Smart Railway Station Energy Management Considering Regenerative Braking and ESS

Transcription

1 Smar Railway Saion Energy Managemen Considering Regeneraive Braking and ESS Ibrahim Sengor, Hasan Can Kılıçkıran, Huseyin Akdemir, Bedri Kekezoglu, and Ozan Erdinç Yildiz Technical Universiy TURKEY and João P. S. Caalão INESC TEC and FEUP, Poro, C-MAST/UBI, Covilha, and INESC-ID/IST-UL, Lisbon, Porugal Absrac Opimum operaion of he energy consumpion of end-users gains more imporance o reduce oal elecriciy bills and in order o more efficienly use energy resources hanks o smar grid concep. Elecrical railway saions are one of he bes places o ake ino accoun for in his manner. This sudy proposes a railway saion energy managemen (RSEM) model. As he main conribuion o he lieraure, regeneraive braking energy (RBE) recovered during he operaion of a mero line is assumed o mee he saion load demand in daily operaion. The RSEM model composed of RBE usage, energy sorage sysem (ESS), and grid suppor is formulaed as a mixed-ineger linear programming (MILP) framework. RSEM model is esed by inroducing he impac of passengers change on RBE for such cases ha wheher RBE and ESS are considered or negleced. Index Terms Energy sorage, mixed ineger linear programming, railway energy managemen, regeneraive breaking energy. NOMENCLATURE The main nomenclaure used in his paper is expressed below. Oher symbols and abbreviaions are defined where hey firs appear. Abbreviaions ERS ESS MILP RSEM RB RBE SOE Indices Parameers CE ESS CR ESS DE ESS DR ESS P load Elecrical Railway Sysem. Energy Sorage Sysem. Mixed Ineger Linear Programming. Railway Saion Energy Managemen. Regeneraive Braking. Regeneraive Braking Energy. Sae-of-energy. Period of he day index in ime unis [min]. Charging efficiency of he ESS. Charging rae of he ESS [kw per min]. Discharging efficiency of he ESS. Discharging rae of he ESS [kw per min]. Railway saion power demand [kw]. P RBE SOE ESS,ini SOE ESS,min SOE ESS,max T λ buy Variables P ESS,ch P ESS,disch P grid P ESS,used P RBE,used SOE ESS u ESS Power generaed from braking energy of rain [kw]. Iniial SOE of he ESS [kwh]. Minimum SOE limi of he ESS [kwh]. Maximum SOE limi of he ESS [kwh]. Number of ime inervals in one hour. Price of energy bough from he grid [e/kwh]. ESS charging power [kw]. ESS discharging power [kw]. Power supplied from he grid [kw]. Power used from ESS [kw]. Power used from RBE [kw]. Sae-of-energy of he ESS [kwh]. Binary variable: if 1 during charging period, else 0. I. INTRODUCTION A. Moivaion and Background Due o he several reasons such as rising fossil fuel prices, necessiy of reducing CO 2 emissions, depleion of fossil fuel reserves and, mos significanly, increasing elecriciy demand, he efficien use of energy plays a pivoal role o cope wih he aforemenioned problems. Recen developmens in he field of efficien use of energy have led o a renewed ineres in elecrical railway sysems (ERSs) in erms of ransporaion [1]. Besides, progressive expansion of ERSs nework have undeniable pressure on he elecrical grid due o he huge amoun of energy consumpion and energy recovery derived from regeneraive braking (RB) [2]. RB means regeneraed energy during braking mode by racion moors and re-using his energy in he form of elecriciy. Moreover, his energy may supply a grea energy poenial for energy efficiency via giving back o he caenary line, energy sorage sysem (ESS), and reversible subsaion [3]. To solve menioned issues in he power sysem of ERSs, smar grid concep should also be implemened o ERSs in all aspecs. Smar grid is a novel concep o deal wih complexiy /17/$31.00 c 2017 IEEE

2 of elecrical grid and efficien usage of elecriciy from he producion o end-user poins considering all generaion and sorage opions, as well as providing communicaion beween sides in a power sysem. This paradigm provides achieving convenien energy managemen soluions including differen ype of producers and consumers [4], [5]. There are many sudies o evolve elecrical infrasrucure of ERSs for smar grid framework; where numerous R&D projecs exis considering his opic from differen poins of view [6]. B. Lieraure Overview There are several sudies dealing wih increase energy efficiency based on soring of RBE in ERSs. Ciceralli e al. [7] suggesed an energy managemen conrol sraegy for wayside ESS o obain maximum RBE during braking mode. The proposed model considered he