Allocation of Electric Vehicles' Parking Lots in Distribution Network

Transcription

1 Allocation of lectric Vehicles' Parking Lots in Distribution Network M. H. Aini, Stuent eber, I, an A. Isla, Meber, I Abstract Increasing growing of electricity ean an environental issues bring huge incentives to electric vehicles (s) arket. High penetration of these vehicles leas to soe challenges for the future power syste. Specifically, utilizing the in an unscheule anner will result in istribution network technical issues such as reliability issues, congestion an loss enhanceent. In this paper, firstly, a probabilistic oeling of s' behavior is achieve. Seconly, reliability-constraint optiu allocation of s' parking lots is propose. To achieve this purpose, a coposite objective function is efine. This function contains three significant istribution network reliability inices an use to allocate parking lots by eans of enhancing reliability. Finally, the propose etho is ipleente on a stanar istribution syste for RBTS bus. Inex Ters oelling, parking allocation, coposite reliability inex, raial istribution network. I. INTRODUCTION N recent years, the environentally-frienly econoy, Irising fossil fuel prices an other incentives for inepenence fro fuel price causes high penetration of s. Investigations in [] iplies that play a pivotal role in the oern transportation evelopent because of its high efficiency in coparison with other vehicles' types such as plug-in hybri electric vehicle (PH), iesel hybri electric vehicle(h), gas H, fuel cell vehicles (FCV), conventional iesel an conventional gas, as epicte in Fig.. On the other han, s are zero eission an bring no treat for environent. But this fast an furious growth of s shoul be in a scheule anner to avoi power syste facing inispensible aages. Stuies on the optiu charging profile to intensify loa in off peak hours have been one in []. In aition, four ipressive factors on charging profile are: vehicle type, istance, charging level an charging start tie [3]. Reference [4], has presente a oel of PH aggregator's behavior in a virtual power arket. Furtherore, s' custoer behavior oeling is one in a atheatically approach in [5]. Benefits of PH's participation in ifferent types of electricity arkets such as regulation arket an spinning reserves are also iscusse [6]. Optiu allocation of s' parking in a istribution network is a challenging topic. Investigation on PH's parking lots allocation to satisfy gri technical constraints such as loss ecreent an voltage profile iproveent is one [7] but reliability constraints are not consiere. The scope of this paper is to allocate s' parking lots consiering istribution reliability. The rest of this paper is arrange as follows: In section II, probabilistic oeling of s' parking is eterine. In section III, proble forulation is aresse. Section IV, is evote to the optiization ethoology. Case stuy incluing test the propose etho on istribution network for RBTS bus is iscusse in section V an the last section, conclues the paper. II. PROBABILISTIC S'PARKING MODLLING A. Probabilistic Moeling of Single In this section, the charging ean of an electric vehicle is extracte base on probabilistic paraeters. The first paraeter which consiere for the charging tie of an is the probabilistic riven istance. Ref [8] propose a lognoral istribution function to generate the probabilistic aily riven istance, enote as M in (). The lognoral rano variables are generate using stanar noral rano variable N in () an are copute using ()[9]. The Box-Muller etho [0] is use to generate norally istribute rano variables N is forulate in (). N =.ln( U ) cos( π. U ) () where N is a stanar noral value (a noral RV with a ean of zero an a variance of one); furtherore, U an U are inepenent an ientically istribute pseuo rano nubers istribute unifor over the range (0,][9]. M e μ σ ( +. N ) = () where μ are the lognoral istribution paraeters an are calculate fro ean an stanar variation of M base on the historical ata, enote as μ, respectively[]. M. H. Aini an A. Isla are with Departent of lectrical an Coputer ngineering of Floria International University, Miai, FL (ails: hai.aini@ieee.org, arisla@fiu.eu) /4/$ I

