Implementation and Evaluation of an Ultracapacitor-Based Auxiliary Energy System for Electric Vehicles

Transcription

1 Implementation and Evaluation of an Ultracapacitor-Based Auxiliary Energy System for Electric Vehicles Micah Ortúzar, Jorge Moreno and Juan Dixon (SM IEEE) Department of Electrical Engineering Pontificia Universidad Católica de Chile Casilla 306, Correo 22, Santiago, Chile Fax: , ABSTRACT. Searching for a better efficiency, an auxiliary energy system for electric vehicles was designed, implemented and tested. The system, composed of an ultracapacitor bank and a buckboost converter, was installed in an electric vehicle, which is powered by a lead-acid battery pack and a 54 kw brushless DC motor. Two control strategies where developed: one based on heuristics and the other based on an optimization model using neural networks. These strategies were translated to algorithms, implemented in a DSP and their performance was evaluated in urban driving. The results were incorporated to an economic evaluation of the system, showing that the reduction in costs would only justify the inclusion of this type of system to a lead-acid battery powered vehicle if the battery life is extended in 50% or more, which is unlikely. The same results where extrapolated to a case in which the lead-acid batteries are replaced by a fuel cell. In this case, costs of different power support systems where evaluated, such as ultracapacitors and high specific power lithium based batteries. The results showed a significant cost reduction when auxiliary energy systems configurations are included, in contrast to a system powered by fuel-cells only. Also, the cost reduction was higher when using ultracapacitors for this purpose.

2 I. INTRODUCTION Recent commercial debut of hybrid vehicles [1, 2] is a proof of their improved efficiency and performance. The word of hybrid stands for the use of two or more energy storage devices that combine three main characteristics: specific energy (Wh/kg) for driving range, specific power (W/kg) for acceleration, and power reversal capability for regenerative braking [3, 4]. Most hybrid configurations use two energy storage devices: one with high energy storage capability, called Main Energy System (MES), and the other with high power capability and reversibility, called Auxiliary Energy System (AES). The MES provides extended driving range and the AES good acceleration and regenerative braking. Even though a vehicle could eventually run with the MES only, the power and efficiency characteristics will be poor. The MES-AES combination gives to hybrids extended range, good power, high regenerative braking capability and better efficiency. These two energy storage devices can be connected in different ways, and cars that use them are classified in two main categories: Parallel Hybrids and Serial Hybrids. Parallel hybrids are vehicles powered from two independent mechanical outputs working in parallel. These power units may move the rear and front wheels independently, or may be combined in a special gearbox to move the same wheels. The last case is shown in Fig. 1(a) and most commercial models, like the Toyota Prius [5], use this configuration. Serial hybrids are vehicles with only one mechanical output (usually an electric motor), as shown in Fig. 1(b), which receives power through static converters from two electrical sources. In this case, any two sources that deliver electric power could be used, as long as they are compatible in terms of electrical variables. This topology is the only possible when energy sources are batteries or fuel cells, because they do not produce mechanical power.

3 Main Energy System (MES) (a) G E A R B O X Traction Motor Internal Combustion Engine Auxiliary Energy System (AES) + Power Inverter _ (Gasoline or Diesel Tank) (High Power batteries, Flywheels, Ultracapacitors) Main Energy System (MES) Traction Motor Power Inverter + _ (High Energy Batteries, Fuel Cells, Microturbines, Diesel Engines, Otto Engines, etc. Auxiliary Energy System (AES) (b) DC-DC Converter (High Power batteries, Flywheels, Ultracapacitors) Fig. 1. Power circuit of a typical (a) parallel hybrid and (b) serial hybrid vehicle. The requirements for an adequate AES are high efficiency, high power-density and reversibility [6]. The most mentioned candidates for this task are high specific-power batteries (Liion, Ni-M-H), ultracapacitors [7] and flywheels [8, 9], of which the only commercially available today are the first two. For the MES, the most popular and proven device has been the internal combustion engine. Nevertheless, there are other devices under study, such as gas turbines, fuel cells [10], and high specific energy batteries, which are suitable for serial hybrid configurations. Recent developments in the areas of fuel cells and high specific energy batteries (PEM cells, Zincair cells, ZEBRA batteries, etc.) suggests a good chance of having a competitive alternative to internal combustion engines for MES in the near future [11]. This means that a serial hybrid configuration, not commercially available yet in passenger cars, might be an interesting option to explore.

