Immersed Interface Adaptive Mesh Refinement Methods for Lithium-ion Battery Simulations

With growing concerns over energy depletion and increasing global carbon dioxide emissions, many national governments have announced commitments to phase out internal combustion engine vehicles by 2040 [3, 4]. This transition highlights the critical and urgent need for advancements in battery technology to promote sustainable energy solutions. Lithium-ion batteries (LIBs) are favoured in the battery sector because they possess a number of useful properties, such as high energy and power density, low self-discharge rate, and a long cycle life, to name a few [5, 6]. However, several challenges still remain that hinder their widespread adoption, including achieving even higher power and energy densities, enabling ultra-fast charging, and addressing safety concerns [7]. Numerical modelling can serve as a powerful means to gain insights into the physical and electrochemical mechanisms involved, potentially enabling the optimisation of battery performance in an efficient and economic manner. However, such numerical simulations are not without challenges. One of the major challenges in extracting more value from simulation lies in developing numerical solvers with the capability of accurately simulating the complex electrochemical processes while also resolving the intricate microstructure of battery components, all in a computationally efficient manner. This thesis develops novel immersed interface adaptive mesh refinement methods for Li-ion battery simulations. These approaches not only reduce the computational cost of solving the well-known P2D model but also enable multi-dimensional simulations that resolve the complex microstructures of battery components. Although immersed interface techniques and hierarchical adaptive mesh refinement (AMR) have been well-established in numerical simulations across a wide range of scientific and engineering disciplines [8–13], they have not yet been introduced to the battery modelling community. This work uses these two numerical techniques within the finite volume method framework to accurately and efficiently solve the battery models, facilitating investigations into the effects of regional geometric heterogeneities on battery performance. The solver is parallelised with the use of MPI, making it compatible be run on a single-core processor as well as on massively parallel supercomputers. Specifically, this work encompasses: the mathematical modelling of the problems of interest, the development of numerical techniques to construct the solver, and the application of the solver to explore how the geometric characteristics of separators and regional geometric heterogeneities induced by electrode-level cracks/voids impact battery performance. As a building block in the solver development, the novel numerical techniques are first implemented to compute solutions for the P2D model. The resulting algorithm reduces the computational cost of solving the full-order P2D model while retaining its accuracy. The solver has been validated against numerical and experimental results across a wide range of discharge/charge rates and various operating conditions. Its computational efficiency has proven to be competitive with state-of-the-art P2D solvers, demonstrating its potential for real-time applications. Subsequently, the method is further developed to be able to solve the multi-dimensional problems. The performance and capabilities of the solver are demonstrated using the pseudo- three-dimensional (P3D) model, which allows some regions inside the battery cell being resolved with explicit representations of their geometries within a complete model of battery physics. The P3D solver has been utilised to investigate how regional geometric hetero- geneities in various battery components affect electrochemical performance. These studies cover the effects of separator microstructure, cracks and voids in electrode materials, crack growth, and electrolyte infiltration into void regions during battery operation. The insights gained from the numerical investigations are of significant value for the material selection, design and optimisation of various battery components, and have not been previously reported in the literature. The capabilities of our solver to resolve arbitrarily complex geometries makes it a valuable design tool for optimising the sophisticated mi- crostructures of separator membranes and electrode materials, as well as for investigating battery degradation mechanisms. The appealing computational efficiency of the solver paves the way for the development of next-generation batteries.

Name dc:type.qualificationname

Doctor of Philosophy (PhD)

Level dc:type.qualificationlevel

Doctoral

Grantor dc:publisher.institution

University of Cambridge

Year dc:date.issued

2025

Author dc:creator

• Lu, Jiawei

Advisor dc:contributor.advisor

• Nikiforakis, Nikolaos

• Computational Physics
• Lithium-ion battery simulations
• Separator microstructure
• Multiphysics
• Finite volume method
• Cartesian grid
• Immersed boundary method
• Adaptive mesh refinement
• Complex geometry
• Parallelisation

Licence

• https://www.repository.cam.ac.uk/bitstreams/301f9716-7137-4ecf-aced-8a935d8c803c/download
• http://purl.org/NET/rdflicense/allrightsreserved

Language dc:language

eng

DOI dc:identifier.doi

https://doi.org/10.17863/CAM.124581

OAI identifier oai:identifier

oai:www.repository.cam.ac.uk:1810/394817