In-Vehicle Testing and Computer Modeling of Electric Vehicle Batteries

A combined simulation and testing approach has been developed to evaluate battery packs in real driving conditions as opposed to purely experimental testing. The new approach is costeffective, greatly accelerates battery development cycle, and enables innovative battery design and optimization. Several fundamental models for valve-regulated lead-acid (VRLA), nickel-metal hydride (NiMH), and Li-ion batteries were developed to simulate in-vehicle performance of battery packs on field trips. Field testing was performed for both VRLA and NiMH batteries using Penn State University’s electric vehicle, the Electric Lion. The computer results of voltage and current responses were compared to the field-test data to provide model validation. Furthermore, the model results for the evolution of internal conditions inside batteries were used to reveal their limiting mechanisms. Introduction In recent years there has been increasing use of battery modeling and simulation in the exploration and development of advanced batteries for electric and hybrid-electric vehicles. A thorough understanding of battery systems from the point of view of performance, safety, longevity, etc. is critical for these applications. Using traditional testing methodologies to evaluate battery characteristics and obtain a performance envelope under a myriad of operating conditions and environments is a time consuming and an almost impossible task. The present work aims to develop a new approach to evaluating electric and hybrid-electric vehicle batteries by integrating first-principle computer simulation with in-vehicle testing. The approach will be demonstrated specifically with lead-acid, Ni-MH and Li-ion batteries, and the focus is placed on obtaining scientifically sound results for battery packs in realworld driving conditions. Such data are critically needed to test the ultimate ability of a battery model to predict actual EV trips. By combining simulation with field testing at our test track and dynamometer facilities, we shall demonstrate that the performance of a battery pack in actual vehicle trips can be accurately and rapidly predicted. The integrated approach will overcome many existing barriers to the development of EV/HEV batteries by offering the following new capabilities: • Enable innovative design and improvement by providing insight into the internal conditions of a battery, such as active material utilization, electrolyte distribution, etc. • Accelerate the development cycle of advanced batteries by allowing engineers to evaluate many design alternatives before building a physical prototype. • Provide accurate predictions of battery state of charge on-board and hence vehicle driving ranges corresponding to specific battery systems, driving profiles, and terrain conditions. • Create digitally based (virtual) battery systems for vehicle simulation and integration. • Facilitate the development of infrastructure such as rapid chargers and battery management systems via high-fidelity computer models rather than physical battery modules and packs. In-Vehicle Testing Test Batteries Test batteries included Horizon lead-acid batteries of 85Ah and Panasonic Ni-MH batteries of 95Ah for electric vehicles. Test Vehicle All field testing was performed in the Electric Lion (see Figure 1). The Electric Lion began life as a 1992 Ford Escort Station Wagon. It has since been converted into a range-extending series hybrid electric vehicle (HEV) by the student members of the Society of Automotive Engineers (SAE). A series HEV is an electric vehicle with an auxiliary power unit (APU) that charges the battery pack. The battery pack consists of 12 12V modules. The pack has a capacity of 15 kWh and a nominal voltage of 144 Volts. Two Solectria GTX-20 AC induction motors, which are connected to the front wheels through one speed transmissions, make up the drivetrain. The motors are controlled by Solectria AC-325 controllers and Penn State purpose-built Z80 microprocessor controls. Figure 1. Test vehicle and track at PTI. Test Track The test track also shown in Figure 1, operated by the Pennsylvania Transportation Institute (PTI), is the site of all federal bus testing. Located at this facility are a one-mile oval track, vehicle durability course, impact pendulum, emission testing facility, as well as other equipment. Besides bus testing, crash tests between vehicles and common obstacles found on the road (such as signs and barriers) are performed. Dynamometer The Penn State HEV laboratory owns a single roll (soon to be converted into a double roll) eddy current dynamometer manufactured by Clayton Industries. See Figure 2. It is controlled by a personal computer using