Effectiveness of Battery-Supercapacitor Combination in Electric Vehicles

Effectiveness of Battery-Supercapacitor Combination in Electric Vehicles S. Pay, Member, IEEE, and Y. Baghzouz, Senior Member, IEEE Abstract This paper investigates the application of a supercapacitor bank when used as a power buffer to smooth rapid power fluctuations in and out of the battery of an electric or hybrid vehicle. The study considers the simple case where the supercapacitor bank is connected directly in shunt with the battery as well as the case where it is connected through a DC/DC converter that is necessary to achieve optimal performance. The work is illustrated through computer simulations during vehicle acceleration and deceleration. Partial experimental data on a prototype circuit is also shown. Proper design of such a hybrid electrical energy storage system is expected to result in substantial benefits to the well being of the battery bank. Index Terms electric and hybrid vehicles, supercapacitors, deep-cycle batteries, regenerative breaking, power transients, DC/DC converters. I. INTRODUCTION significant portion of energy is dissipated in the brakes A when driving conventional gasoline-powered vehicles in urban areas, where periodic acceleration-deceleration cycles are required. Therefore, recovering this energy through regenerative breaking is an effective approach for improving vehicle driving range and this can only be accomplished by electric vehicles (EV) or hybrid-electric vehicles (HEV). Regenerative breaking in these vehicles captures some of the kinetic energy stored in the vehicle s moving mass by operating the vehicle s traction motor as a generator that provides braking torque to the wheels and recharge the batteries [1]. The battery bank of an EV is sized for peak power demand, and this often compromises the desired weight and space specifications. On the other hand, The auxiliary power unit (APU) of an HEV is designed to provide the normal average power required by the vehicle, while the battery is sized to provide power surges needed during acceleration and hill climbing and to accept momentary powers during breaking. While EVs and HEVs are more efficient than conventional vehicles in urban areas, the electric load profile consists of This work was supported in part by the U.S. Department of Energy, Nevada Operations Office, through the Nevada Environmental Research Park (NERP) Program. S. Pay is with Harris Engineering Consultants, Las Vegas, NV 89119, USA, (e-mail: spay@lvcm.com). Y. Baghzouz is with the Department of Electrical and Computer Engineering, University of Nevada, Las Vegas, NV 89154 USA (e-mail: eebag@ee.unlv.edu). high peaks and steep valleys due to repetitive acceleration and deceleration. The resulting current surges in and out of the battery tend to generate extensive heat inside the battery, which leads to increased battery internal resistance thus lower efficiency and ultimately premature failure [2], [3]. The problem of battery overheating and loss of capacity is more acute when batteries are near full state-of-charge (SOC) since they cannot accept large busts of current from regenerative breaking without degradation at this stage. Supercapacitors (also referred to as ultracapacitors or electrochemical capacitors) have much greater advantage over batteries when capturing and supplying short bursts of power due to their higher power density limits, and ability to charge and discharge very quickly. Hence adding a supercapacitor bank will assist the battery during vehicle acceleration and hill climbing, and with its quick recharge capability, it will assist the battery in capturing the regenerative braking energy. This significant advantage a battery-supercapacitor energy storage/supply system gained attention in recent years in transportation systems [4]-[8] as well as other applications [9], [10]. Applying supercapacitors also allows for a smaller battery size, and there is almost no limit to number of their charge-discharge cycles (since there are no chemical reactions involved in their energy storage mechanism). Furthermore these devices require no maintenance and do not use toxic materials. Special considerations must be taken into account when integrating such a hybrid energy storage system to achieve optimal performance. While direct connection of the supercapacitor across the battery terminals does reduce transient currents in an out of the battery, the best way to utilize the supercapacitor bank is to be able control its energy content through a power converter. The paper reviews the direct supercapacitor-battery shunt connection, after a short section addressing component modeling issues. The desired connection is then addressed by using a DC/DC converter in the boost mode when discharging, and in the buck mode when charging the supercapacitor bank. The analysis is illustrated by computer simulations of an actual system. Part of the simulations is verified by experimental data on a prototype model. II. COMPONENT MODELING This section reviews the modeling of the main power system components in an electric vehicle; namely, the battery

