Sequential multi-objective Bayesian optimization of air-cooled battery thermal management system with spoiler integration

Effective battery thermal management is crucial for the performance and safety of lithium-ion batteries. This study introduces a novel three-dimensional battery thermal management system (BTMS) structure incorporating a Be<acute accent>zier surface designed spoiler, requiring multiple variables for its definition and optimization. The study further proposes data efficient machine learning (ML) driven optimization strategy, namely, sequential multi- objective Bayesian optimization (SMBO). The BTMS consists of a battery module with eight prismatic cells, utilizing a forced air-cooling system with a single inlet and outlet. The SMBO approach sequentially optimizes battery cell spacing and spoiler geometry to minimize the average cell temperature and temperature differences between the cells, while reducing energy consumption, which is quantified by pressure drop. The optimized designs demonstrate superior cooling performance, achieving average temperature of 312.76 K and a maximum temperature difference of 4.04 K among battery cells, representing reductions of 3.72 K and 14.28 K (77.95 %), respectively, compared to the baseline conventional design. Furthermore, when compared to designs optimized solely for cell spacing with similar pressure drop values, the SMBO approach achieves an average temperature of 314.86 K and a maximum temperature difference 5.36 K, representing reductions of 0.76 K and 2.36 K (30.57 %), respectively, while maintaining a pressure drop of 15.88 Pa, an increase of only 0.1 Pa (0.63 %). These results are achieved while maintaining energy efficiency, highlighting the potential of aerodynamic interventions and advanced optimization techniques for developing high-performance BTMS solutions for electric vehicles.