acual volage and curren value of ESS ogeher wih power sysem losses, and based on forecasing of rain moion parameers such as ineria forces and acceleraion. Neverheless, RBE was used for acceleraion of oher rain a he saion and passengers impacs were no evaluaed while rain is operaed in braking and mooring mode. Khayyam e al. [8] developed a railway energy managemen sysem archiecure based on he smar grid concep. In his archiecure, dynamic opimal energy usage was aimed whils aking ino accoun rain loads, on-board and wayside ESS as well as disribued generaion unis. However, he saion loads and passenger effec on RBE were no considered. Novak e al. [9] considered a hierarchical energy managemen sraegy ha has an energy sorage sysem and a microgrid, for single power fed railway saion. The economic benefis of aforemenioned managemen sysem was invesigaed in erms of differen scenarios wheher microgrid was included or no. Also, he conrol problem of energy consumpion level saed in [10] and [11] was solved. Noneheless, line opology such as curves and slopes was no considered in [9]. Lu e al. [12] suggesed a power managemen sraegy o enhance he energy saving of a diesel muliple-uni rain based on dynamic programming and nonlinear programming. I was saed ha he proposed sraegy reduced he cos of fuel consumpion by 7%. Alhough line opology and passenger numbers were considered o obain energy consumpion and generaion by he rain, RBE was no reused for he saion loads. Nasr e al. [13] performed a smar DC microgrid sysem o preven wase of RBE based on energy soring. This sudy was aimed ha RBE was reused for he saion loads such as elevaors, escalaors, lighing ec., no railway applicaions so as o increase benefi from energy recovery. However, he impac of varying price and passenger number were no evaluaed in [13]. Pankovis e al. [14] analyzed he feasibiliy of implemening smar grid o he railway applicaion aking ino accoun wind and PV based generaion unis, RBE, and ESS by using fuzzy logic and geneic algorihm. However, line opology and passenger effecs were no considered as well as uilizaion of energy for a saion loads. There are also several oher sudies no referred here considering he opic from differen poins of view. C. Conribuions This paper propounds a Mixed Ineger Linear Programming (MILP) model of Railway Saion Energy Managemen (RSEM) concep including ESS, RBE, and differen price schemes such as dynamic price, ime-of-use, fla price, for he purpose of an overall operaional assessmen of a railway saion. Keeping in mind he valuable conribuions made by prior sudies, his paper inends o make conribuions saed below: RBE is used wih he purpose of feeding inernal demand of a railway saion. The number of passengers ha change depending on he inensiy during he day, is aken ino accoun while compuing RBE. The effecs caused by differen price schemes on RSEM are invesigaed by carrying ou case sudies. D. Organizaion The res of he paper is organized as follows: The operaion of railway vehicle and mahemaical formulaion of problem are expressed in Secion II. Thereafer, he case sudies and relaed resuls are presened and discussed in Secion III. Finally, concluding remarks and recommendaions for fuure sudies are summarized in Secion IV. II. METHODOLOGY The block diagram of RSEM sraegy is demonsraed in Fig. 1. The RSEM sysem rules he operaion of a smar railway saion in a subway line aking ino accoun RBE, price signal received from he uiliy, and ESS. In RSEM sysem, only he inernal demand of saion is considered and energy consumpion of rain is assumed o be fed via racion ransformers. The res of his secion gives informaion abou mahemaical model of rain moion and he proposed energy managemen model. A. Mahemaical Model of Train Moion In order o deermine he poenial of RBE, he mahemaical model of rain moion is used. This subsecion presens he model of rain moion. The rain moion is based on Newon s one dimensional moion laws and direcly affeced by no only he line opology bu also he characerisics of racion devices: n F i = m a (1) i=1 In (1), F i represens he oal forces ha affecs on rain moion, m is he roaing rain mass, and a is he acceleraion of he rain. The forces affecing he rain moion are divided ino wo main caegories as F r and F ag, and are illusraed in Fig. 2. F r : Force generaed by racion moors. I is considered as posiive in racion mode, while negaive in braking mode.