2 To oel the an eparture tie, the Gaussian istributions is use, as the best estiate of rano resiential consuer behavior [9]-[3]. In (7), the forulas use to eterine these two factors are illustrate. t t = μ + σ = μ + σ. N eparture eparture eparture. N (7) Fig.. A coparison between efficiency of vehicles [] μ μ = ln μ σ + σ σ = ln + μ The aount of μ are extracte fro the historical ata of riving istance in [4], consiere 40 iles an 0 iles, respectively. The secon paraeter which is effective on the s perforance is the energy consuption per ile riven, shown in (4) []: (3) β = α. k (4) where α an β are constant coefficients epenent on the type of an k represents the fraction of the total energy input supplie by the battery []. In this paper, because of the concentration on the electric ean of, k is assue to be one. The axiu riving istance of when battery is fully charge is enote as M ax an calculate by the following equation: M ax BCAP = (5) where BCAP is the axiu battery capacity. Now the expecte energy ean of the, enote by ean is calculate as[]: ean BCAP ; M M = M. ; M < M ax ax The thir paraeter is the expecte uration of charging base on the charging rate, battery capacity ( BCAP ) an probabilistic an eparture tie. (6) where N an N are rano variables calculate fro(), μ are ean an stanar variation of tie base on historical ata, μ eparture eparture are ean an stanar variation of eparture tie, respectively. Typical values of these statistical paraeters are extracte fro [3] as shown in Table I. The entione paraeters, t an t eparture, shoul satisfy constraint t eparture > t, if this constraint i not satisfie, the ties generate again. After generating these ties the probabilistic uration of charge are calculate as below: turation = teparture t (8) Now, the esire state-of-charge (SOC) shoul be calculate by the entione paraeters. ean turation. chr SOCesire = Min SOC init +, SOCinit + BCAP BCAP B. Parking lot oeling This section introuces the input an output of parking lot. Noinal capacity of parking an charging rate are two input variables. Table II illustrates the three charging oes which are use in this paper [4]. The capacity of each parking is eterine consiering istribution network which parking shoul be locate in. To ake the oel ore practical, four classes of s are consiere. Table III shows these classes an their technical specifications. These classes are extracte fro [6] with soe changes in orer to aaption with the propose parking oel. Base on the entione paraeters the probabilistic oel of the parking lot is eterine by eans of probabilistic oel of single. This oel outputs are parking ean in a aily interval which is specifie in each hour an hourly plugge in s. In orer to consier the arket share of each class, the aily ean of each class is eterine by the propose oel, then these eans weighte suation leas to total loa oel of the parking lot. Total plugge in s are consiere in calculating reliability inices in the placeent process. (9)

3 3 TABL I STATISTICAL DRIVING PATTRN PARAMTRS [3] Paraeter μ σ μeparture σ eparture Value TABL II CHARGING RATS IN A PROPORTION OF BATTRY CAPACITY Charging Moe Charging Rate Class 3 4 Slow Charging Quick Charging Fast Charging TABL III CLASSS SPCIFICATIONS BCAP α ( kwh ) ( kwh ) ile BCAP/hour 0.3 BCAP/hour.0 BCAP/hour Market Share (Percent) IV. OPTIMIZATION USING GNTIC ALGORITHM Because of the iscrete essence of placeent an sizing proble, it has nubers of local inius. To optiize this type of objective function, heuristic algoriths ay be effective. In this paper genetic algorith (GA) is utilize to iniize the propose objective function containing three proinent reliability inices [7]. In this paper fitness function an objective function consiere to be the sae. The chroosoes inclue binary nubers which eterine the nuber of buses that parking shoul be installe at the. Because of the caniate LPs in the test syste are just resiential an coercial buses an these LPs oes not excee 5, each chroosoe for every parking bus nuber is five bits so the population size is five. Other GA paraeters, cross over fraction, utation probability an population size set 0.8, 0. an 0 respectively. The optial allocation ethoology is shown in Fig.. III. PROBLM FORMULATION In orer to optial placeent of parking lots, it is exigent to efine an objective function. The propose objective function can be shape base on various purposes. In this paper, the reliability aspect of parking