High-Performance Computing in Battery Development: From Pore Scale to Continuum

TL;DR

This work addresses electrolyte filling in porous Li-ion battery electrodes and its impact on electrochemical performance due to gas entrapment and wettability. It presents a HPC-driven workflow that couples pore-network modeling (OpenPNM) and lattice Boltzmann methods (Palabos) with electrochemical simulations (BEST) to predict filling behavior and battery performance. Results show notable gas entrapment with LBM, while PNM saturates fully; perforated electrode microstructures improve wettability and state-of-charge homogeneity, enhancing high-current operation. The study demonstrates scalable coupling of low- and high-fidelity models and highlights how MPI-enabled HPC enables efficient exploration of large parameter spaces for co-design of microstructure and filling processes.

Abstract

An application for high-performance computing (HPC) is shown that is relevant in the field of battery development. Simulations of electrolyte wetting and flow are conducted using pore network models (PNM) and the lattice Boltzmann method (LBM), while electrochemical simulations are conducted using the tool BEST. All aforementioned software packages show an appropriate scaling behavior. A workflow for optimizing battery performance by improving the filling of battery components is presented. A special focus is given to the unwanted side effect of gas entrapment encountered during filling. It is also known to adversely affect the electrochemical performance of batteries and can be partially prevented by appropriate microstructure design such as electrode perforation.

Figures (5)

• Figure 1: Flowchart of the workflow. The data flow and the relationships between the different steps are illustrated. The boxes are colored with respect to the software used. Gray labels depict steps that are not directly related to any specific software tool.
• Figure 2: Network extraction from structurally-resolved image data. Left: 2D pore space segmented into two pores (yellow and green). The disk sizes (blue) correspond to the inscribed diameters of the pores and are connected by a throat (opaque rectangle). Right: Extracted full network embedded into the corresponding 3D image of a battery cathode. The pore color corresponds to the pore volume. The throat diameters are depicted for topological reasons only and do not relate to the physical size.
• Figure 3: Simulation results for electrolyte filling using LBM and PNM. Left: Entrapped gas (gray) in the pore space at the end of the filling process. Electrolyte and active material are fully transparent. Right: Pressure-saturation curves determined using LBM (dots) and PNM (lines) simulations for a selection of contact angles $\theta$.
• Figure 4: Electrochemical aspects. a) SOC distribution of lithium in unstructured electrode (top) and structured electrode (bottom). The structuring improves the homogeneity of the SOC distribution. b) Electrochemical performance for operation currents of $j\geq 6\text{mA/cm}^2$.
• Figure 5: Performance Test. Left: Scaling of fully resolved and homogenized LBM electrolyte filling simulations using Palabos. Right: Scaling of a matrix multiplication in Python using NumPy in a multi-threaded environment to illustrate the scaling of non-MPI computations in the workflow.