Long-Term Calendar Aging Across Commercial Lithium-Ion Cell Chemistries—Modeling and Early Prediction

Lithium-ion batteries are widely used in electric vehicles, stationary grid systems, and various consumer electronics today. In these cases, batteries remain idle for extended periods and are only in operation for a relatively small portion of their lifetime. Batteries degrade while at rest due to calendar aging induced mechanisms, and quantifying the effect of calendar aging is crucial for accurate battery capacity loss predictions. In this study, we formulate, validate and compare three different models to predict lithium-ion battery capacity fade from a long-term calendar aging dataset [V. N. Lam, X. Cui, F. Stroebl, M. Uppaluri, S. Onori, and W. C. Chueh, Joule, 9, 146 (2025)], namely semi-empirical, symbolic regression and data-driven models, are compared and validated using this dataset. We investigate each model’s ability to interpolate and extrapolate aging behavior across different storage conditions and cell types testing their transferability. We observe that the three models accurately predict the end-of-life (EOL) of a cell. However, the data-driven model demonstrates a superior ability to predict the capacity fade trajectory for an unseen cell. Moreover, the data-driven model exhibits reasonable accuracy when predicting the EOL and capacity trajectories across various cell-types.