Multi-objective topology optimization of a liquid-cooled heat sink for electric vehicle SiC power modules considering thermal and hydraulic performance

Silicon carbide (SiC) power modules are increasingly used in electric vehicle traction inverters because of their low losses, high switching frequencies and high-temperature capability. However, increasing power density concentrates semiconductor losses and creates localized hot spots, while higher coolant flow incurs greater pressure drop and pumping power. This work proposes a multi-objective design procedure for the liquid-cooled heat sink of a multi-chip SiC power module. First, module losses calculated using an electro-thermal model were assigned to a three-dimensional conjugate heat-transfer model, whose baseline pin-fin predictions agreed with steady-state measurements within 3%. Second, density-based topology optimization generated five channel configurations under different objective weights. Twenty weight–flow cases were used to train Kriging models of maximum module temperature and pumping power, both achieving leave-one-out cross-validated R 2 values of 0.998. Finally, the Non-dominated Sorting Genetic Algorithm II (NSGA-II) generated the Pareto front, and the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) selected a compromise at a thermal weight of 0.896 and 8.15 L/min. The continuous recommendation was mapped to the nearest generated topology, ω T : ω P = 0.9:0.1, for three-dimensional numerical confirmation. At 4915 W and 8.15 L/min, the selected topology produced maximum module and cold-plate temperatures of 171.6 and 108.8 °C, compared with 171.9 and 111.7 °C for the pin-fin baseline. Across 8–14 L/min, the module-temperature difference was 0.3–1.5 °C, comparable with the mesh sensitivity, whereas pressure drop and pumping power were approximately 61%–62% lower. The principal predicted benefit is therefore reduced hydraulic-energy demand with comparable module-temperature control.