Commercialization of lithium-air batteries faces many challenges, such as electrolyte decomposition, short cycle life, low energy and power density, etc. However, commercialization of Li-O2 batteries for aeronautics is much more challenging due to additional safety constraints on cyclability and performance (high specific power and specific energy). For this presentation, we will discuss inter-related aspects of physics-based modeling of a pack: cell and battery model calibration; and parameter estimation using statistical methods. In addition, we will evaluate and present optimal battery designs for high discharge current density, high discharge time, and low battery mass using simulation-based optimization. The Finite Element Model (FEM) used to simulate a Li-O2 cell is based on the work of Bevara [1]. The different aspects of the model are based on: porous electrode theory and concentrated electrolyte theory; quantum tunneling model for the resistance of conformal layer of discharge product (Li2O2) [1]; Butler-Volmer kinetics for electrochemical reaction; Fick’s diffusion for oxygen transport; and an oxygen dissolution model is applied at the air/electrolyte interface [2]. The electrolyte properties such as ion conductivity, ion diffusion, oxygen diffusion, and mass density of the electrolyte were taken from Molecular Dynamics (MD) simulations [3]; while the other model parameters, which includes mass of cell components, were calibrated to match experiments at high discharge current densities. The cell mass includes the anode, cathode, separator, electrolyte, and other components (such as current collector). This calibrated model is used to perform parametric studies on cathode thickness, porosity, tortuosity, carbon particle size, electrolyte transport and material properties, partial pressure of oxygen, discharge time, and discharge current density to study optimal designs for high specific power and energy. Traditionally, pack (several cells are connected in series and/or parallel) simulations are performed using equivalent circuit, analytical expressions, or empirical models. Here, we develop a multi-physics model to include cell level variations in pack by considering factors such as difference in pressure drop, temperature variation, and variation in manufacturing. Using “accurate” cell power and mass estimates from the multi-physics model, optimal pack design solutions will be explored and presented at the meeting. Continuum model parameters are usually taken from experiments, from first principle simulations, or are assumed. They include errors such as model error, which arise from using a simplified model, from neglecting unknown processes or interactions, or from using wrong assumptions. Bayesian inversion techniques [4] will be used to find probability distributions in parameter values such as reaction rate, reaction active area, and Li2O2 resistance and the resulting error in the predicted values of specific power will be presented.
Read full abstract