Cycling lithium-ion batteries causes capacity fade, but also changes the shape of the open-circuit voltage (OCV) curve, due to loss of active material (LAM) and loss of lithium inventory (LLI). To model this change, we recently proposed a novel empirical calendar aging model that is parameterized on component states of health (s) instead of capacity fade only.
Mechanistic cycle aging model for the open-circuit voltage curve of lithium-ion batteries
Mechanistic cycle aging model for the open-circuit voltage curve of lithium-ion batteries
Authors: Alexander Karger, Julius Schmitt, Cedric Kirst, Jan Singer, Leo Wildfeuer, Andreas Jossen
Highlights
- Empirical algorithm for degradation rates enables first mechanistic cycle aging model
- Degradation rates for all components mainly driven by DOD, SOC and temperature
- Modeling aged OCV curves reduces voltage error by factor 8, compared to no update
- Predicted capacity fade values exhibit an average error of 1.04% on validation data
- Hidden degradation of electrodes does not explain onset of non-linear capacity fade
Cycling lithium-ion batteries causes capacity fade, but also changes the shape of the open-circuit voltage (OCV) curve, due to loss of active material (LAM) and loss of lithium inventory (LLI). To model this change, we recently proposed a novel empirical calendar aging model that is parameterized on component states of health (s) instead of capacity fade only.
In this work, we present a mechanistic aging model for cycle aging, allowing prediction of capacity fade, OCV curve change and component degradation. The model is parameterized on cycling data of 59 commercial lithium-ion batteries with NCA cathode and silicon–graphite anode, which were cycled for 2500 equivalent full cycles under varying conditions. We propose a stepwise approach to identify the most relevant stress parameters causing LLI and LAM, where we also separate between loss of accessible graphite and silicon in the blend anode. Stress parameter dependence is modeled with linear combinations of exponential functions and the model predicts capacity fade with mean absolute error(MAE).
For all test conditions, LLI is the dominating degradation mode and loss of accessible graphite is negligible. Reconstructed OCV curves reduce the median voltage MAE by a factor of 8, compared to not updating the OCV.
Access the paper here.
Mastering EV Battery Aging: Prevent Fires and Lithium Plating
Join our webinar on avoiding battery fires and non-linear aging with the new anode-potential model for faster, safer charging and improved battery lifespan.
Watch recording