3 Fig. 1. The block diagram of RSEM Fig. 2. The forces acing on rain moion F ag : Toal forces ha play negaive role agains rain moion. I consiss of line gradien, line curve, and resisance caused by own rain moion. F r F ag = m a (2) Equaion (1), can be rearranged by subsiuion of oal forces ha ac on rain moion wih F r and F ag. F ag = F r + F gr + F c (3) Toal forces ha have a negaive effec on rain moion is obained by sum of F r, F gr, and F c which symbolize resisance caused by own rain moion, gradien of line, and curve of line, respecively, in (3). Herein, F r is modelled by using well-known Davis formula [15]. P = (m.a + F ag). v η g. η m. η i. 3, 6 + P a (4) I = P V l (5) In (4), P indicaes he insananeous power of rain and i is assumed as posiive while rain acceleraes. Conversely, when rain brakes, P is considered as negaive due o he generaed power in racion moors on rain. Also, P a represens auxiliary loads of rain while η g, η m, and η i are efficiency of gear, racion moors, and inverers respecively. Lasly, v symbolizes rain speed a ime. I should also be noed ha I represens he insananeous curren and V l indicaes he line volage in (5). B. Energy Managemen Model The main aim of energy managemen model is o minimize he oal daily cos of railway saion elecriciy consumpion. Therefore, he objecive funcion comprises only power bough from grid (P grid signals (λ buy ) as a variable a ime. Beside, he price ) are ime dependen. min P grid T.λbuy (6) I is worh o noe ha his sudy focuses on operaional cos of he railway saion. Hence, invesmen coss of communicaion sysem and any oher power sysem componens are no considered. In addiion, here are more han one opion o decide ime granulariy T such as 1h, 30 min, 15 min, ec. 1) Power Balance: P grid + P ESS,used = P load + P ESS,ch, (7) Equaion (7) enforces ha power demand of railway saion loads (P load ) and ESS (P ESS,ch ) can be me from eiher grid (P grid ) or ESS (P ESS,used ), or combinaion of hem. Addiionally, deails of ESS model are presened in he nex subsecion [5]. 2) ESS Modelling: P RBE,used P ESS,used + P ESS,ch P ESS,disch SOE ESS = P ESS,disch. DE ESS, (8) CR ESS. u ESS, (9) DR ESS. (1 u ESS ), (10) = SOE 1 ESS + CE ESS (. P ESS,ch ) + P RBE,used T P ESS,disch T, 1 (11) SOE ESS = SOE ESS,ini, if = 1 (12) SOE ESS SOE ESS,max, (13) SOE ESS SOE ESS,min, (14) Equaion (8) saes ha used ESS power (P ESS,used ) so as o mee railway saion loads is a funcion of discharged efficiency and discharged power capaciy of ESS (P ESS,disch. DE ESS ). Addiionally, considering ESS canno be charged and discharged a he same ime, (9) and (10) regulae ESS charging and discharging operaion by using a logical consrain. ESS charging rae (CR ESS ) resrics power provided from RBE (P RBE,used ) and charging power of ESS (P ESS,ch ) in (9). Similarly, maximum power aken from ESS (P ESS,disch ) is consrained by discharged rae of ESS (DR ESS ) in (10). For any ime inerval, he SOE of ESS (SOE ESS ) comprises charging energy aken from RB and elecrical grid, discharging energy used for railway saion loads, and SOE of ESS in he previous ime inerval (SOE 1 ESS ). In he beginning of operaion, SOE of ESS is equalled o iniial SOE of ESS (SOE ESS,ini ) hanks o (12). Las bu no leas, consrains (13)

4 Fig. 3. The roue of Isanbul M1 ligh mero line and (14) ensure ha keeping SOE of ESS beween maximum (SOE ESS,max ) and minimum (SOE ESS,min ) value. 