bank, the supercapacitor bank, and the electrical load. More details on electrical component modeling can be found in power electronics textbooks such as Ref. [11]. A. Battery Bank Batteries are quite difficult to model as they undergo thermally-dependent electrochemical processes while delivering and accepting energy. Thus the electrical behavior of a battery is a nonlinear function of a number of constantly changing parameters, such as internal temperature, state-ofcharge, rate of charge/discharge, etc The capacity of a battery depends on the discharge rate as wells as temperature. This relationship is described by Peuket s equation relating the discharge current I (A) to the time t (hr) it takes it to discharge, I = α (β/t), where α and β are constants. Given the battery capacity C To at temperature T o, the capacity at some other temperature is computed by C T = C To {1 + σ(t-t o )} where σ is a constant. An approximate model that is often used for batteries is a Thevenin equivalent circuit that consists of the open circuit voltage in series with an effective internal resistance. Both voltage and resistance values are functions of the battery SOC, and these relations are generally supplied the manufacturer. SOC is defined the percentage of energy left in a battery (after supplying a certain amount of amp-hours) relative to its full capacity. The open-circuit voltage is often approximated by a linear function of the SOC: V oc = a 1 + a 2 SOC, at some specific temperature (e.g., 80 o F). The battery internal resistance has static and dynamic values that depend of battery SOC, whether the battery is being charged or discharged and rate of charge/discharge. In short duration studies, however, the amount of amp-hours in and out of the battery is a small fraction of the battery capacity. Hence it is fair to assume that battery internal voltage is constant during such periods, and a quasi-steady state model with fixed open-circuit voltage and internal resistance constitutes an acceptable battery model [12],[13]. Note that two resistance values are used in this case, one during charging and another during discharging. B. Supercapacitor Bank As in conventional capacitors, the resistance and inductance of the terminal wires and electrodes of supercapacitors are represented by a series R-L circuit. Further, non-perfect insulation between the device electrodes results in leakage current that is represented by a large shunt resistance. The difference between conventional and supercapacitors is that the latter are much more efficient, i.e., the series resistance is a lot lower and the shunt resistance is much higher in value. The self-discharge time constant of supercapacitors several orders of magnitude larger than that of conventional capacitors. More sophisticated models suitable for dynamic studies are found in Ref. [14]. The study under investigation is a short-duration analysis of the power (or current) distribution between the battery bank and supercapacitor bank during acceleration and deceleration. Hence, the leakage resistance can be ignored without much error, and the supercapacitor bank can simply be represented by a series R-C circuit. C. Electrical Load The electrical load in electric vehicles consists mainly of an inverter-fed induction motor for motive power. During regenerative breaking, the motor is turned into a generator by reducing the frequency of its terminal voltage, thus reversing power flow and producing braking torque. Detailed modeling of inverter-fed motor drives is found in standard power electronics and drives textbooks. As far as the power source in concerned, power demand is sufficient for analysis. Since the DC bus voltage is not allowed to vary significantly from its nominal value, current demand gives a good approximation of power demand. Thus the load can be modeled simply by a time-varying current source that reverses direction as the vehicle switches from coasting or acceleration to regenerative braking. III. DIRECT SUPERCAPACITOR CONNECTION Supercapacitor integration with the battery-load circuit can be challenging when trying to optimize the presence of this additional sub-system. The simplest way is to connect the supercapacitor directly in parallel with the battery bank, after first pre-charging it to the battery terminal voltage. Such a connection is shown in Fig. 1 where the load current denoted by i L is defined to flow downward (i.e., positive during acceleration and coasting, and negative during regenerative breaking). Fig. 1. Parallel connection of supercapacitor bank, battery bank, and electrical load. Given a certain load current profile representing some short drive cycle, the battery current i b and supercapacitor current i c are found by basic circuit rules such Kirchoff s voltage and current laws: i + i = i c b L ( 1) ( 2) v= vc icrc = vb ibrb dvc ic = C ( 3) dt where v c and v b represent the internal capacitor and battery voltages, respectively. Subsititution of (1) and (3) in (2) yields the first order equation of v c :

