Characteristics analysis and structure optimization of a hybrid micro-jet impingement/micro-channel heat sink

Abstract

To solve the heat dissipation problem of high-power electronics, a novel hybrid micro-jet impingement/micro-channel heat sink is proposed in this work. Different from the continuous and straight rectangular micro-channel, the inclined and interrupted channel walls (fins) are adopted to enhance the fluid disturbance inside the heat sink. To obtain the thermal–hydraulic performance of the hybrid heat sink, an overall 3D modeling and numerical simulations are conducted instead of local cell numerical simulation. In addition, to further improve the heat transfer and flow resistance performance, the effects of variation of structural parameters on heat transfer and flow resistance properties are investigated, including the fin inclination angle, fin length and fin height. The results of structural parametric analysis reveal that there are interactions between different structural parameters, but with different levels of significance for different evaluation indicators. Then, to obtain the optimal combinations of structural parameters for different applications, the multi-objective genetic algorithm coupled with artificial neural network (ANN) is utilized to search for the optimal value of each structural parameter. The heat transfer thermal resistance, pressure drop and mean absolute temperature deviation of the heated surface are utilized as the optimization objectives to evaluate the cooling and resistance characteristics. The optimization results obtained by three-objective optimization and two-objective optimization are analyzed and compared. Finally, seven optimal compromise solutions are selected from the Pareto front by using TOPSIS algorithm with different weight factor matrices. For the optimal compromise solutions, the maximum relative errors between the predicted values of the ANN and numerical simulation values are only 3%, which further demonstrates the accuracy of the artificial neural network model and the effectiveness of using multi-objective optimization to guide heat sink design.