3) RBE Modelling: Herein, P RBE symbolizes RB power ha is available for charging of ESS. Regarding he maximum allowed charging capaciy of ESS some amoun of P RBE migh be curailed. Keeping in mind all possible cases, RBE model is buil by defining a variable which is called as P RBE,used. Equaion (15) draws an upper limi for RBE obained from rain braking. P RBE P RBE,used, (15) III. TEST AND RESULTS The calculaion of poenial RBE regarding he mero line is carried ou using RAILSIM 8 sofware [16]. Aferwards, a MILP model consruced so as o decrease daily elecriciy cos of a railway saion using RBE is assessed in GAMS v sofware wih CPLEX v.12 solver [17]. In his sudy, M1A ligh mero line as one of he busies mero lines in Isanbul is simulaed. Fig. 3 picures he overall look of he M1 mero line. In his sudy, only 19.7 kilomeres of M1A line which has 18 saions, is simulaed in RAILSIM. To make he simulaion as realisic as possible, real daa including racion moor sizes, opological feaures of line such as gradien and curve, aken from Mero Isanbul Co. is considered. I is regarded in his paper ha he number of passengers dynamically changes per hour. Therefore, i is allowed ha dynamic variaions in passengers affec RBE. Daily passenger profile obained from he records of Mero Isanbul Co. is given Fig. 4. The proposed RSEM is implemened jus for one railway saion, namely Bahcelievler. The energy demand of his saion was recorded and he compued power values depending on acual daa is illusraed in Fig. 5. Herein, railway saion loads are composed of escalaor, elevaors, lighning, heaing, venilaion, and air condiioning. Because of he fac ha duraion of RB process is very shor, i is assumed in his sudy as 1 minue; herefore, he sampling ime of recorded daa is rearranged according o his. In his sudy, i is considered ha railway saion loads are fed by grid and ESS. Also, ESS is assumed o be charged via RBE or grid. RB power profile of he relaed saion is given in Fig. 6. I is worhy o noe ha one-way power flow is assumed o be from grid o he saion loads. In oher words, here is no energy-selling back o grid. The proposed ESS model is supposed o have 100 kwh Fig. 4. The daily passenger profile of M1 ligh mero line TABLE I THE PRICE SIGNAL OF TIME-OF-USE Time Inerval Price Signal [e/kwh] 23:00-07: :00-18: :00-23: Fig. 5. The daily load demand profile of Bahcelievler railway saion capaciy. Besides, charging and discharging rae of ESS is considered as 100 kw per minue. I is regarded ha ESS has 50 kwh iniial SOE and charging/discharging efficiencies are Addiionally, deep discharging limi is aken as 20 kwh. Las bu no leas, i is considered ha oal elecriciy cos of RSEM operaion is independen from invesmen cos of ESS. Assuming ha he saion has a smar meer infrasrucure, i is acceped ha RSEM can dynamically adjus he elecriciy consumpion demand relaed o he price signal coming from he uiliy wih is own inernal insrumens. In his sudy, hree pricing schemes are considered as dynamic price signal, ime of use, and single price. Time varying price signal used in his sudy is shown in Fig. 7, ha is a normalized version of he scheme given in [18]. Moreover, single price is couned as e/kwh and deails of ime of use signal are given in Table I. Fig. 8 picures he power flows and he change in SOE of ESS in he Bahelievler railway saion beween he hours 5:00 and 6:30 for case sudy wih he dynamic prices. Since he RSEM model is analyzed in minues, i is given only for a specific ime inerval of Fig. 8, for he sake of he clariy. The reason for choosing his range is ha all he load flow possibiliies for ESS ha can be observed are wihin his range. Figure 8(a) shows he insananeous power flow of ESS locaed

5 Fig. 9. The variaions in SOE of ESS during a day for he case sudies relaed o dynamic price. Fig. 6. The daily RB power profile for Bahcelievler Saion Fig. 7. Dynamic Price Signal Fig. 8. The variaion of: (a) Power decomposiion of ESS, (b) Power bough from grid, (c) SOE of ESS in Bahcelievler railway saion. Insananeous drawn power from grid by ESS is shown in red, he power ransferred from RBE o ESS is shown in green, and blue graph illusraes he insananeous power ha ESS gives so as o feed inernal load demand of saion. Figure 8(b) shows he insananeous power ha he saion draws from he grid o mee is inernal load demand. In addiion, Fig. 8(c) depics he variaions in SOE of ESS. As can be seen from Fig. 8, ESS sores energy beween 05:00 and 05:20 