POSER 2016, PRAGUE MAY 24 1 Deign of Control for Battery Storage Unit Converter Martin GALÁD 1 1 Dept. of Mechatronic and Electronic, Univerity of Žilina, Univezitná 1, 010 26 Žilina, Slovakia martin.galad@fel.uniza.k Abtract. Battery torage will play an important part of ytem upplied by photovoltaic or other renewable energy grid-connected ource. It i life-important to ue appropriate battery management technique to achieve deired efficiency, long-life of battery unit and favorable cot. o enure thee demand, a precie control of batterie hould conit of accurate State-of-Charge (SOC) etimation, effective balancing and charging control. he correct value of SOC i etimated uing combination of Coulomb Counting and Kalman Filter method. Performance of control for - converter with Double tiered capacity balancing circuit i analyzed. Battery management i verified uing MALAB/Simulink. Keyword Battery management, charge, dicharge, balancer, tate of charge, VRLA battery. 1. Introduction Battery torage i a crucial part of many application. For example, the increaing importance of energy accumulation in grid-connected photovoltaic power plant implie the meaning of precie battery management. Battery management hould handle not only charging and dicharging batterie but alo the afety of battery tring, utilization the mot of poible torage capacity of batterie, the prediction of age of batterie to etimate overall effectivene of the ytem [1]. he management hould conit of charge control, SOC and State-of-Health (SOH) etimation, balancing control, temperature and afety control. In thi paper, control of tatic battery torage connected to the grid will be demontrated. A battery torage are ued 4 VRLA batterie. o etimate the SOC, Current Counting and Kalman Filter approache are combined [2]. An important part of battery torage unit i balancer circuit. Performance of battery i getting wore by aging, ignificant influence have the temperature, not fully charged battery or long time inactivity with low SOC. Balancer enable effective uage of battery unit capacity. It ha everal function uch are cell voltage leveling, afety or extenion of battery life [3]. Final control algorithm for - converter i verified by MALAB/Simulink. 2. VRLA Battery Storage Unit Battery ytem conit of bidirectional / converter which provide the charging and dicharging of VRLA. Simplified battery management diagram i hown in Fig. 1. Battery management of the torage unit conit of SOC etimator, which i the mot important part. Etimated SOC i ued required by charge and balancer control. bu - converter Battery tring with balancer Charging and balancer control Fig.1 Battery management diagram U, I meaurement SOC etimator 2.1 - Converter and parameter of VRLA Buck-boot topology wa choen for the - converter. Input capacitance C in wa et to 0.68 mf, output capacitance C out to 42μF and inductance L bb to 3mH. Batterie are charged uing contant voltage with current limit et to 1.5A a i defined in dataheet [4]. In ab.1 are hown parameter of ued VRLA battery. he battery torage unit ha nominal voltage 48 V. Nominal Voltage Nominal Capacity Internal reitance Recommended charging Voltage (20 C) VRLA battery parameter 20h rate 10h rate 5h rate 1h rate 12V 22mΩ Floating ue Cycle ue Maximum charging 1,5A ab. 1 VRLA parameter 7,2Ah 6,5Ah 5,9Ah 4,3Ah 13,5-13,8V 14,4-15V
2 M. GALÁD, DESIGN OF CONROL FOR BAERY SORAGE UNI CONVERER 2.2 Battery Sytem Balancer Decription here are many balancer topologie and can be divided into two main type. Paive balancer ue the diipation to exce the tored charge and equalize the voltage among each battery cell or battery. Intead of energy diipation, active balancer tranfer the energy from the mot charged cell to other uing toring element uch are capacitor, inductor or tranformer. In Fig. 2 are hown compared topologie and in ab. 2 are compared parameter. Compared balancer are paive reitive (PR), witched capacitor (SC), ingle witched