A vital aspect of Ultra-Fast Charging (UFC) Li-Ion battery packs is their thermal management system, which directly influences safety, performance, and cell longevity. Immersion cooling technology offers superior effectiveness compared to indirect cold plate cooling, as it allows for faster heat dissipation and has the potential to significantly mitigate thermal runaway propagation, enhance overall pack performance, and extend cell life.
To achieve faster design optimization and deeper insights, high-fidelity Multiphysics-Multiscale simulations are essential. In this study, Equivalent Circuit Model (ECM) based electro-thermally coupled CFD simulations are utilized to optimize an innovative busbar design that facilitates the removal of individual cells. Additionally, high-fidelity 3D transient flow-thermal simulations have been employed to optimize coolant flow direction, inlet positions, cell spacing, and separator design, ensuring efficient flow distribution within the module. While these high-fidelity CFD models accurately represent flow and thermal behavior, their computational intensity often limits the speed of design optimizations.
To address this challenge, the study focuses on developing Reduced Order Models (ROM) derived from high-fidelity CFD simulations, aimed at improving the prediction capabilities of Battery Management Systems (BMS) through real-time simulations. The methodology for creating a Linear Parameter Varying (LPV) ROM, along with multiple Linear Time Invariant (LTI) matrices for expedited parametric studies, is explored in detail. The ROM fully integrates electro-thermal aspects of immersion cooling systems, where dielectric liquid flows directly along the axial direction of the cells. Training data for the LPV ROM is generated by running transient step response thermal simulations on converged steady-state flow solutions at various flow rates. An experimental module setup, comprising 144 cells immersed in dielectric fluid, is prepared to validate the model. The validation results confirm the ROM’s accuracy and robustness, achieving a tenfold reduction in computational time with minimal loss in solution accuracy.