4 These facts motivated the development of a serial hybrid configuration, with ultracapacitors as AES [12], and with lead-acid batteries as MES. The prototype was implemented in an electric vehicle transformed from a conventional Chevrolet LUV truck (similar in weight and shape to a Chevrolet S-10) with a Brushless DC traction motor (32 KW nominal power and 53 KW maximum power) [13]. The MES is formed by a battery pack of 26 lead-acid batteries connected in series (356 Vdc), which are currently being changed by a single Sodium Nickel-Chloride (ZEBRA) battery (371 Vdc). The AES was implemented with a 20 Farad ultracapacitor bank and a Buck-Boost converter, with a nominal voltage of 300 Vdc, and a nominal current of 200 Adc. This paper presents the complete power converter design and implementation, including DSP (Digital Signal Processor) control. Final tests were performed with the vehicle running on lead-acid batteries only, and powered by a hybrid configuration of lead-acid batteries (MES) and ultracapacitors (AES). For the hybrid configuration, the performance was analyzed using two different control strategies. In addition, a chapter with deducted economic implications complements the technical evaluation for those cases, and also for a hypothetical Fuel Cell case, which is used as MES. II. STATIC CONVERTER TOPOLOGY The Auxiliary Energy System (AES) needs a static DC-DC converter, as showed in Fig. 1(b). The DC-DC converter must be able to transfer energy from the battery to the ultracapacitors and vice-versa. It also transfers energy from and to the ultracapacitors during acceleration and regenerative braking respectively. Figure 2 shows a simplified schematic of the DC-DC converter and the two related sources: the 20-Farad ultracapacitor (VU) and the battery pack (VB). R2 include the internal resistances of ultracapacitors (ESR) and L S, and Rint the battery internal resistance. The positive and negative power terminals connect this system to the electric drive system.

5 VB Battery Rint i BAT V C + C T2 20 Farad ultracapacitor bank + R2 + T1 Positive power terminal D2 D1 L S V S VU Negative power terminal Fig. 2. DC-DC Converter schematic. The main parts of the converter are the inductor Ls, the capacitor C, the IGBTs T1 and T2, and the Diodes D1 and D2. A 12 khz fixed-frequency PWM is applied on either IGBT to transfer energy back and forth. The converter has two operation modes: Buck and Boost. Buck operation consists of transferring energy from the Battery pack (or the power terminals) to the ultracapacitors by triggering IGBT T2. Boost operation results from triggering IGBT T1, and energy is transferred from the Ultracapacitor to the Battery pack or power terminals. In either case, the amount of current transferred, will depend on the Ultracapacitor voltage, the system parameters (resistances, battery voltage) and the duty cycle of the PWM applied. Under steady state, and for mean current values during periods of several milliseconds, the converter can be modelled as an ideal DC-DC transformer. During Boost operation, steady state voltages and currents are described by equations (1) and (2), respectively [14]. For simplicity, these equations do not take into account the diodes and IGBTs voltage drop effect. VS V = (1) 1 C ( δ )

6 Ib VU ( 1 δ ) R int + R2 0 VB ( 1 δ ) 2 VU VU ( 1 δ ) ( 1 δ ) VB VB 0 < 0 (2) where, VU : ultracapacitor voltage VB : battery voltage Rint : battery s internal resistance R2 : L S resistance plus ultracapacitor ESR δ : duty cycle of the PWM applied (0 < δ < 1). As it can be seen from (1) and (2), Boost operation can be modelled as a one-way-conducting DC transformer, where the transformer ratio, as seen from the Ultracapacitor side, is 1/(1 - δ ). Equations (3) and (4) describe steady state voltages and currents during Buck operation (energy transferred from Battery to Ultracapacitor). V s = δ Vc (3) Ib ( VB δ VU ) 2 ( R2 + Rint δ ) 0 ( VB δ VU ) ( VB δ VU ) 0 < 0 (4) As in the previous case, Buck operation may be modelled as a one-way-conducting DC transformer, where the transformer ratio seen from the battery side is δ.

