Laboratory-based battery ageing studies rely primarily on constant-current cycle loads to age the battery at defined conditions. In the next step, the battery degradation can be modelled by correlating the resulting degradation to the varied constant-current cycle conditions, such as temperature, charging and discharging current, cycle depth, and soc range. However, the load profiles of batteries in the application are far from constant current. The applications usually result in transient and dynamic current profiles, proceeding from the application’s dynamic power demand and usually contain multiple pauses and interruptions or recuperation phases. The dynamic profiles are simplified as constant-current cycles to study the effect in a laboratory ageing study.
When applying an empiric ageing model on field data, the dynamic and transient current, temperature and state-of-charge profile need to be condensed into a collective of constant-current cycles, for which the model can calculate the individual degradation. This reduction towards a cycle collective from time series can be made with the rainflow algorithm, which finds overlaying and included cycles beside interruptions and pauses. Whether or not a cycle is interrupted by another one or simply a pause is irrelevant to finding the overlaying large cycle, usually dominating the overall degradation.
This step of simplification, from transient and dynamic profiles towards constant-current profiles and when using the rainflow algorithm for applying an ageing model, is questioned in this work. We have developed two studies which investigate the effect of a few longer (5 to 15-minute) interruptions in the discharging and charging half cycle and multiple shorter interruptions (1mHz to 1Hz) in the discharging half cycle. For the first study, a reference measurement was done without interruptions. The second study used a reference measurement with the same mean current as the interrupted current profiles. This approach enables the analysis of the impact of averaging dynamic current profiles on the degradation.
Compared to no pauses, we have concluded that more prolonged interruptions (5 to 15 minutes) in the discharge and charge cycle reduce the degradation. Additionally, longer pauses (15 min) result in a higher reduction of degradation than shorter pauses (5 min). Regarding the shorter interruptions (1mHz to 1Hz) of the discharging half cycle, no significant influence of the interruption length or frequency has been found. Besides a high cell-to-cell deviation (n=2 cells per test point), probably caused by production deviation between cells and charges, averaging the dynamic current and applying a constant current agrees with the interrupted profiles.
With the results of the two ageing studies, the reduced ageing in study 1 is expected to result from the interruption of the charging half cycle. In contrast, pauses in the discharging half cycle do not matter and can be depicted by an averaged discharging current. The interruption in the charging half cycle is applied at a relatively high charging current of 0.75C, which might trigger at least partial lithium-plating due to inhomogeneities. Interruptions are believed to reduce the inhomogeneities, resulting in lower degradation.