Multi-fidelity Electro-thermal Optimization of Multiport Converter employing SiC MOSFET and Indirect Liquid Cooling

This article presents a systematic multi-fidelity optimization framework applied to a 60kW Multiport Converter (MPC) for Electric Vehicle drivetrains. The proposed design methodology integrates the multidisciplinary design process through the integration of system optimization, multi-domain modeling, and existing Computer Aid Design (CAD) tools. At the first design stage, low-fidelity models of multi-objective functions (the static converter losses, the weight of inductor and the input ripple current) have been developed to find the optimal set of number of phases, switching frequency and inductor size. A set of optimal solutions in the form of a Pareto front is found by means of Non-dominated Sorting Genetic Algorithm (NSGA-II). At the second design stage, a medium-fidelity model coupling the dynamic electro-thermal behavior of a SiC MOSFET half-bridge module has been used to predict the instantaneous losses and junction temperature over a typical current load profile. Finally, a high fidelity model of the indirect liquid-cooling heatsink has been designed in ANSYS AIM to find the geometry parameters of the cold plate. The proposed design framework is an intermediate step to utilize better materials and select power components properly for a future hardware prototype.