Multi-timescale Parametric Electrical Battery Model for Use in Dynamic Electric Vehicle Simulations

Abstract

Simulation of electric vehicles (EVs) over driving schedules within a fully dynamic EV simulator requires battery models capable of accurately and quickly predicting state of charge (SOC), $I$ – $V$ characteristics, and dynamic behavior of various battery types. An electric battery model utilizing multiple time constants, to address ranges of seconds, minutes, and hours, is developed. The model parameters include open-circuit voltage, series resistance, and equivalent RC circuits, with nonlinear dependence on battery SOC. The SOC captures effects from discharge and charge rate, temperature, and battery cycling. Thermal modeling predicting real-time battery temperature is introduced. One focus of this paper is presenting a systematic and generic methodology for parameter extraction as well as obtaining SOC factors through reasonable test work when evaluating any given lithium-ion (Li-ion), nickel-metal hydride, or lead-acid battery cell. In particular, data sets for a Panasonic CGR18650 Li-ion battery cell are tabulated for direct use. The Li-ion battery model is programmed into a MATLAB/Simulink environment and used as a power source within an existing comprehensive dynamic vehicle simulator. Validation of the Simulink model is through a battery testing apparatus with a hardware-in-the-loop driving schedule that cycles real batteries. Results from simulations and measurements of Li-ion battery packs show that the proposed battery model behaves well and interacts appropriately with other subcomponents of the vehicle simulator.