Equivalent circuit model for electrical performance characterisation of lithium-ion batteries under thermo-mechanical loads

Abstract

The diverse applications of lithium-ion batteries (LiBs) and the increasing technology for new commercial applications in electric vehicles (EVs) have driven extensive research, expecting to enhance their performance, longevity, and safety. Most of the studies cover the impact of temperature, and only a few encounter the side effects of mechanical vibration. In automotive applications, temperature and vibration have been largely overlooked separately in existing models, leading to unpredictable results. However, when both parameters are reviewed together, they encompass significant variations in resistance, voltage and power. This paper proposes to fill the gap by developing an equivalent circuit model (ECM) that couples temperature and vibration under varying conditions of state of charge (SOC). While many ECMs have been proposed in the literature to improve the battery management system (BMS), none of them have explored the impact of vibrations. The methodology involves the experimental quantification of the cell characteristics by electrochemical impedance spectroscopy (EIS) and hybrid pulse power characterisation (HPPC). The ECM revealed that vibration increased all resistive parameters, particularly at high SOC and low temperature. The quantified variation due to vibration loads is given as the incremental values of the ohmic resistance (R0), interfacial resistance (Rint), diffusion losses (R1) and capacitance (C1) (∆R0, ∆Rint, ∆R1, ∆C1), showing an increase with different trends at different temperature and SOC values. At 100% SOC, ∆R0 varies between 0.67 mꭥ at 0 °C to −0.01 mꭥ at 40 °C. When the temperature increases, the effects of vibration are less noticeable, proving that temperature is more common in the resistance. Nevertheless, the changes in diffusion resistance and capacitance were complex, with some parameters decreasing under specific conditions. The ECM was validated using three urban dynamometer driving schedule (UDDS) cycles performed at various SOC and temperatures, yielding an average root mean square error of 14 mV, which is of comparable accuracy with other ECMs in the literature.