Electro-thermal-aging modeling of Li-ion batteries with active-material loss

Most aging models for lithium-ion batteries (LIBs) focus on the loss of lithium inventory (LLI) while neglecting the loss of active material (LAM), which limits their ability to capture long-term capacity fade and heat generation. To address this gap, a coupled electro-thermal-aging (ETA) model is developed by explicitly incorporating both LLI and LAM, along with an analysis of the associated heat-generation behavior. The modeling framework begins with the construction of an LLI-only baseline model. A quantitative LAM indicator is then derived through differential voltage analysis (DVA), which exhibits a stable power-law relationship with cumulative charge throughput. Because experimental data for LAM parameter mapping are sparse, the training dataset is expanded through physics-informed augmentation, and a hybrid multilayer perceptron-Gaussian process regression approach is used to obtain accurate and continuous parameter estimates. The complete ETA model is validated using experimental measurements and COMSOL simulations, followed by an evaluation of the evolution of individual heat-generation components over extended cycling. Relative to the baseline, the ETA model reduces voltage and capacity prediction errors by 40.63% and 21.96%, respectively. The thermal analysis reveals a nonlinear increase in total heat generation with aging, driven predominantly by activation polarization heat, with ohmic heat increasing more moderately and reversible heat remaining nearly unchanged. Overall, the ETA framework strengthens understanding of how multiple degradation pathways influence the electrochemical and thermal behaviors of LIBs, thereby informing the design of more effective thermal-management strategies across the battery life cycle.