7 Equations (2) and (4) do not take into account the ripple component of currents. Equations (1) thru (4) will help understand the converter behaviour under different conditions and will help elaborate an adequate control strategy. III. STATIC CONVERTER DESIGN AND IMPLEMENTATION A careful design is required for the implementation of the static converter. Some of the most important tasks where: the design of the inductor Ls, and the water-cooled heatsink. Other key components were the flat conductors within the static converter, their semiconductors, DC capacitor and snubbers. 3.1 Design of Inductance L S. The inductance L S allows the transient energy storage during operation of the DC-DC converter. Its design is also related with the current ripple amplitude, which is one of the variables that have to be minimized, because it produces undesirable EMI, mechanical vibrations and losses due to current induction on surrounding conducting material. Figure 3 shows a typical steady-state current waveform for a Buck operation. As it is well known, the commutation frequency f and the inductance Ls are the two design configurable values that, together with the fixed battery voltage, will determine the maximum ripple amplitude [14]. The frequency was set to a value of 12 khz (considered low enough to keep commutation losses low). Therefore, in order to keep a maximum ripple-amplitude value of 5 A (2.5 % nominal current), the Ls inductance value had to be of at least 1.4 mh (considering a nominal battery voltage of 350 V). On the other hand, its weight, volume and series resistance had to be as small as possible.

8 ( ) ( + ) di di b b dt dt i b t 0 T-t 0 T Fig. 3. Current thorough ultracapacitors ripple waveform. In order to minimize the skin effect at high frequencies, a laminated conductor (0.5 mm thick and 12 cm wide) was selected, and because of its better conductivity to weight ratio (better than copper) aluminium was the chosen material. The achieved inductance was of 1.6 mh, with resulting total resistance of 37 mω. The inductance is capable of transferring currents up to 200 A for several seconds without considerable heating. The resulting inductance weighted 22 kg, which is adequate considering the currents it must withstand. Figure 4 shows the design and constructed prototype of this element. Fig. 4. inductance Ls: schematic and photo of the constructed prototype.

9 Fig. 5. Schematic and photo of the heatsink prototype. 3.2 Design of Heatsink for Power Transistors. Another important design element was the heatsink, needed to evacuate energy from conduction and commutation losses of the semiconductors. The thermal resistance had to be low enough to maintain the semiconductor junction temperature below the maximum allowed of 150º [15] while delivering maximum power. Several air-cooled and water-cooled heatsinks where considered, but in most cases the thermal resistance was higher than the maximum allowed of 0.05 ºC/W (according to the thermal model) [15]. Besides, the big size did not adjust to the design constraints. Hence, a water-cooled heatsink was designed and constructed from aluminium material. It consisted of two complementary parts with several parallel channels of equal length and width to ensure equal distribution of water flow. The designed heatsink, shown in Fig. 5, has a theoretical thermal resistance calculated at 0.01ºC/W [16], which is much lower than the required 0.05ºC/W. 3.3 Other components. The semiconductors used where IGBTs from POWEREX s Intellimod line, with incorporated gating circuits. The electrical connections between the capacitor C into the converter and the semiconductors had to have the least parasitic inductance to avoid over voltage conditions

10 during commutations. Therefore, they where constructed from laminated copper, which reduces parasitic inductance of conductors [15]. R-C snubbers where also incorporated to absorb excess energy from parasitic inductances during commutations. A picture of the constructed Buck-Boost converter is shown in Fig. 6. Here, the main components, such as capacitor, snubbers, semiconductor and heatsink, as well as the laminated connections are displayed. Fig. 6. Picture of the constructed static converter (without the cover) showing main components.

