Publications

Predicting the state of health (SoH) of Lithium-ion battery cells is essential for optimizing their utilization. Although electrochemical degradation models and data-driven approaches are promising for SoH estimation, they often rely on experiments with repeated idealized cycles and extended relaxation periods, leading to non-robust models for dynamic real-world conditions.

This study presents a novel investigation on how different cycling sequences (ageing history) influence future degradation rates and SoH estimation. In a unique experimental design, four identical automotive-grade NMC pouch cells were initially subjected to different accelerated cycling conditions, followed by common protocols until the degradation curve passed its knee point.

A detailed analysis was conducted on capacity evolution, swelling and apparent capacity fade. Results indicate that ageing history must be considered in SoH modelling, to achieve robustness under dynamic conditions and improve system safety. Under some conditions, the knee point onset was delayed by at least 50%, while in others, it did not occur at all. Additionally, reversible short-term capacity fade is strongly dependent on the cell’s state, with fresh cells exhibiting a drop of approximately 15% which gradually decreases to around 1% by the end of the test campaign. 

These findings highlight key considerations for transitioning from laboratory-based to real-world applicable SoH models. Finally swelling shows strong correlation with SoH (R² = 0.97), and is therefore a valuable alternative SoH indicator, especially in online in-service estimation techniques where controlled capacity checks under repeatable conditions are not available.