , [10] SUMO, Simultion of urn moility, pps/mediwiki/sumo/, Jnury 2010, retrieved ( [11] J. Lrminie nd J. Lowry, Eletri Vehile Tehnology Explined. John Wiley & Sons, [12] S. Kruß, Mirosopi modeling of trffi flow: Investigtion of ollision free vehile dynmis, Ph.D. disserttion, Mthemtishes Institut, Universität zu Köln, [13] S. Kruß, P. Wgner, nd C. Gwron, Metstle sttes in mirosopi model of trffi flow, Phys. Rev. E, vol. 55, no. 5, pp , My [14] SUMO, Sumo user doumenttion, mediwiki/sumo/index.php?title=sumo User Doumenttion, Jn 2010, retrieved [15] A. Cmpell, A. Rengn, nd J. Steffey, The simultion of 42-volt hyrid eletri vehiles, Progrms/ grdute/msim/msimprojetreports/mcp2.my.2001.report.do, retrieved [16] U.S. Deprtment of Energy, 1999 generl motors EV1 w/nimh, fsev/ev results/ev1 ev.pdf, retrieved [17] S. Goluff, Optimiztion of plug-in hyrid eletri vehile, Mster s thesis, Georgi Institute of Tehnology, the Netherlnds, [18] A. R. Bord, 2000 zero emission vehile progrm - stff report, http: // retrieved