capacitor (SC), double tiered capacitor (DC), witched inductor (L), witched tranformer (r) and converter balancer (Con) [5]. o balance the batterie, DC balancer circuit i choen and implemented in the ytem becaue of it favorable power loe, implicity, average but atifactory time of balancing. DC ha alo maller balancing time than SC according [5]. Only drawback i the cot. he overall circuitry of battery charger i hown in Fig.3. Balancer control i imple, witche are witched oppoitely. U bu C in I char L bb C out B 1 B 2 B 3 C 1 C 2 C 4 C 3 B 4 ESR Fig.3 Charger with balancer PS SC SC C1 C4 C2 C3 DC r Con C L Fig.2 Comparion of balancer topologie opology PR SC SC DC L r Con Balance time + ++ ++ ++ +++ ++ +++ Complexity +++ +++ + +++ ++ ++ ++ Efficiency - +++ +++ +++ ++ ++ +++ Power m-h m-h m-h m-h m m-h Size +++ ++ ++ ++ + + + Cot +++ ++ ++ ++ + + + -2bad + atifactory ++ good +++ excellent - mall m - medium h - high ab. 2 Comparion of balancer circuit parameter L 2.3 SOC Decription here are many method to etimate SOC. he mot ued i Current Coupling. It i baed on the fact that capacity have pecific amount of charge [6]. Counting the charge during dicharge can lead to the value of SOC. It i defined by: 1 tt di SOCt t SOCt C t dt (1) where SOC t and SOC t+ t are value of SOC in time t and t+ t, C the capacity of VRLA, η efficiency of dicharging or charging and I dicharge/charge current. hi method ha one major drawback. It i the error due the ample intenity and dependence of accurate ene of current. hi mean that SOC can be etimated for cycle performance without correction or modification becaue the error grow with operational time. Alo, dynamical charge/dicharge profile increae the error [6]. o minimize thi error, the SOC obtained by Current Coupling i updated by SOC etimated by Kalman Filter after 10. Kalman Filter i baed on the battery model prediction and it correction to achieve the mot poible olution.. he electrical equivalent model are commonly ued becaue of le complexity and neceity of pecial knowledge of battery chemitry although the electrochemical model are more accurate. Accuracy of SOC etimated by Kalman Filter depend ignificantly on the choen battery model, proce noie and meaurement noie. Model i decribed uing tate equation: x Ax Bu w k k1 k k y Cx v k k k where x k/k-1 i the tate vector, u k control vector, y k meaured vector, w k the meaurement noie, v k the proce noie, A the tate matrix, B the control matrix and C the obervation matrix. (2)
SOC [%] POSER 2016, PRAGUE MAY 24 3 Randle equivalent model of VRLA battery i choen for Kalman filtering for it atifying accuracy and relatively implicity unlike other model. It i hown in Fig. 4. It conit of bulk capacitance decribing the capacity of VRLA C b, R in repreent the inner erie reitance of VRLA, R t C t circuit define the dynamical tranient behavior [7]. R p C b U oc R in R t C t U t I U bat 0 0 CR b p A 0 0 CR t t CbRp Ct R t Cb B C Ct Cb C t 1 0 0 Fig. 4 Randle equivalent model of VRLA Equation which decribe Randle battery model are: U U U R I U bat oc t in oc RpI U RC t p t b RtI Ut Ut RC oc where U t i the voltage which repreent tranient behavior of battery, U bat voltage on terminal of VRLA, I current, R p parallel reitance which repreent the elf-dicharge. Kalman Filter equation for linear ytem are: x Ax Bu kp k 1 k Pkp APk 1A Q S CPkpC R K 1 PkpC S x x K y Cx P k kp k kp k KC P kp where x kp i the prediction tate vector, P kp covariance prediction, x k the new tate vector, P k the new covariance, R the meaurement noie, Q the proce noie, K the kalman gain, I identity matrix and S the innovation of covariance [8]. Matrixe for Randle model are defined: (3) (4) 3. Simulation Reult SOC and Balancer imulation were done. SOC wa etimated during everal dicharge with current 7A. Simulation tep wa et to 1. Fig. 5 how dicharge profile of VRLA. In Fig. 6, etimated SOC from Current Coupling, Kalman Filter and their combination i hown. Detailed view on SOC etimation reult i on Fig. 7. Accuracy of combined SOC method i uitable, the only drawback i the delay caued by Kalman Filter. 50 40 30 20 10 U bat [V] I bat 0 t[] 95 90 85 Fig. 5 Dicharge profile of VRLA CCupdKF CC KF 80 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 t [] Fig. 6 SOC etimation reult
