Inconsistent evolutionary analysis of multi-level physical model for lithium-ion batteries considering calendering process uncertainties

Abstract

Due to the variation in the calendering process, there are inconsistencies between battery cells that cannot be eliminated, which affects the energy utilization efficiency, safety, and reliability of the battery system. The deviations occur in the calendering process resulting in uncertainties in the structural parameters of the electrode. To analyze the evolutionary mechanism of inconsistency arising from the uncertainties in the calendering process, a stochastic, multi-level physical model approach combined with sparse polynomial chaos expansions is presented. The physical model is verified by voltage and temperature experiments. This approach enables to establish the relationship between calendering parameters variation and the inconsistency performance between cells. The performance of battery inconsistency has been quantified as a function of calendering process deviations at the manufacturing level. The most important uncertainty sources are identified by global sensitivity analysis. Based on the multi-level physical model approach, the battery module connected in series is constructed to study the effect of the calendering variations on the energy utilization efficiency of the battery module. The results suggest that the electrode thickness has the most significant effects on the battery capacity, resistance, and temperature. Meanwhile, the deviations of the mass load and electrode thickness need to be effectively controlled to achieve a better consistent performance between cells. Such information allows for more efficient design and targeted quality control, thereby reducing production costs.