where dv dt c + αv = αv + βi c b L ( + ) ( + ) ( 4) 1 Rb α =, β = 5 C R R C R R b c b c The solution to (5) can be written as shown in Eqn. (6) below: αt αt αt vc = Ke + vb + βe ile dt ( ) ( 6) where K is determined by setting the initial value of v c to v b. Note that it is not possible to control power flow in and out of the supercapacitor bank in the circuit above since its terminal voltage is forced to be equal to the that at the battery terminals at all time. Current division between the battery and supercapacitor bank is determined solely by the two branch internal resistances and internal voltages. The basic controls for the static power converter in Fig. 2 can be summarized as follows: prior to vehicle use, the supercapacitor must charged by the battery bank or by an off board power supply. During the initial stages of vehicle acceleration, power flow out of the supercapacitor should be matched to that of the load demand as long as the device current rating is not exceeded. This requires the controller to adjust the ON state pulse with of S 1 accordingly. As the capacitor continues to discharge, the battery current should gradually increase and ultimately reach the load current when the energy stored in the capacitor reaches low levels. During regenerative breaking, the supercapcitor should be charged at the maximum possible rate (by modulating switch S 2 ) so that a small fraction of the load current flows into the battery bank. The current injected by the load is then diverted slowly into the battery as the capacitor approaches full charge. IV. SUPERCAPACITOR CONNECTION THROUGH POWER CONTROLLER The above straight connection clearly indicates that optimal use of the supercapacitor bank requires a power flow controller between the two energy storage subsystems. The objective is to maintain the battery current as constant as possible with slow transition from low to high current during transients to limit battery stress. On the other hand, the supercapacitor ought to charge as fast as possible without exceeding maximum current from regenerative breaking, and to discharge most of its stored energy during acceleration. Energy flow in and out of the supercapacitor can be controlled with a pulse-with-modulated (PWM) DC/DC converter with a simple topology as shown Fig. 2 [7], [8]. The supercapacitor is discharged during acceleration at a rate controlled by modulating switch S 1. In this boost mode, energy is delivered to inductor L f when S 1 is turned ON (State 1), then transferred to the load through diode D 2 when S 1 is OFF (State 2). During deceleration, the supercapacitor is charged at a rate controlled by modeling switch S 2. In here, energy is transferred to L f when S 2 is turned ON (State 3), then to the supercapacitor through diode D 1 when S 2 is turned OFF (State 4). The analytical expression of the supercapacitor current can be determined by the second order differential equation d i dt Rc dic 1 + + ic f ( t) (7) L dt CL 2 c = 2 f f where f(t) = 0 in States 1 and 4, and Rb dil f ( t) = (8) L dt f in States 2 and 3. The power controller governs the flow of energy in and out of the supercapacitor only during fluctuations in power demand. Consequently, quasi-steadystate relations between the converter input and output parameters do not exist in this case, and one has to resort to circuit simulation software such as PSpice. Fig. 2. Supercapacitor integration through power flow controller. IV. NUMERICAL ILLUSTRATION To illustrate the analysis in the sections above, the performance of supercapacitor bank that was recently designed and installed as part of the energy storage system of a series hybrid-electric bus is investigated by computer simulations. The APU of this hybrid vehicle is a hydrogen-powered internal combustion engine [15]. But for simplicity, the analysis simulates the vehicle power circuit with the APU shut off, thus representing the vehicle engine status when the battery bank is at or near full charge. The battery system consists of two banks connected in parallel, and one of the banks is shown in Fig. 3 along with the supercapacitor bank. A short description of the battery and supercapacitor subsystems follows. Each of the two battery banks consists of 28 deep-cycle valve-regulated-lead-acid (VRLA) battery units connected in series. Each unit is rated at 12 V, and a capacity of C To = 85 Ah @ C/3 (at T o = 80 o F). Other battery parameters are listed below. static internal resistance during charging: 4 mω for SOC 80%, and 10 mω at SOC = 90% recharge current limit: 400 A, Peukert s equation constants: α = 1.33 and β = 256, capacity-temp. dependence parameter: σ = 0.004,