hours because of he low energy cos of he respecive ime inerval due o dynamic pricing. Because of RBE of rain ha comes o saion a 06:05, RSEM model prefers o use sored energy in ESS insead of buying from grid jus before ha ime. The reducion in power drawn from grid for he relaed ime inerval can be seen from Fig. 8(b). Figure 9 picures he corresponding resuls of ESS SOE where i is obvious ha he charging and discharging saes of ESS are direcly affeced by RBE usage. I is noeworhy ha dashed line indicaes he case includes only ESS while he oher one, sraigh line, shows he case ha RBE and ESS insrumens ake place a he same ime. In he menioned figure, alhough he boh cases are examined under dynamic price signal, he case wihou RBE leads o less variaion in SOE of ESS. However, i can be also seen from he same figure ha when RBE is inroduced, SOE of ESS alers more frequenly relaed o prior case. Table II encapsulaes he case sudies assessed in his paper. I can be saed ha uilizaion of wheher only ESS or ESS ogeher wih RBE posiively affecs he oal elecriciy cos of railway saion. The wors case scenario is assumed as railway saion has no ESS while is smar meer of railway saion is operaed under single price signal, and RBE is wased hrough resisances placed over he rains. Alhough mos of he acual railway saions are no managed by using dynamic price signal, evaluaions marked ha he bigges cos advanages in cases wih dynamic price signal. In RSEM applicaion, one opion for decreasing he oal elecriciy cos of railway saion is o equip he saion wih ESS, which provides nearly 2.75% drop in cos. Ye anoher and more effecive one is o reuse he RBE, which can provide a significan reducion in cos as 15.74% compared o he previous case. IV. CONCLUSION Depleion of fossil fuel reserves ogeher wih increasing public knowledge and sensiiviy agains climae change made energy efficiency a more imporan issue o deal wih decreasing

6 Case Sudy Wihou ESS Wihou RBE Wih RBE + ESS TABLE II COMPARISON OF DIFFERENT CASE STUDIES Dynamic Price TOU Price Single Price Toal Cos [e] Cos Reducion [%] Toal Cos [e] Cos Reducion [%] Toal Cos [e] Cos Reducion [%] 316,132 Base Case Base Case Base Case 307, , carbon emission levels due o he energy usage. ERSs are one of he bes candidae for using energy in a more effecive way wih heir promising poenial of RBE. This sudy aimed o model a RSEM srucure consising of RBE, ESS, and smar meer capable of being operaed by various pricing schemes, so as o regain he RBE occurring during deceleraing periods of rains. During minimizing he elecriciy cos of a railway saion, he passenger effec and differen pricing schemes were aken ino accoun, which can be lised as he main conribuions of his sudy. In addiion, hree differen pricing schemes were evaluaed by consrucing a MILP model of RSEM. Moreover, RBE occurring in Bahcelievler railway saion was compued wih RAILSIM sofware. I is noeworhy ha his paper assumed power can only flow from grid o railway saion. Therefore, any selling price or direcional power flow beween grid and saion were no considered in his paper. In his paper, several cases creaed according o he variaions of ESS and RBE were examined under differen pricing schemes. The case ha railway saion neiher includes ESS or has he abiliy o uilize RBE, was seleced as he base case for each kind of price signal. From he elecriciy consumpion poin of view, a more economical operaion of railway saion was obained when he saion has access o ESS. Furhermore, he economic operaion could be improved wih he inroducing he RBE compared o he prior cases. The resuls also showed ha he mos significan reducion as 15.74% can be achieved by using ESS and RBE a he same ime under dynamic price signal. I is worh o underline ha all analysis made in his sudy are based on only deerminisic approach. Considering his, he auhors would like o sae ha an comprehensive evaluaion of a RSEM model including demand response