An immersed interface Adaptive Mesh Refinement algorithm for Li-ion battery simulations. II. Multi-dimensional extension and separator modeling

Abstract

We present a parallelizable, Adaptive Mesh Refinement (AMR)-compatible solver for computing solutions to multi-dimensional battery models, with added capability to resolve the complex geometries of battery components. Development is carried out within the finite volume framework, with diffuse and sharp (cut-cell) Cartesian immersed interface methods used to model interactions at material interfaces, allowing mesh generation to be carried out rapidly. The solver integrates seamlessly with hierarchical AMR, achieving accelerated computational efficiency while preserving solution accuracy. The parallelizable nature of the solver means that it can be run on massively parallel supercomputers to further reduce computational time. The performance and capabilities of the solver are demonstrated using the pseudo-three-dimensional model, which allows us to present for the first time in the literature a numerical study that directly investigates the effects of separator membrane microstructure on battery electrochemical performance, where the separator microstructures are resolved within the model. The solver was carefully validated under various operating conditions, with grid-aligned and non-grid-aligned battery boundary shapes, on uniform and AMR grids. The use of AMR was shown to significantly reduce computational time for multi-dimensional problems. The solver was also shown to demonstrate good “strong scaling” parallel performance. When using the solver to investigate the effects of separator microstructure, the influences of pore size and constrictivity on electrochemical performance were examined. Through a showcase study performed using realistic separator microstructures, the potential of the solver to be used as an effective tool for design and optimization of next-generation batteries was also demonstrated.