Joint inference of diffusivity and additional electrochemical parameters in extended ICM/SPM frameworks

Develop an efficient and reliable inference methodology that, within physics-based battery models used by the Inference from a Consistent Model (ICM) approach (such as the single particle model), simultaneously estimates the concentration-dependent solid-state lithium diffusivity D(c) and additional parameters representing electrolyte transport limitations, reaction overpotentials, and Ohmic and charge-transfer resistances from voltage–current data. Ascertain computational strategies—potentially including ultra-fast surrogate forward solvers—to make this contemporaneous multi-parameter inference tractable and robust.

Background

The paper introduces the ICM methodology, which infers the concentration-dependent diffusivity by fitting a physically consistent single particle diffusion model directly to measured voltage–current data. While successful under assumptions where overpotentials and Ohmic drops are negligible, real electrodes often exhibit electrolyte transport limitations, non-negligible reaction overpotentials, and resistive losses.

The authors note that extending ICM to incorporate these additional phenomena increases model complexity and parameter count. They state that, unless these extra parameters are independently characterised, one must pose and solve a more complex inference problem that jointly identifies diffusivity and the added parameters. They explicitly remark that this multi-parameter, contemporaneous inference remains open and suggest that ultra-fast surrogate models may be needed to make the forward solutions fast enough for practical optimisation.