sraegies and disribued generaion unis wih a sochasic approach is planned as a fuure sudy. ACKNOWLEDGMENT The auhors would like o hank METRO ISTANBUL Co. for heir valuable cooperaion as sharing he real daa belong M1A and allowing o use RAILSIM sofware. This work was suppored by FEDER funds hrough COMPETE 2020 and by Poruguese funds hrough FCT, under Projecs SAICT-PAC/0004/ POCI FEDER , POCI FEDER , UID/EEA/50014/2013, UID/CEC/50021/2013, UID/EMS/00151/2013, and SFRH/BPD/103744/2014. Also, he research leading o hese resuls has received funding from he EU Sevenh Framework Programme FP7/ under gran agreemen no REFERENCES [1] A. J. López-López, R. R. Pecharromán, A. Fernández-Cardador, and A. P. Cucala, Assessmen of energy-saving echniques in direc-currenelecrified mass ransi sysems, Transporaion Research Par C: Emerging Technologies, vol. 38, pp , [2] R. R. Pecharroman, A. Lopez-Lopez, A. P. Cucala, and A. Fernandez- Cardador, Riding he rails o dc power efficiency: Energy efficiency in dc-elecrified meropolian railways. IEEE Elecrificaion Magazine, vol. 2, no. 3, pp , [3] A. González-Gil, R. Palacin, P. Bay, and J. Powell, A sysems approach o reduce urban rail energy consumpion, Energy Conversion and Managemen, vol. 80, pp , [4] C. W. Gellings, The smar grid: enabling energy efficiency and demand response. The Fairmon Press, Inc., [5] O. Erdinc, N. G. Paerakis, T. D. Mendes, A. G. Bakirzis, and J. P. Caalão, Smar household operaion considering bi-direcional ev and ess uilizaion by real-ime pricing-based dr, IEEE Transacions on Smar Grid, vol. 6, no. 3, pp , [6] E. P. de la Fuene, S. K. Mazumder, and I. G. Franco, Railway elecrical smar grids: An inroducion o nex-generaion railway power sysems and heir operaion. IEEE Elecrificaion Magazine, vol. 2, no. 3, pp , [7] F. Ciccarelli, A. Del Pizzo, and D. Iannuzzi, Improvemen of energy efficiency in ligh railway vehicles based on power managemen conrol of wayside lihium-ion capacior sorage, IEEE Transacions on Power Elecronics, vol. 29, no. 1, pp , [8] S. Khayyam, F. Ponci, J. Goikoexea, V. Recagno, V. Bagliano, and A. Moni, Railway energy managemen sysem: Cenralized decenralized auomaion archiecure, IEEE Transacions on Smar Grid, vol. 7, no. 2, pp , [9] H. Novak, M. Vašak, and V. Lešić, Hierarchical energy managemen of muli-rain railway ranspor sysem wih energy sorages, in Inelligen Rail Transporaion (ICIRT), 2016 IEEE Inernaional Conference on. IEEE, 2016, pp [10] M. Vašak, M. Baoić, N. Perić, and M. Bago, Opimal rail roue energy managemen under consrains and fixed arrival ime, in Conrol Conference (ECC), 2009 European. IEEE, 2009, pp [11] M. Vašak, M. Baoić, M. Bago, and N. Perić, Consrained energy opimal conrol of rains: Concepual implemenaion, in 30h Conference on Transporaion Sysems wih Inernaional Paricipaion Auomaion in Transporaion, 2010, pp [12] S. Lu, S. Hillmansen, and C. Robers, A power-managemen sraegy for muliple-uni railroad vehicles, IEEE ransacions on vehicular echnology, vol. 60, no. 2, pp , [13] S. Nasr, M. Iordache, and M. Pei, Smar micro-grid inegraion in dc railway sysems, in Innovaive Smar Grid Technologies Conference Europe (ISGT-Europe), 2014 IEEE PES. IEEE, 2014, pp [14] P. Pankovis, J. Pouge, B. Robyns, F. Delhaye, and S. Brisse, Towards railway-smargrid: Energy managemen opimizaion for hybrid railway power subsaions, in Innovaive Smar Grid Technologies Conference Europe (ISGT-Europe), 2014 IEEE PES. IEEE, 2014, pp [15] D. Seimbille, Design of power supply sysem in dc elecrified ransi railways-influence of he high volage nework, [16] SYSTRA Consuling, Inc., RAILSIM 8 User Manual,, 2008, online; accessed 11 March [17] GAMS, CPLEX 12 Solver Descripion, hps://gams.com/laes/docs/solvers/cplex/index.hml, 2012, online; accessed 8 March [18] K. M. Tsui and S.-C. Chan, Demand response opimizaion for smar home scheduling under real-ime pricing, IEEE Transacions on Smar Grid, vol. 3, no. 4, pp , 2012.