the Virtual Test Track software. With this software, different types of tests can be performed including constant force, constant power, and even a 1⁄4 mile drag race. Data on the power, torque, speed, and acceleration of the vehicle during these tests is collected and can be displayed graphically or exported to an excel file to be plotted. Figure 2. Clayton dynamometer Computer Modeling Since 1994 we have been developing mathematical models for Lead-Acid, Ni-MH, Li-ion and Li-polymer cells using advanced computational fluid dynamics (CFD). CFD technology is a numerical tool used to analyze and optimize fluid flow, mass and thermal transport, and related phenomena (e.g. chemical reactions) that may simultaneously take place in a complex system. The broad scope, power, convenience, user interface, and preand post-processing capabilities developed for CFD over the last decade have made the technique very attractive. Over the past few years, we have successfully adapted the CFD modeling technique for a variety of battery systems such as VRLA, Ni-MH and Li-ion batteries for electric and hybrid vehicles, alkaline primary cells for portable electronics, primary Li/SOCl2 batteries for military applications, and Ni-H2 batteries for aerospace application . Our current modeling capabilities can include multiple electrode reactions, charge transfer, multi-component species transport via diffusion, convection and migration, solid state diffusion, gas generation and transport, and heat generation and transport. CFD codes are particularly suited for comprehensive modeling of electrochemical, thermal and gassing behaviors. The CFD battery codes are also Matlab and Simulink compatible so that they are readily integrative in vehicle simulations. The computational efficiency is between 10 and 100 times (10-100x) faster than real-time testing. Recently, an Internet-based battery simulation environment was also created to provide convenience for using battery models. This online simulation system allows users to submit input files, execute simulations, and receive results via the Internet in an encrypted format anywhere, anytime. With fully interactive preand post-processing interfaces as well as visualization and animation tools, this system is particularly useful for battery designers and vehicle builders to collaborate via the Internet in real-time. Results and Discussion In the following, in-vehicle testing of Horizon lead-acid and Panasonic Ni-MH batteries on the dynamometer is described and test data are compared to the corresponding computer simulations. Subsequently, a simulation of battery pack performance during a field trip is presented along with its comparison with actual data. Finally, additional modeling capabilities are illustrated through applications to several battery systems. In-Vehicle Evaluation Figure 3 displays results of drive testing and corresponding simulation of Horizon lead-acid and Panasonic Ni-MH batteries in Electric Lion on the dynamometer, respectively. It can be seen that the computer simulations reproduce reasonably well the general trends and fluctuations of voltage responses for both types of batteries. Note also that the actual testing took about one hour in both cases, but the computer simulations required only one minute of CPU time on a standard PC. Figure 3. Comparison of in-vehicle test data and simulation results: (a) Horizon leadacid battery, and (b) Panasonic Ni-MH battery. Figure 4 further shows a case study to predict battery current and voltage characteristics of a lead-acid battery pack during an S-10 pickup field trip. It is encouraging to see that the predictions follow quite closely with the measured fluctuations. Figure 4. Predicted and measured current and voltage responses of a lead-acid battery pack in an S-10 electric pickup field trip. The field data were obtained from Hawaii NDC. Additional Modeling Capabilities The valve-regulated lead-acid (VRLA) battery model has also been extensively validated against laboratory testing under complex EV battery test procedures such as the dynamic stress test (DST). A detailed account of this work was presented in Ref.. A sample result is displayed in Figure 5, which demonstrates the high-fidelity of our model in simulating a commercially available lead-acid battery module undergoing the complex DST. It can also be seen from Figure 5 that the experimental DST cycling has to be terminated at the 80% depth of discharge (DOD) (at about 76 min) in order to avoid battery overdischarge and hence permanent damage, while the virtual simulation can continue the cycling beyond 80% DOD 0 10 20 30 40 50 60 11.0 11.5 12.0 12.5 13.0 Comparison of Simulation with Test Data Dynamometer Test 2 Voltage vs. time Simulation Battery 1