SOC [%] V bat I bat SOC [%] 4 M. GALÁD, DESIGN OF CONROL FOR BAERY SORAGE UNI CONVERER 87.15 87.1 87.05 87 CCupdKF CC KF Fig. 10 SOC of batterie Bat 1 Bat 2 Bat 3 Bat 4 Initial SOC 45 43 39 37 Final SOC 69,21 68.42 66,95 66,39 ab. 3 Comparion of battery voltage 86.95 1908 1910 1912 1914 1916 1918 1920 1922 1924 1926 1928 t [] Fig. 7 Detailed SOC etimation reult Balancer imulation wa alo analyzed by imulation. Capacitie of balancer were et to 470mF and protection reitance which limit the balance current were et to 30mΩ. he witching frequency wa et 1 khz. Reult of imulation are hown in Fig.8, 9, 10. Batterie SOC were et to (37, 39, 43, 45 %) at the tart of imulation. Balancer i activated during contant current charging (1,5 A) at time 10 and imulation time wa et to 5000. Initial and final value of SOC are compared in ab. 3. Simulated performance i ufficient for energy torage with mall charge/dicharge dynamic. 2 1.5 4. Concluion Control of - converter for battery torage unit wa analyzed in the paper. hi control conited of accurate SOC etimation, charging control and balancing of battery tring. Combination of Current Counting and Kalman Filter method of SOC wa ued. he value of SOC etimated by Current Counting wa updated by the value obtained by Kalman Filter every 10. It ha advantage of both approache; it i relatively imple and till accurate. A balancing circuit, the DC wa choen. It ha good parameter and can be ued during charge and dicharge. In the future, temperature of battery will be added to the SOC etimation and alo parameter of equivalent battery model will be etimated to enhance the accuracy of SOC prediction. hi control will be part of tatic gridconnected ytem with energy torage. 1 0.5 0 t [] 14 13.5 13 12.5 12 11.5 Fig. 8 Charging profile Bat1 Bat2 Bat3 Bat4 11 t [] 70 65 60 55 50 45 40 Fig. 9 Voltage among batterie Bat1 Bat2 Bat3 Bat4 35 t [] Acknowledgement Reearch decribed in the paper wa upervied by Prof. P. Špánik, EF KME in Žilina and upported by the Slovak grant agency VEGA for project no. VEGA 1/0579/14 - Reearch of methodology for optimization of lifetime of critical component in perpective electronic appliance through the ue of ytem level imulation. Reference [1] LINDEN, D., REDDY,.B. Handbook of Batterie, hird Edition, McGraw-Hill, 2001 [2] CHANG, W., he State of Charge Etimating Method for Battery: A Review, ISRN Applied Mathematic, vol. 2013, Article ID 953792, 7 page, 2013. doi:10.1155/2013/953792 [3] KREIN, P.., BALOG, R. S. Life Extenion hrough Charge Equalization of Lead-Acid Batterie, elecommunication Energy Conference, 2002. INELEC. 24th Annual International, 2002 [4] FIAMM 20721/2 Dataheet [5] DAOWD, M., OMAR, N.BOSSCHE, P., MIERLO, J. A Review of Paive and Active Battery Balancing baed on MALAB/Simulink, International Review of Electrical Engineering (I.R.E.E.), Praie Worthy, 2011 [6] LINDNER, D., NIEDEMAYR, F. Akku4Future, Report for Workpackage 6, Fraunhofer Italia 2014 [7] JONGERDEN, M.,HAVENKOR, B., Battery modeling, echnical report, R-CI-08-01, CI, 2008
POSER 2016, PRAGUE MAY 24 5 [8] GREWAL, M. S., ANDREWS., A. P. Kalman Filtering: heory and Practice Uing MALAB, Second Edition, John Wiley, 2001 About Author... Martin GALÁD wa born in Lučenec, Slovakia in 1988. He earned hi bachelor degree in Electrical Engineering in Univerity of Žilina in Žilina, Slovakia in 2010. He got hi Mater degree in Electrical Drive in Univerity of Žilina in 2012. Nowaday, he i a PhD. tudent at the Department of Mechatronic and Electronic, Faculty of Electrical Engineering, Univerity of Žilina and hi field of tudy are tand-alone power ytem, battery management.