InterPore2025

Upscaling of lithium-ion battery models: from the pore-scale to the cell-scale through homogenization

20 May 2025, 09:20

15m

Oral Presentation (MS19) Elastic, electrical, and electrochemical processes and properties in porous media MS19

Speaker

Mr Alessio Lombardo Pontillo (Politecnico di Torino, Italy)

Description

Lithium-ion batteries (LiBs) are currently the leading energy storage technology for applications ranging from portable devices to aerospace vehicles. They are also expected to play a significant role in net-zero energy buildings, with global battery demand projected to increase by approximately 27% annually until 2030. Their advantages include a high energy-to-weight ratio, low self-discharge, and decreasing production costs, making them a top choice for future non-fossil fuel-powered systems [1]. However, improving LiB performance and lifetime requires addressing challenges such as non-uniform current distribution, which is strongly influenced by electrode morphology.
Graphite, the most commonly used material for anodes, has a flake-like anisotropic structure due to its hexagonal atomic arrangement. This geometry is crucial to the electrochemical processes in LiBs but is often oversimplified in models, which typically assume spherical particles. While this approximation may suffice for cathodes, it fails to capture the complex behavior of graphite anodes, leading to inaccuracies in performance predictions. Therefore, accurately reconstructing the microscopic geometry of the electrode is vital for understanding charge and discharge processes on a macroscopic scale [2].
This study focuses on the geometrical characterization of graphite electrodes in a half-cell setup. The goal is to develop an upscaled model using homogenization techniques, enabling faster and more computationally efficient simulations. The team began by identifying a particle shape that balances realism and computational cost. An ellipsoidal shape was chosen, with two larger dimensions derived from particle size distribution (PSD) data and aspect ratios evaluated from scanning electron microscope (SEM) images. This approach provides a more realistic representation of graphite particles compared to spherical approximations, as confirmed by experimental porosity evaluations.
The electrode geometry was created through a semi-automated process using Python-based tools. Initially, the software Yade-DEM was used to generate a packing of spheres based on PSD data. These spheres were then transformed into flattened ellipsoids using Blender, and a rigid-body simulation compacted the layer to reduce voids. The central portion of the packed layer was extracted and used as input for pore-scale simulations in COMSOL 6.1. This Python-controlled workflow allows the creation of various electrode morphologies and particle size distributions.
Homogenization techniques were applied to derive macroscopic properties from microscopic behavior. This involved rewriting equations in dimensionless form to extract parameters like the Damköhler and Péclet numbers, which determine the feasibility of scale separation. A closure problem was solved on a periodic unit cell, containing polydisperse ellipsoids and semi-ellipsoids, to calculate effective diffusivity and conductivity values for the solid and liquid phases [3-4]. These effective properties were then used to create and solve a homogenized model.
The homogenized model produced results comparable to those of the computationally intensive pore-scale simulations but required significantly less time and resources. The flexible script enables the generation of diverse electrode geometries, facilitating the creation of a large dataset across a wide parameter space. This dataset could be used to train neural network models, offering a quick surrogate method to explore various electrode configurations. This approach represents a significant step toward optimizing LiB design and performance.