A1318
Title: Global sensitivity analysis of a nonlinear battery replica model for design under uncertainty
Authors: Robertas Alzbutas - Kaunas University of Technology, Lithuanian Energy Institute (Lithuania) [presenting]
Edvinas Juozapaitis - Kaunas University of Technology (Lithuania)
Abstract: A nonlinear reduced electrochemical model is analyzed as a battery replica under parameter uncertainty. Constrained inverse parameter identification from current-voltage data is combined with Monte Carlo uncertainty propagation and global sensitivity analysis in charging and discharging regimes. Sobol first-order and total-order indices quantify the contributions of individual parameters and their interactions to the variability of voltage predictions, while rank-based sensitivity measures are evaluated for comparison. The results indicate that the positive-particle radius and the effective electrode area are the dominant contributors to output uncertainty, whereas electrode thickness and selected negative-electrode parameters have a moderate effect on the model response. The comparison shows that linear sensitivity measures are less informative for the considered nonlinear system, while rank-based and variance-based measures provide more stable and interpretable parameter rankings. The resulting workflow integrates numerical simulation, inverse modelling, and uncertainty quantification into a computational framework for parameter prioritization, calibration, and validation of nonlinear dynamical models for uncertainty-aware battery design and control.
