Battery Degradation and Performance

Stanford 11Year Calendar Aging Study Researchers at Stanford published a dataset (available open-source) with 8 commercial cells from 5 manufacturers calendar aged over 13 years, showing significant time-varying nonlinearity, cell-to-cell variability, and cell-type dissimilarity. They highlighted the shortcomings of widely-accepted storage time and temperature dependencies on calendar aging, i.e., power-law exponents and the Arrhenius law,, respectively. They also compared 3 different models to predict the EOL and capacity trajectories of unseen cells, unseen temperatures & unseen cell types: semi-empirical SE, symbolic regression and data-driven DD) models.

For long storage times, capacity fade predictions require more sophisticated models than the Arrhenius law and power-law exponents. In unforeseen conditions, the SE model excels in predicting EOL, whereas for unseen cell types the DD model was effective in predicting the degradation trajectory.

Source: 1 X Cui et al., Journal of the Electrochemical Society 2025, 2 V. Lam et al., Joule (2025

Extracted from the PDF; charts and images may read imperfectly.