lots' placeent in this function is consiere. The input of objective function is total neee capacity of parking lots in all of the stuie feeers an istribution network topology. The objective function of s' parking placeent in a raial feeer is to axiize istribution network reliability satisfying soe constraints. Two efficacious reliability inices are use in orer to evaluate the istribution network's reliability; syste average interruption frequency inex (SAIFI), syste average interruption uration inex (SAIDI) an average energy not supplie (ANS) are utilize in orer to have a coprehensive objective function [5]. The propose objective function illustrate in the following equation, inicates coposite reliability inex through weighte aggregation of these three inices [6]: SAIFI SAIDI ANS CRI = ( ωsaifi ) + ( ωsaidi ) + ( ωans ) SAIFI T SAIDIT ANST (0) where CRI is a coposite reliability inex an coefficient ω x > 0 inicates the weight for corresponing reliability inex an subscript T inicates the target value of each inex [6]. These coefficients have the sae aount 0.33, an the target value of SAIFI, SAIDI an ANS are consiere 5, 5 an 450, respectively. In the suggeste proble, ultifarious constraints are consiere such as the nuber an capacity of require parking lots, nuber of caniate loa points to satisfy civic architectural qualifications as well as being near resiential an coercial concentrate loa points to be afforable for investors. Fig.. Flowchart of parking lot allocation algorith

4 4 Fig.3. Distribution syste for RBTS bus [8] V. CAS STUDY In orer to evaluate the propose algorith, it is teste on istribution syste for RBTS bus fro [8]. This raial test syste consists of four feeers with four types of custoers, resiential, coercial, governent/institutions an sall user that coercial an resiential ones were consiere as caniates for s' parking placeent. The test syste is shown in Fig.3. As a result, first an secon feeers are connecte as aneuver points with norally open switches. In aition, thir an fourth feeers are also connecte in the sae way. It is exigent to eclare just resiential an coercial buses are consiere as caniate for placeent containing loa points nuber -3, 6-7, 0-, 5-9 an. VI. SIMULATION RSULTS Four scenarios were efine to evaluate the propose etho. A. Scenario In this scenario the fast charging etho was consiere. Five parking lots with total capacity of 500 s are allocate. B. Scenario In this scenario the slow charging etho was consiere. Total capacity is the sae as scenario. This scenario is efine to achieve a sensitivity analysis on charging rate effect on coposite reliability inex value. C. Scenario 3 In this scenario the fast charging etho was consiere. Total parking lots capacity is 000 s. This scenario is eterine tin orer to investigate the effect of neee parking lot capacity increent on coposite reliability inex value an allocation process. D. Scenario 4 In this scenario the slow charging etho was consiere. Total parking lots capacity is 000 s. The coposite reliability inex value is expecte to be increase after installing parking at loa points. Table IV shows the optiu locations of 's parking lots an table V illustrates the etaile siulation results. It can be inferre fro table V that increasing charging rate leas to CRI reuction. With coparing the result of first an secon scenario, it is observe that reucing charging rate leas to coposite reliability inex value increent. In aition, a coparison between thir an forth scenario also confirs this fact. In other wor fast charging etho causes reliability iproveent. Furtherore, coparing between the result of first an thir scenario shows that increasing in total s nuber results in reliability eterioration. So two ipressive factors on istribution network reliability which resulte by s' parking placeent are total nuber of s an charging etho. Fig.4. shows the CRI values for ifferent scenarios. VII. CONCLUSION This paper iscusse about reliability-constraint allocation of s' parking lots is a raial istribution network. Genetic algorith utilize to achieve the best allocation of parking lots. The propose algorith has potential to be carrie out so that reliability iproveent of istribution network is obtaine. Allocation of parking lots is investigate base upon a coprehensive coposite reliability inex. Genetic algorith