Computational Homogenization of Mechanical and Electro-Chemical Properties in Structural Battery Electrolytes

Doctoral thesis, 2024

In the quest for weight efficient energy storage solutions, the structural battery is an emerging technology under development. It is a multifunctional composite material that can carry mechanical loads, and simultaneously store and release energy. This is made possible due to carbon fibers' ability to act not only as structural reinforcement materials, but also as electrode components. While conventional batteries rely solely on liquid electrolyte to allow for ion transfer between the electrodes, structural batteries exploit the so-called structural battery electrolyte (SBE). The SBE consists of two continuous phases; a porous polymer skeleton and a liquid electrolyte. The role of the liquid electrolyte is to allow for ion transfer, while the porous polymer skeleton provides load-bearing functionality. In short, the structural battery consists of carbon fibers (acting as electrodes) embedded in an SBE (electrolyte/matrix).

The first part of the thesis studies the multifunctional performance of various SBE microstructures by performing virtual material testing on numerically generated artificial Representative Volume Elements (RVEs). In particular, the effective ionic conductivity is obtained by solving a diffusion equation with Fick's law, and the effective stiffness by assuming linear elasticity. As a direct extension of this framework, coupled diffusion and large deformation in the SBE is also considered; i.e., ionic transport in an SBE subjected to mechanical loads using finite strain theory. Here, the aim is to compute the deformation-dependent effective mobility.

The second part covers the development of a multi-scale modeling framework for electro-chemically coupled ion transport in SBEs. After establishing the governing equations, Variationally Consistent Homogenization (VCH) is employed to obtain a two-scale model. If the sub-scale RVE problem exhibits negligible transient effects for length scales relevant to the studied application, then the assumption of micro-stationarity can be introduced. This opens up for the possibility to devise a numerically efficient solution scheme for the macro-scale problem that is based on a priori upscaling of the effective response.

Lastly, Numerical Model Reduction (NMR) is exploited to enable solution of fully transient electro-chemically coupled two-scale problems without assuming micro-stationarity. The goal is to exploit NMR by training a surrogate model, via Proper Orthogonal Decomposition (POD), that replaces the RVE simulations. The surrogate model takes the form of a system of Ordinary Differential Equations (ODE). The final NMR framework leads to a computationally efficient solution scheme for solving fully transient two-scale problems.