11 3.4 Integration to Vehicle For integration of the AES, several hardware and software protections were incorporated to the design. Fuses where installed for protection against failure. A diode was also incorporated in parallel to the fuse that connects the converter and batteries. This was done because a fuse breakdown during boost operation could result in a dangerous voltage rise in the converter capacitor, possibly producing capacitor blast and/or energy discharge through arc. Anyhow, the mentioned fuse is monitored by the control processor to disable operation when failure occurs. Figure 7 shows the power circuit schematic with the AES installed. Brushles- DC Traction Motor Power Inverter _ + Diode 125 A Fuse _ + DC-DC Buck-Boost Converter Fuse 160 A Inductor Ls _ 26 Lead- Acid Batteries A Fuse Ultracapacitor Bank + _ AES Fig. 7. Resulting power circuit in electric vehicle. The 132 ultracapacitors where packed in five groups, connected in series and installed in the vehicle. Dissipative equalizing devices where also installed on each ultracapacitor unit. Figure 8 shows pictures of the vehicle, the installed converter and ultracapacitors.

12 Fig. 8: Installation of the ultracapacitor system. (a) the EV, (b) the Buck-Boost converter, (c) inductance L S and ultracapacitors in the back, (d) one ultracapacitor box. IV. CONTROL SYSTEM AND ENERGY MANAGEMENT ALGORITHMS The control of AES, such as data monitoring, current-control and energy-management algorithm, is performed by a DSP (Digital Signal Processor TMS320f241) and its associated circuitry. This unit acquires all relevant signals, processes data, and generates the PWM signals to commutate IGBTs in the DC-DC converter. In addition, a communication protocol with a personal computer was implemented in this processor to perform real-time monitoring of relevant data. Two energy management algorithms where implemented to verify their influence in system behaviour and overall efficiency: a Heuristic-based Algorithm and an Optimized Algorithm.

13 4.1 Heuristic Algorithm The first algorithm, shown in Fig. 9, is based on Heuristics and follows two main rules [17]: first, a rule of energy balance that ensures long-term convergence of capacitor state of charge; and second, a rule that consists on limiting the battery current to predefined maximum positive and negative values, limiting its power to values slightly above mean power consumption. Fig. 9. Heuristic control algorithm schematic. The first rule states that energy content in the ultracapacitors must have an inverse relation to vehicle speed. Thus, when the vehicle runs at low speeds, energy is reserved to accelerate. On the other hand, when the vehicle runs at high speeds, space to store energy from braking is made available. This is achieved by injecting or extracting current from the ultracapacitors to reach a predefined state-of-charge reference, which depends on the vehicle speed.

14 The second rule consists on limiting the current extracted from (or injected to) the battery pack. Currents outside those limits will come from the AES through the ultracapacititors. These current limits automatically change when the battery is fully charged, avoiding overvoltages during regenerative braking. The two aforementioned rules are complementary because when the vehicle accelerates, current must be drawn from capacitors to maintain the energy balance, which usually coincides with large amount of current being consumed by the traction drive. 4.2 Optimized Algorithm The second algorithm, shown in Fig. 10, was developed with optimization tools. It obtains from a Neural Network (NN) the learned value of the most efficient current from the AES at all times [18]. Fig. 10. Optimal control applied on Neural Networks algorithm schematic.

15 The NN was trained using various sets of data, in each of which the most efficient AES current for a given load current was calculated. The most efficient AES currents are determined using Optimal Control techniques, where the path or load current is known. However, different AES currents will result in different battery currents and therefore the most efficient set of currents must be found, maintaining border conditions such as ultracapacitor state of charge. A model of the battery and the AES (ultracapacitors plus Buck-Boost converter) were used to determine the efficiency of these devices working at different conditions. As a result of this training, the network acquires the knowledge necessary to determine the most efficient AES current under different conditions. The optimality will depend on how many sets of data (or the driving conditions) are used to train the network. The evaluation process and results obtained are analyzed in the next sections. V. URBAN OPERATION TESTS Once all supporting parts of the system (power and control) where tested and ready, the evaluation process was prepared. The goal was to determine and quantify the improvements in vehicle performance due to the use of the Auxiliary Energy System (AES). This assessment allows seeing the technical and economic contribution of this kind of equipment to pure electric vehicles and the possible application to hybrid vehicles. The variables to be measured where: available power increase (kw) and energy efficiency increase (km/kwh) due to the use of the AES. Therefore, a protocol to measure these variables was established.