is eploye to eterine optiu allocation base upon the aforeentione objective function. The nuerical results show that istribution feeers with no resiential an coercial custoers are not feasible to allocate parking lots. In conclusion, a stanar test syste is opte to show the effectiveness of the propose etho. As siulation results eonstrate, parking lots location is raatically sensitive to total nuber of s an charging etho. In aition, it is just the technical aspect an the econoical aspect of placeent shoul be stuy in future works. Thus, unicipalities shoul consier the reliability aspect, as a oinant factor of parking lot installation at istribution networks in their network expansion roaaps. The siulation results also shows that unscheule allocation of parking lots ay leas to an ineligible reliability of istribution syste. TABL IV PARKING LOTS' LOCATIONS Scenario nuber Coposite reliability inex value parking location (loa points nuber)., 0,, 5, 9.37,6-7, ,, 6, , 7, 0-, 8

5 5 TABL V LABORAT SIMULATION RSULTS Scenario nuber Charging Metho Total s CRI value ANS SAIFI SAIDI WITHOUT FAST SLOW FAST SLOW Fig.4. CRI Values for Different Scenarios VIII. RFRNCS [] S. G. Wirasingha,, R. Greban an A. ai, Source-to-Wheel (STW) Analysis of Plug-in Hybri lectric Vehicles, I Transaction on Sart Gri, vol. 3, No., 0. [] X. Yu, Ipacts assessent of PH Charge Profiles on Generation xpansion Using National nergy Moeling Syste, in Proc. I Power & nergy Society General Meeting - Conversion an Delivery of lectrical nergy in the st Century, 008, pp. 5. [3] Zahra Darabi, Mehi Ferowsi, "Aggregate Ipact of Plug-in Hybri lectric Vehicles on lectricity Dean Profile", I Trans. On sustainable energy, vol., no. 4, Oct. 0. [4] M. Shafie-khah, M. Parsa Moghaa, M.K. Sheikh-l-slai an Mehi Rahani-Anebili, Moeling of interactions between arket regulations an behavior of plug-in electric vehicle aggregators in a virtual power arket environent, nergy, vol. 40, no., pp , April 0. [5] M.H.Aini, B.Nabi, M.Parsa Moghaa an S.A.Mortazavi, " valuating the ffect of Dean Response Progras an Fuel Cost on PH Owners Behavior, A Matheatical Approach", Secon Iranian Conference on Sart Gri, ICSG0. [6] W. Kepton, A Test of Vehicle-to-Gri (VG) for nergy Storage an Frequency Regulation in the PJM Syste, Results fro an Inustry- University Research Partnership, 008. [7] M. Moraijoz an M. Parsa Moghaa, Optiu Allocation of Parking Lots in Distribution Systes for Loss Reuction, I PS GeneralMeeting, 0. [8] S. Meliopoulos, Power syste level ipacts of plug-in hybri vehicles, Power Systes ngineering Research Center (PSRC), 009 [9] A. D. Doínguez-García, G. T. Heyt, S. Suryanarayanan, Iplications of the Sart Gri Initiative on Distribution ngineering (Final Project Report-Part), PSRC Docuent -05, Sep. 0. [0] Reliability Test Syste Task Force of the Application of Probability Methos Subcoittee, I reliability test syste, I Transactions on Power Apparatus an Systes, vol. PAS-98, no. 6, pp , Noveber 979. [] P. Sharer, R. Leyier, an A. Rousseau, Ipact of rive cycle aggressiveness an spee on Hs fuel consuption sensitivity, Argonne National Lab, 007 [Online]. Available: entations.htl [] G. Li an X-P. Zhang, Moeling of Plug-in Hybri lectric Vehicle Charging Dean in Probabilistic Power Flow Calculations, I Transaction on Sart Gri, vol. 3, No., March 0. [3] S. Letenre, P. Denhol, P. Lilienthal, New Loa, or New Resource?, Public Utitlity Fortnightly, pp. 8-37, Dec [4] Patricia Saneier, Sion Felsenstein,; lectric vehicle infrastructure ABB Market potentials lectric vehicle infrastructure, using the Swiss exaple ; ABB Sales Switzerlan Business Developent, October 009; ABB Internal report. [5] R.. Brown, lectric PowerDistribution Reliability. NewYork:Marcel Dekker, 00. [6] L. Wang an C. Singh, " Reliability-constraine optiu placeent of reclosers an istribute generators in istribution networks using an ant colony syste algorith ", I Transaction on Systes, Man an CyberneticsPart C: Applications an Reviews, vol. 38, No. 6, Noveber 008. [7] D.. Golberg, Genetic algoriths in search, optiization an achine learning.reaing. MA: Aison-Wesley, 989. [8] R.N.Allan, R.Billinton, I.Sjarief, L.Goel an K.S.So, A reliability test syste for eucational purposes- basic istribution syste ata an results, I Transactions on Power Systes, vo.6, No., May 99.