V oc vs. SOC parameters: a 1 = 11.80, a 2 = 1.32, battery subsystem rating: 336 VDC, 170 Ah@C/3. The supercapacitor consists of a string of 150 cells. Each cell is rated at 2.5 V and 2,500 F. Additional data follows: cell series resistance = 1 mω, cell leakage resistance: 300Ω, cell peak voltage: 2.7 V, cell rated current: 400 A, supecapacitor subsystem rating - rated voltage: 375 V, peak voltage: 405 V, capacitance: 16.67 Farads, energy storage capability: 1.2 MJ. Since no capacitor cells are alike, they need to be balanced as the voltage distribution becomes a function of internal parallel resistance and cell capacitance. To distribute the total stack voltage evenly across the capacitor bank, bypass resistors sized to dominate the leakage current are placed in parallel with each cell. The battery pack, however, is not equipped with a charge equalizer. The battery condition are monitored with a data acquisition system that logs terminal voltage and internal temperature of each cell and alerts the operator in case of excessive heating or out-of-range voltages. acceleration and deceleration. The following test procedure was conducted: Pre-charging the supercapacitor string and connecting it in parallel with the battery unit. Inject a constant current of 200 A until the voltage reaches the maximum allowed value of 14.2 V, then reduce the current to maintain this voltage for 10 seconds. Shut of the load for 30 seconds. Connect a 200 A load for 10 seconds, then stop. The measured load current, battery current, capacitor current and system voltage are shown in Fig. 4. Note that the presence of the supercapacitor led to significant reduction of battery current, especially during the initial seconds of imposing load current. Fig. 4 shows the corresponding current and voltage profiles calculated by Eqn. (1)-(7) where the load current is approximated by i L = 320(e -1.25 t - e -0.16 t ) A for 0 < t 10 sec., and i L = -205 A for 40 < t 50 sec. The graphs of both figures are nearly perfect duplicates. Fig. 3. Supercapacitor bank, battery bank. To validate the performance of the supercapacitor-battery system during charge and discharge, a prototype model consisting of one 12 V battery cell in parallel with a string of 6 supercapacitor cells of the same model and make described above, was tested using special equipment that can both generate and absorb specific current profiles that simulate Fig. 4. Battery and supercapacitor currents during charging and discharging, experimental data, calculated data. Testing with a power controller to govern the energy flow from the supercapacitor was not conducted due to lack of converter availability, and only computer simulations using

PSpice are performed at present. To show the performance of the hybrid system with a DC/DC controller, a vehicle acceleration cycle is simulated by a load current having a waveshape that increases from 0 to 300 A within 10 seconds, then decreases exponentially for 1.5 seconds, then stays constant at 66 A for 9.5 seconds. The converter parameters are set as follows: L f = 100 mh, R f = 10 mω, f sw = 1 khz, C f = 100 mf, the switch duty ratio is set to increase in discrete steps every 1.5 seconds as long as the load current is rising. Figure 5 below shows the simulated supercacitor and battery currents and terminal voltages. Note that the power controller allowed the supercapacitor to discharge from 360 V down to 135 V (i.e., its energy content dropped to nearly 14% of the initial charge). This made the supercapacitor an effective energy buffer as delivered the most significant portion of the load current during the first 10 seconds of acceleration, and the battery bank current made a relatively smooth transition for vehicle stand-still to constant speed. For comparison purposes, Fig. 6 shows the corresponding currents and voltages when the battery and supercapacitor banks are connected without a power controller. Some of the additional remarks that can be made include the following when comparing both figures: a) The battery peak current is reduced by 40%. b) The DC bus voltage regulation is improved by 30%. c) The supercapacitor s SOC is 3.5 times lower with than without the power controller after acceleration. Fig. 6. Simulated battery and supercapacitor currents and voltages during acceleration (w/o power converter). V. CONCLUSIONS Adding a supercapacitor bank to a battery- or fuel celldriven vehicle makes sense and advantages by far outweigh the disadvantages. A direct parallel connection will reduce battery stress by assisting with transient currents during acceleration and deceleration, but will not make full use of the supercapacitor as a true power buffer. Optimal use of the supercapacitor requires a power controller that requires only two static power switches, two power diodes, an inductor and filter capacitor, but the best control strategy is not fully developed due challenging control issues. Future work will report on the development of the power converter that is currently in the design stages and on experimental data from actual vehicle driving cycles. VI. ACKNOWLEDGEMENT The work reported in this paper was funded by a grant from the U.S. Department of Energy, Nevada Operations Office, through the Nevada Environmental Research Park (NERP) program. Fig. 5. Simulated battery and supercapacitor currents and voltages during acceleration (with power converter). VII. REFERENCES [1] Y. Gao, L. Chen and M. Ehsani, Investigation of the effectiveness of regenerative braking of EV and HEV Proc. Society of Automotive Engineers, 1999, paper No. 1999-01-2910.