16 As the AES has been conceived for urban driving, a 14 km urban route was established for testing. The circuit had slow and fast driving portions with stops every one or two hundred meters. The stops where introduced to simulate congested urban driving conditions. During the tests, a monitoring system stored currents and voltages from batteries and ultracapacitors, and vehicle speed. Several tests where performed, computing total energy used in four different conditions: i) without regeneration, ii) with battery-only regeneration, iii) with AES using the heuristic-based algorithm, and iv) with AES using the optimized algorithm. The results of these tests are summarized in Table 1. Table 1. Urban tests results summary. Drive City Circuit (Km) Kwh Used Ah Used Km/KWh Km/KWh Improvement Batteries without Regeneration Batteries with Regeneration % Batteries with AES (SOC Control) % Batteries with AES (Optimal Neural Network Control) % The results shown in Table 1 demonstrate that a measurable improvement is achieved with the use of AES in battery powered EVs. Even better, the use of optimization tools in the energy management algorithm allows getting superior results when compared to those obtained using an algorithm based on heuristics. Without the AES, the intensive cycling of batteries wears them down and produces important energy loses. By contrast, the AES maintained battery current under 35 A and over -5 A, avoiding high loses and battery deterioration.

17 Figure 11(A) shows that, without support from the AES, the battery voltage drops below 300 V with load currents higher than 80 A. By contrast, with AES, battery voltage drops only to 315V with similar currents, as shown in Fig. 11(B). On the other hand, the same figures show huge regenerative battery currents without the AES (higher than -80 A). With AES, regenerative currents go almost entirely into the ultracapacitor, increasing the energy efficiency of the system, and protecting batteries from damage AES Current (A) Load Current (A) Battery Current (A) (a) Time (s) Battery Voltage (V) (b) Time (s) AES Current (A) Load Current (A) Battery Current (A) Battery Voltage (V) Figure 11. System currents and battery voltage for comparison: (a) with the AES disabled, (b) with the AES enabled.

18 VI. ECONOMIC EVALUATION An AES produces an improvement in maximum power capability, vehicle efficiency (km/kwh) and autonomy. These facts would probably extend the battery life due to reduction in maximum power demanded. They will also improve performance and produce cost-reduction benefits, which may be measured to calculate the cost-benefit relation of the AES inclusion in future EV configurations. Therefore, a cost-benefit analysis is presented, including only cost-related benefits generated by this equipment. The analysis was made in terms of total mean costs $/km of an EV powered with lead-acid batteries and compared to the costs of the same vehicle using an Ultracapacitor-based AES, as the one described on this paper. To obtain the total mean cost ($/km) of a vehicle, all present and future costs are calculated and added using a discount rate, which represents the uncertainty of future costs throughout time. The result represents the present value (PV) of all costs, which may also be expressed as a monthly payment throughout the vehicle s lifetime using an interest rate representing the cost of capital. This payment divided by the amount of kilometres per month will represent the total mean cost ($/km) of the vehicle. The analysis will be performed over a lifetime of 12 years, which corresponds to 240,000 kms. The costs considered are: the vehicle s chassis, power train and accessories 1 ; original and replacement batteries throughout lifetime 2 ; AES (ultracapacitors 3 + static converter 4 ); cost of spent energy 5, annual maintenance 6 and the residual value 7 at the end of the period analyzed. It is assumed that costs of components and maintenance will not change throughout lifecycle. Electric energy prices are supposed to follow a projected trend 5 and possible deviations from this trend are not considered.

