Accurate Estimation of State of Charge Using Reduced Order Electrochemical Model

Abstract

To estimate the dynamics of Li-ion cells (state of charge, cell voltage, etc.), various electrochemical models based on uniform reaction kinetics have been developed. Due to uniform reaction rate assumption, accurate prediction of cell behaviour is difficult. Also, many detailed physics-based models have been developed to improve accuracy of estimation but due to the higher computational cost, real time estimation of the cell dynamics is still limited. By keeping the above limitations in mind, present work focuses on developing an accurate model with low or moderate computational cost. Our model considers the analytical form for the non-uniform reaction rates and the polynomial approximation for concentration profiles and then uses efficient computational methodology in Python to simultaneously solve the involved partial and ordinary differential equations. Due to non-uniform consideration of reaction kinetics and the computational methodology adopted, the model is called as Non-Uniform Modified Reduced Order Model. This reduced order model accurately predicts the test data of large format commercially available Li-ion batteries for various C-rates. Further the robustness of model is proven by reproducing the results published using full pseudo-2-dimensional model in commercially available Multiphysics software for C-rate as high as 5C.