Optimization of the microstructure of carbon felt electrodes by applying the lattice Boltzmann method and Bayesian optimizer

Abstract

In this study, we propose a computational methodology to optimize the microstructure of the carbon felt electrode employed in redox flow batteries. Our optimization objective is to maximize the electrolyte utilization rate in a quantitive manner, which is affected by the fibrous electrode's microstructure. By combining stochastic generation of electrode microstructures, the digital compression of the electrodes, the Lattice Boltzmann Method, and a Bayesian optimization approach, we established our computational workflow that predicts an optimized set of parameters for electrode design. The optimization algorithm assesses electrode microstructure properties, including the specific surface area, the hydraulic permeability, and the practical reactive volume. The optimization results demonstrate that a high compression ratio with thick aligned fibers favors better electrode performance. For a highly compressed felt electrode, the pore size generally decreases while a small amount of large-sized pores remain in the microstructure, facilitating the convection in the electrolyte flow. The obtained properties from our optimal microstructure and from an unfavorable microstructure are injected into a homogeneous model to simulate the electrochemical performance during charge and discharge, in which the optimal case demonstrates a better reversible capacity than the not optimized one.