Uncertainty quantification in energy and thermal management of immersion-cooled battery safety systems in electric vehicles using polynomial chaos expansion

High-fidelity Multi-physics models offer detailed insights into thermal behaviors and conjugate heat transfer phenomena within immersion-cooled battery systems. However, their computational complexity significantly limits direct analysis of parametric uncertainties, necessitating an efficient probabilistic approach for practical safety assessments. This study develops comprehensive uncertainty quantification framework using Polynomial Chaos Expansion for immersion-cooled battery systems. A validated surrogate model captures intricate electrochemical-thermal interactions within lithium-ion battery, enabling optimized handling of uncertain electrochemical parameters. Subsequently, second surrogate model investigates uncertainties in thermal interactions during immersion cooling process, ensuring robust modeling of heat transfer dynamics. Adaptive Sparse Polynomial Chaos Expansion based on Least Angle Regression is employed to address computational demands imposed by inherent uncertainties in immersion cooling processes. Sensitivity analysis, utilizing Sobol indices, identifies critical parameters that significantly influence temperature uniformity. An uncertainty-aware optimization framework is established, and Kernel Density Estimation results show a 100 % probability of maintaining a temperature difference below critical 5 K threshold when operating at an optimal coolant inlet temperature of 20 °C and a flow rate of 2 L per minute. This research substantially advances theoretical and practical foundations for developing robust, optimized battery thermal management strategies, promoting enhanced reliability, safety, and prolonged battery life in electric vehicles.