Multi-objective design optimization of pin-fin cold plates for electric vehicle battery packs using convolutional neural networks and genetic algorithms

Lithium-ion batteries generate heat while charging and discharging, requiring battery thermal management systems (BTMS) to mitigate adverse thermal effects impacting battery performance, safety, and longevity. As future electric vehicles (EVs) demand battery packs of higher energy densities that charge faster, more advanced BTMS are becoming increasingly critical. This work contributes to surrogate-model-based design optimization methodologies that are accelerating the design process of BTMS and their integrated liquid-cooled cold plates. This work proposes a novel surrogate-model-based design optimization approach to enable rapid design iterations of cold plates and satisfy ever-stringent EV thermal management needs. The approach is demonstrated through a case study addressing the optimal channel geometry of a pin-fin cold plate for a modular pack design used in the battery pack of Ford’s Mach-E EV model. This work’s approach produces optimal cold plate channel geometries based on battery-module-specific design features, including battery cell heat generation profiles, packaging materials, and heat spreading characteristics. Latin Hypercube sampling and numerical three-dimensional conjugate heat transfer models are used to construct a training database for deep encoder-decoder hierarchical (DeepEDH) convolution neural network (CNN) models. These trained DeepEDH-CNN models replace the numerical simulations during a multi-objective optimization with the Non-dominated Sorting Genetic Algorithm II (NSGA-II). The optimal pin-fin positions and radii yield a cold plate design that reduces maximum temperatures, temperature gradients, and cooling power consumption by 4.87 K, 5.1 K (22.2 %), and 7.93 Pa (9.0 %) respectively. This work demonstrates the capabilities of the proposed approach as a generalized methodology for BTMS design optimization, holding the potential to accelerate the thermal performance enhancements demanded by EV fast charging and next-generation battery packs.