Review and Performance Comparison of Mechanical-Chemical Degradation Models for Lithium-Ion Batteries

TLDR

Lithium‑ion battery energy capacity declines due to irreversible degradation mechanisms, yet a comprehensive comparison of existing degradation models is lacking, hindering the selection of appropriate approaches. The study aims to compare various literature degradation models within a unified single‑particle framework. The authors implement these models in a single‑particle simulation to evaluate their behavior. Many models can be fitted to small data sets, but interactions among them can accelerate degradation and alter trends; operating conditions influence model performance, and only a large data set combined with multiple models can capture diverse degradation behaviors.

Abstract

The maximum energy that lithium-ion batteries can store decreases as they are used because of various irreversible degradation mechanisms. Many models of degradation have been proposed in the literature, sometimes with a small experimental data set for validation. However, a comprehensive comparison between different model predictions is lacking, making it difficult to select modelling approaches which can explain the degradation trends actually observed from data. Here, various degradation models from literature are implemented within a single particle model framework and their behavior is compared. It is shown that many different models can be fitted to a small experimental data set. The interactions between different models are simulated, showing how some of the models accelerate degradation in other models, altering the overall degradation trend. The effects of operating conditions on the various degradation models is simulated. This identifies which models are enhanced by which operating conditions and might therefore explain specific degradation trends observed in data. Finally, it is shown how a combination of different models is needed to capture different degradation trends observed in a large experimental data set. Vice versa, only a large data set enables to properly select the models which best explain the observed degradation.