[2] P.T. Moseley, High-rate, valve-regulated lead-acid batteries suitable for hybrid electric vehicles? Journal of Power Sources, Vol. 84, 1999, pp. 237-242. [3] Electrosource, Battery Handbook, Horizon C2M Batteries, 1999. [4] E. Faggioli, P. Rena, V. Danel, X. Andrieu, R. mallant, and H. Kahlen, Supercapacitors for the energy management of electric vehicles, Journal of Power Sources, Vol. 84, 1999, pp. 261-269. [5] J.C. Brown, D.J. Eichenberg, W.K. Thompson, L.A. Viterna, and R.F. Soltis, Ultracapacitors store energy in hybrid electric vehicles, NASA Tech Briefs, April, 2000, pp. 63-64. [6] Smith, T.A.; Mars, J.P.; Turner, G.A., Using supercapacitors to improve battery performance, Proc. IEEE 33 rd annual Power Electronics Specialists Conference, pp. 124-128. [7] B.J. Arnet and L.P. Haines, Hig-power DC-to-DC converter for supercapacitors, Proc. IEEE Power Conversion Conference, Osaka, Japan, 2002, pp. 1160-1165. [8] L. Bertoni, H. Gualous, D. Bouquain, D. Hissel, C. Pera, J.M. Kauffmann, Hybrid auxiliary power unit (APU) for automotive applications, Proc. IEEE 56 th Vehicular Technology Conference, 2002, pp. 1840-1845. [9] C. Chen, "High pulse system through engineering battery-capacitor combination," Proc. 2000 American Aeronautics & Astronautics, Inc., Paper # AAA-2000-2935, pp. 752-755. [10] Barrade, P.; Rufer, A, Supercapacitors as energy buffers: a solution for elevators and for electric busses supply, Proc. Power Conversion Conference, 2002, pp. 1160-1165. [11] P.T. Krein, Elements of Power Electronics, Oxford University Press, 1998. [12] M.A. Merkle, Variable Bus Voltage Modeling of Series Hybrid Electric Vehicles, M.S. Thesis, Virginia Tech University, 1997. [13] Steven Pay, Hybrid Electric Vehicle Regenerative Braking Using Ultracapacitors, M.S. Thesis, University of Nevada, Las Vegas, Dec., 2000. [14] S. Buller, E. Karden, D. Kok and R.W. De Donker, Modeling the dynamic behavior of supercapacitors using impedance spectroscopy, IEEE Trans. Ind. Applications, Vol. 38, No. 6, 2002, pp. 1622-162. [15] Y. Baghzouz, Y.J. Fiene, J. Van Dam, L,. Shi, E. Wilkenson, and R.F. Boehm, Modifications to a hydrogen/electric hybrid bus, Proc. American Aeronautics and Astronautics Engineers, 2000, paper No. AAA-2000-2857. VIII. BIOGRAPHIES Steven Pay received his B.S. and M.S. degrees in electrical engineering from the University of Nevada, Las Vegas in 1994, and 2000, respectively. He is currently an electrical engineer with Harris Consulting Engineers, Inc. His interests include power systems analysis, distribution system design and power quality. His is a registered Professional Engineer in the State of Nevada. Yahia Baghzouz received is B.S., M.S. and Ph.D. degrees in electrical engineering from Louisiana State University, Baton Rouge, LA, in 1981, 1982 and 1986, respectively. He is currently professor of Electrical Engineering, and Associate Director of the Center for Energy Research at the University of Nevada, Las Vegas. His interests are in power quality, power electronics and renewable energy. He is a registered Professional Engineer in the State of Nevada.