Abstract
Ensuring the mechanical integrity and safety of lithium-ion battery cells under abusive conditions is critical for applications such as Electric Vehicles (EVs). This work presents a new unified and generalised Finite Element Method (FEM) framework for lithium-ion battery cell modelling that can serve both detailed investigations and scalable applications. This integrated approach addresses the need for both physical accuracy and computational efficiency and provides a structured pathway for researchers and engineers to simulate battery mechanical behaviour and failure without requiring extensive experimental data. The proposed approach begins with the development of a detailed heterogeneous model that accurately replicates the jelly-roll structure to establish benchmark conditions under four loading scenarios: indentation, three-point bending, compression, and impact. These models are qualitatively assessed against digitised experimental data from the literature, demonstrating their reliability in predicting mechanical responses and stress concentrations. This detailed model then serves as a benchmark for developing a simplified homogeneous model using a Design of Experiments (DOE) approach, optimizing its parameters for computational efficiency while preserving predictive accuracy. The transition from heterogeneous to homogeneous modelling enables efficient simulation of larger-scale systems, such as battery modules, without compromising failure prediction. Furthermore, the methodology provides a valuable tool for researchers and engineers who may lack access to experimental data, allowing for virtual testing and optimization of battery designs under mechanical abuse conditions. The proposed approach contributes to a deeper understanding of localized failure phenomena and supports enhanced safety assessments for lithium-ion batteries.