Surrogate-assisted multi-objective optimization of hybrid thermoelectric–liquid battery thermal management systems

Battery thermal management is critical for ensuring the safety, performance, and longevity of lithium-ion batteries in electric vehicles. This study presents a hybrid battery thermal management system (BTMS) integrating thermoelectric coolers (TECs) with serpentine mini-channel cold plates for cylindrical lithium-ion battery packs. A system-level multi-objective optimization framework combining a genetic algorithm with an artificial neural network surrogate model is developed to simultaneously minimize maximum cell temperature, temperature difference, total power consumption, and system mass. Three representative designs were identified from the resulting Pareto front, and the knee-point design was selected for detailed evaluation, achieving a maximum temperature of 28.88 °C, a temperature difference of 2.04 °C, a power consumption of 64.27 W, and a system mass of 2.42 kg, representing reductions of 10.45 %, 14.51 %, 15.98 %, and 1.82 %, respectively, compared to a baseline design. Independent CFD re-evaluation of several Pareto-optimal solutions, including the selected design, confirms the surrogate model’s predictive accuracy, and sensitivity analysis identifies TEC input current as the strongest driver of thermal performance. Under both the NEDC and the WLTC, the optimized BTMS maintains stable thermal performance, with maximum temperature and temperature difference increasing by only 2.45 °C/1.25 °C and 1.58 °C/0.99 °C, respectively. The hybrid configuration also reaches a safe operating temperature approximately 53 % faster than the baseline following elevated initial conditions and remains robust under elevated coolant inlet temperatures (up to 45 °C) and high discharge rates (up to 2C). The proposed framework contributes to improving the thermal management of battery energy storage systems, with practical implications for electric vehicle range and battery longevity.