Modeling Calendar Aging of Lithium-Ion Batteries: Analysis of SEI-Growth Models and Their Experimentally Efficient Parameterization
Abstract We compare and evaluate five electrochemical solid-electrolyte interphase (SEI)-growth models from the literature, implemented in PyBaMM, using laboratory calendar-aging data of 39 3 Ah LiFePO4/C cells spanning 29 months. We first assess whether these models can predict aging at a single storage condition and then test whether they inherently capture the observed temperature and state of charge (SoC) dependence across a wide range of storage conditions. Four of the five models can be parameterized to fit 29 months of calendar aging at a single storage condition. An Arrhenius relation sufficiently describes SEI growth in a temperature range between 0°C and 40°C. However, none of the models inherently capture the SoC-dependency of calendar aging, indicating limitations in their treatment of anode-potential dependency. To address this, we propose a hybrid SEI-growth model combining electron-migration-limited and interstitial-diffusion-limited SEI-growth mechanisms. Our hybrid model successfully predicts calendar aging across a wide range of SoCs and temperatures and reduces the error by approximately 58% compared to the best-performing literature model. Finally, we analyze the experimental effort required for parameterization and show that, for this cell, approximately 280 days of calendar-aging data at two elevated temperatures and three SoC levels are sufficient.