Estimation of thermophysical properties of a pouch-type Li-ion battery using an inverse methodology

Abstract

Abstract The growing popularity of electric vehicles highlights the crucial role of batteries. Effective battery thermal management is crucial for improving performance, reliability, and safety, especially in tropical areas where overheating is a key challenge. This necessitates a comprehensive understanding of the thermophysical properties of batteries. The present study concerns the estimation of the temperature-dependent orthotropic thermophysical properties (kx , ky , kz , cp ) of the active material in a pouch-type Li-ion battery using an inverse methodology. An experimental study is conducted on a commercial AMP20M1HD-A Li-ion battery to measure the surface temperature at various locations using thermocouples. The forward model consists of the three-dimensional unsteady conduction problem and is solved in COMSOL using experimental boundary conditions. The data generated is used to train an Artificial Neural Network, which acts as a replacement for the forward model. The Metropolis Hasting-Markov Chain Monte Carlo algorithm along with the Bayesian inference inverse model is used for analyzing the posterior distribution and the average estimates for thermophysical properties are obtained. The temperature dependence study shows a significant correlation between temperature and battery thermophysical properties. The accuracy of the employed inverse model is validated by obtaining the surface temperature using the estimated thermophysical properties and comparing it with the measured surface temperature.