19 The base case is the test vehicle already described in the Introduction (Chevrolet LUV ). This case is compared to the same vehicle with the AES, installed for peak power support. The AES is composed of a 20-Farad Ultracapacitor bank, a buck-boost converter and an energy control system. For the evaluated alternative, two different assumptions where tested for the increase in battery life: first, a rather optimistic 50 % extra life and second, a more realistic 20% extra life. Table 2. Total mean costs comparison with batteries as main energy source. Component Batteries Only Batteries (50%+) + Ucaps Batteries (20%+) + Ucaps Vehicle (1) $ 8,000 $ 8,000 $ 8,000 Bateries (2) $ 16,181 $ 11,595 $ 13,897 Ucaps (3) + static converter (4) $ 0 $ 5,160 $ 5,160 Total cost of energy (5) $ 3,288 $ 3,024 $ 3,024 Maintenance (6) $ 2,856 $ 2,856 $ 2,856 Residual value (7) -$ 450 -$ 724 -$ 578 PV of total costs (8) $ 29,874 $ 29,910 $ 32,358 Total average cost ($/km) $ $ $ Cost change percentage - 0.1% 8.3% Table 2 shows all of these costs and the corresponding present value and total average cost for each case analyzed. Total average costs, calculated for each case, show that AES convenience is relative to the battery life extension it produces. If batteries last 50% longer (an optimistic scenario) with the AES installed, the total mean cost is almost the same as the cost of the vehicle working on batteries only. However, the advantage of better vehicle performance in terms of power and efficiency will increase customer satisfaction. On the other hand, if batteries last 20% more cycles (a more realistic scenario), then the total average costs are 8.3% higher for the vehicle equipped with the AES. Hence, in terms of costs only, the system described in this paper would only justify its inclusion in a lead-acid battery equipped vehicle if battery life extension were equal or higher

20 than 50%. If the customer satisfaction factor were included in the analysis, then battery life extension may not need to reach a 50% life extension to be cost-effective. Presently, a new, state-of-the-art, Sodium/Nickel-Chloride (ZEBRA) battery is being installed in the Chevrolet LUV EV. This battery increases the range of the vehicle to around 150 kms per charge and weights only 240 kgs. With a cycle life of around 1,600 cycles, only one ZEBRA battery (with a total cost of around US$ 12,000) could cover the entire 12 years operation period. As this battery weights less than half the weight of previous lead-acid battery pack, a better efficiency (km/kwh) should also be expected. The situation is being evaluated to determine costbenefit results when compared with the previous lead-acid hybrid EV. Another interesting scenario was analyzed, which is currently being explored by automakers: the implementation of fuel cell-powered vehicles. For this analysis, three power configurations where evaluated: the vehicle powered by fuel cells only, fuel cells plus Li-ion batteries and fuel cells plus ultracapacitors. The efficiency for a fuel cell vehicle (running on gas hydrogen from electrolysis) from the power grid to the wheels, can be derived from the composition of typical efficiencies of all conversion processes involved: electrolysis (72 %), fuel cell (54 %) and electric drivetrain (89 %), with a total integrated efficiency of 34 % [19]. The energy efficiency achieved by the wheel-to-road conversion was assumed at 8 km/kwh, equivalent to run 300 miles with 60 kwh coming out of the wheels [19]. An extra 18 % efficiency improvement was considered for the fuel cell plus batteries configuration (because of regeneration savings), and a 24 % of extra improvement for the configuration running on fuel cells plus ultracapacitors (because of better specific power than Li-ion batteries). These numbers where obtained from experience acquired while testing the AES with and without regeneration.

