Multiphysics models have shown potential to support the design of battery systems at beginning of life and control during operation conditions, but still show limitations on the lifetime prediction. Thus, despite significant advances in recent years, the determination and accurate parameterization of descriptive degradation models for battery systems remains an open problem which primarily relies on semi-empirical approaches.
A common approach on the estimation of cell degradation is the use of incremental capacity analysis which, generally, groups degradation into three modes: loss of active material (LAM), loss of lithium inventory (LLI) and internal resistance increase. Nonetheless, as several degradation mechanisms contribute simultaneously to more than one mode and the synergy between mechanisms is in some cases not fully understood, the identification of descriptive model parameters is challenging.
In this work, we present an inverse-modelling approach based on a pseudo-2D battery model to aid the parameterization of degradation models by combining different degradation metrics (state-of-health/energy, LLI, LAM…) from experimental measurements. We critically explore the main degradation models available in literature and support the study with sensitivity analysis to identify the critical parameters from a model perspective. Given the computation cost of cycle optimizations and sensitivity studies, available approaches for cycle-extrapolation are also explored. Moreover, the effect of the uncertainty in the manufacturing parameters (e.g., thickness and active material distribution) in cell performance is also considered to provide confidence intervals of the simulation results.