Anaysis of the validity of P2D models for solid-state batteries in a large parameter range

TL;DR

This paper evaluates the accuracy and limitations of pseudo two-dimensional (P2D) models versus 3D models for solid-state batteries across a broad parameter space, focusing on sensitivity analysis and model comparison.

Contribution

It provides a comprehensive comparison of geometric resolutions in electro-chemo-mechanical models, highlighting their advantages, drawbacks, and regions of significant deviation.

Findings

Global sensitivity varies with parameters, affecting model accuracy.

Significant deviations between models occur in specific parameter regions.

Potential for combined modeling approaches to improve efficiency and reliability.

Abstract

Simulation models are nowadays indispensable to efficiently assess or optimize novel battery cell concepts during the development process. Electro-chemo-mechano models are widely used to investigate solid-state batteries during cycling and allow the prediction of the dependence of design parameters like material properties, geometric properties, or operating conditions on output quantities like the state of charge. One possibility of classification of these physics-based models is their level of geometric resolution, including three-dimensionally resolved models and geometrically homogenized models, known as Doyle-Fuller-Newman or pseudo two-dimensional models. Within this study, the advantages and drawbacks of these two types of models are identified within a wide range of the design parameter values. Therefore, the sensitivity of an output quantity of the models on one or a combination of parameters is compared. In particular, the global sensitivity, i.e., the sensitivity in a wide range of parameter values, is computed by using the Sobol indices as a measure. Furthermore, the local sensitivity of the difference in the output quantities of both models is evaluated to identify regions of parameter values in which they contain significant deviations. Finally, remarks on the potential interplay between both models to obtain fast and reliable results are given.