Performance optimization of discrete sidewall features mini-channel heat sinks via laser powder bed fusion and machine learning

The ability to fabricate complex geometric configurations of the laser powder bed fusion technique offers a significant potential for fabricating a mini-channel heat sink with discretized features on the sidewall. Based on this advantage, a novel discretized sidewall features MCHS (DSF-MCHS) was proposed, and an optimal design with multi-objective and multi-parameter was carried out using a genetic algorithm under 20% design conditions. The optimized DSF-MCHS and the corresponding continuous sidewall features MCHS (CSF-MCHS) were fabricated using AlSi10Mg. The performance of pressure loss and temperature rise was evaluated through numerical simulations and experiments. The results show that the average temperature rise in DSF-MCHS increases by 27.2% and the pressure drop decreases by 64.9% compared to CSF-MCHS over the considered flow range, indicating a significant advantage in reducing the pressure drop for DSF-MCHS. In addition, a neural network model was trained based on the available dataset to predict the achievable pressure drop reductions under varying temperature rise increase conditions within the design space, guiding design optimization. The proposed DSF-MCHS offers significant potential for energy conservation and high efficiency for heat transfer applications.