Multi-objective optimization of cooling plate with hexagonal channel design for thermal management of Li-ion battery module

An optimization workflow combining Latin hypercube sampling (LHS), surrogate models and multi-objective optimization is explored and applied to augment the effectiveness of a hexagonal mini-channelled cooling plate proposed previously in mitigating the overheating issues of 20Ah pouch batteries. Three design parameters are selected and deliberately randomized to explore a range of geometric configurations that might not be readily apparent through traditional intuition. Alongside, two thermo-fluidic parameters are varied with an optimization objective to minimize maximum battery temperature, pressure drop and enhance heat transfer performance simultaneously. Initially, 160 sample points are generated using LHS, followed by simulations conducted using COMSOL Multiphysics for a fixed discharge rate of 3C and an ambient temperature of 35 °C. Then, Multi linear regression, Response surface approximation, Support vector machine, and Kriging model are considered in search of the best surrogate. Results highlighted that the Kriging model could make predictions more accurately and is used further. Later, a Pareto analysis is performed using non-dominated sorting genetic algorithm II to generate 100 optimal solutions influencing the three conflicting objectives. Finally, the best compromise solutions are obtained using a K-means clustering algorithm and are numerically validated. The case with a side length of 18.99 mm, a channel width of 4.99 mm, a branching angle of 91.51° and a coolant temperature of 25 °C is ideal. This aided in reducing the pressure drop by 62.32% and enhancing heat transfer coefficient by 64.41%, with a minimal change in maximum temperature compared to that of the initial design under the same mass flow rate limit of 0.003 kg s−1. In addition, the SHapley Additive exPlanations technique is employed to unravel the design and operating variable's impact on the objective functions. Overall, the proposed framework could be a valuable contribution to optimizing the cooling channel design and potentially boosting the battery lifespan.