21 Table 3. Total mean costs comparison with Fuel Cell as main energy source. Component FuelCell Only FuelCell + Batteries FuelCell + Ucaps Vehicle (1) $ 8,000 $ 8,000 $ 8,000 Bateries (9) $ 0 $ 5,706 $ 0 Fuel Cell (10) $ 20,000 $ 4,000 $ 4,000 Ucaps (3) + static converter (4) $ 0 $ 0 $ 5,160 Total cost of energy (5) $ 3,670 $ 3,110 $ 2,960 Maintenance (6) $ 2,856 $ 2,856 $ 2,856 Residual value (7) -$ 238 -$ 238 -$ 238 PV of total costs (8) $ 34,287 $ 23,433 $ 22,737 Total average cost ($/km) $ $ $ Cost change percentage % -33.7% Costs in Table 3 show clearly how fuel cells represent an important percentage in the cost of structure. Therefore, a small fuel cell drastically reduces mean costs, which can be seen in the case of combination with batteries or ultracapacitors. The use of a much smaller fuel cell (20 kw in hybrid cases) is compensated by the power support from batteries or ultracapacitors during peak power demand. However, this cannot be sustained for a long time due to the limited amount of energy stored in these devices. Hence, the use of hybrid configurations would limit the amount of continuous time allowed to drive at maximum power, making it unsuitable for sustained high speeds or hill climbing. Nevertheless, these configurations could still perform more than well in urban conditions and even in highways at reasonable speeds (assuming good aerodynamics), and the mean cost is drastically reduced in 33.6 % and 31.8 % for combinations with ultracapacitors and batteries, respectively. VII. CONCLUSIONS An Auxiliary Energy System (AES) based on ultracapacitors and a DC-DC converter has been designed, implemented and evaluated. The Buck Boost topology selected for the DC-DC converter design has behaved adequately, achieving in very small equipment, a satisfactory thermal

22 control, low parasitic inductances and low current-ripple amplitude. The AES was installed in a Chevrolet LUV pick up truck powered by lead-acid batteries. A control and monitoring system was implemented on a DSP from Texas Instruments. Two different control algorithms where implemented in the control module. The first one, based on heuristics, establishes an inverse relation between the energy stored in the capacitors and the vehicle s kinetic energy. The second one uses a neural network, which has been off-line trained to imitate the numeric solutions of an optimization model. A series of tests were performed with and without the AES, using both algorithms. The results where evaluated from an economic approach, which showed that a battery life-increase of about 50% was required to compensate for the AES costs. Similar analyses were performed for a hypothetic fuel-cell-powered hybrid vehicle under three scenarios: i) without the AES (just a bigger fuel cell to cope with peak power demand), ii) using an ultracapacitor-based AES, and iii) using a Li-ion battery-based AES. Results showed that the case without AES was notoriously more expensive than the other two alternatives. On the other hand, the ultracapacitorbased AES showed to be the least expensive combination. VIII. ACKOWLEDGEMENTS The authors want to thank Conicyt through Project Fondecyt and , for the support given to this work. IX. REFERENCES [1] John Voelcker, Top 10 Tech Cars, IEEE Spectrum, Vol. 41, Nº 3, March 2004, pp [2] Michel Wehrey, What is New with Hybrid Electric Vehicles, IEEE Power and Energy Magazine, Vol. 2, Nº 6, Nov/Dec 2004, pp [3] Victor Wouk, Hybrids: Then and now, IEEE Spectrum, Vol. 32, Nº 7, July 1995, pp [4] Osamu Fuji, The Development and Application of Hybrid Vehicles. 19th Electric Vehicle Symposium, Busan, Korea, October, 2002.

