A multi-physics, fully liquid-cooled battery pack model development for winter-summer driving using a holistic reverse-engineering method

Abstract

Controlling battery temperature can reduce cell aging, internal resistance, and overheating, and improve pack performance. These require advanced battery thermal management systems (BTMS) for all-weather driving. A full liquid-cooled pack has hundreds to thousands of cells and coolant channels. Therefore, designing a full pack requires a thorough understanding of battery thermal, flow, and electrical responses under various driving and thermal conditions. This work presents a holistic reverse-engineering method to model and validate a production-based, liquid-cooled, 75-kWh lithium-ion battery pack, including its BTMS via multi-physics simulation. The model includes 4416 cells, 28 side-coolant lines, 784 coolant flow channels, and all plate bends. The channel geometries (width, height, bend angle, radius, and length) are optimized using a genetic algorithm. Firstly, a design-of-experiment is performed by changing the inlet coolant flow rate ($V_{cool}$ = 0–16 L/min) to measure steady-state and transient pressure drops. A sensitivity analysis of the channel geometries to the coolant flow characteristics is performed for the pack's flow model validation. A full battery-electric SUV equipped with the battery pack and dual e-motors was tested under a 60 km/h driving (winter test with ambient temperature $T_a$ = −10 °C) and repeated WLTC and (FTP75+HWFET) cycles in summer (25–40 °C). The pack performances were recorded under battery heating (initial $T_{a,i}$ < initial $T_{b,i}$) and cooling ($T_{a,i}$ > $T_{b,i}$) modes. The battery model is based on the 2RC equivalent-circuit model, calibrated against an electrochemical NCA/Gr-SiO_x cell model to accelerate simulation time. Using the optimized flow and cell models, the model accurately (over 90 % accuracy) predicts the pack's responses (voltage, state of charge, flow, temperature) under steady-state and dynamic conditions. The detailed approach to building a comprehensive pack model can serve as a guideline for future BTMS development.