23 [5] David Hermance and Shoichi Sasaki, Hybrid Electric Vehicles take to the streets, IEEE Spectrum, Vol. 35, Nº 11, November 1998, pp [6] Per Rutquist, Optimal Control for the Energy Storage in a Hybrid Electric Vehicle 19 th Electric Vehicle Symposium, Busan, korea, October, [7] Glenn Zorpete, Super Charged (an article about Ultracapacitors), IEEE Spectrum, Vol. 42, Nº 1, January 2005, pp [8] Richard Post and Stephen Post, Flywheels, Scientific American, Vol. 229, Nº 6, December 1973, pp [9] Richard Post, Kenneth Fowler, and Stephen Post, A High Efficiency Electromechanical Battery, Proceedings of the IEEE, Vol. 81, Nº 3, March 1993, pp [10] Tom Gilchrist, Fuel Cells to the Fore, IEEE Spectrum, Vol. 35, Nº 11, November 1998, pp [11] Fact Sheets: Vehicles and Fuels, Electric Vehicle Batteries, Energy Efficiency and Renewable Energy, U.S. Department of Energy, may [12] Jin-uk Jeong, Hyeoun-dong Lee, Chul-soo Kim, Hang-Seok Choi and Bo-Hyung Cho, A Development of an Energy Storage System for Hybrid Electric Vehicles Using Supercapacitor. 19th Electric Vehicle Symposium, Busan, Korea, October, [13] J. Dixon, M. Ortúzar, R. Schmidt, G. Lazo, I. Leal, F. García, M. Rodríguez, A. Amaro and E. Wiechmann, Performance Characteristics of the First, State-of-the-art Electric Vehicle Implemented in Chile, 17th Electric Vehicle Symposium, Montreal, Canada, October, [14] M. Rashid, DC Choppers, Power Electronics. Circuits, Devices and Applications, 2 nd edition. Prentice-Hall, 1993, Chapter 7, pp [15] Powerex, Inc. IGBTMOD and Intellimodtm Intelligent Power Modules, Applications and Technical Data Book. First edition. Published by Powerex, Inc [16] A. Chapman, Heat Tansfer, Third Edition. Macmillan Publishing Co. Inc., 1974, pp 72-76, [17] J. W. Dixon and Micah Ortúzar, Ultracapacitors + DC-DC Converters in Regenerative Braking System. IEEE Aerospace and Electronic Systems Magazine, Volume 17 - Nº 8, pp.16-21, August [18] J. Moreno, J. Dixon and M. Ortúzar, Energy Management System for an Hybrid Electric Vehicle, Using Ultracapacitors and Neural Networks. IEEE Electric Power Propulsion Conference, Paris, France, October on CD-ROM. [19] S. Eaves and J. Eaves, A Cost Comparison of Fuel-Cell and Battery Electric Vehicles, Journal of Power Sources, 130(1-2), pp (2004). [20] C. C. Chan and Y. S. Wong, Electric Vehicles Charge Forward, IEEE Power and Energy Magazine, Vol. 2 Nº 6, Nov-Dec 2004, pp [21] M. Sund and P. Trice (2001).Maxwell Technologies to Supply PowerCache. Ultracapacitors to General Motors for Hybrid Electric Vehicles., Press release:

24 [22] Energy Information Administration, Annual Energy Outlook 2005, Report # DOE/EIA- 0383(2005). [23] Taylor, D. (2003) Simplified Life Cycle Cost Analysis of Plug-in HEVs, Engine Dominant HEVs and Conventional Vehicles in EPRI Hybrid Electric Vehicle Working Group. [24] U.S.Treasury, (2005) Daily Treasury Yield Curve Rates, march 2005, X. NOTES 1 Arbitrary estimated value of vehicle without energy source of US$8000. It represents an approximate cost of structure, accessories and drive train. 2 Cost of batteries for 12 years of operation, based on $150/kWh [20]. On batteries only configuration (base case), batteries life is km. On the 50%+ case, batteries life is km; for the 20%+ case, batteries life is km. 3 Projected costs of ultracapacitors is US$30 per unit (2,700 Farad, 2.5 V each) [21]. 4 Arbitrary estimated cost of static converter is US$ Cost of energy is based on projections by Energy Information Administration [22]. 6 Cost of maintenance is an arbitrary estimated value of US$400/year. 7 The residual value represents the approximate price for a 12 year old vehicle in good conditions or its parts. 8 The Present Value (PV) of costs is the sum of all discounted costs. An 8% discount rate was used for vehicle s future costs [23]. A 3.2% discount rate was used for cost of capital, equivalent to the rate of a (1 year to maturity) security from the US Treasury [24]. 9 The estimated life of a Li-ion battery for a hybrid vehicle is 6 years. 10 Fuel cells projected cost is $200/kW [20].