Optimal design of finned surface structures to reduce thermal contact resistance

Solid-solid contact heat transfer widely exists in the industrial field. In this work, we propose utilizing finned surface structures to reduce solid-solid thermal contact resistance thanks to the increased contact area. To accurately predict the thermal resistances of finned surface structures, we employ a physics-augmented machine learning (PAML) approach that combines machine learning (ML) with the thermal circuit model (TCM). The fin sizes are further optimized to minimize the thermal resistance by coupling the deduced correlations with the genetic algorithm (GA) and particle swarm optimization (PSO). For the rectangular fin, the theoretical solution of the optimal fin height is derived. Given the remaining geometrical parameters, we reveal the optimal fin heights as functions of thermal conductivity and thermal contact conductance. Furthermore, the multi-objective genetic algorithm (MOGA) is employed to optimize both the fin height and thermal resistance. Then the optimal structure is selected by TOPSIS (Technique for Order Preference by Similarity to an Ideal Solution) from the Pareto front. Compared with the planar contact, the finned surface structures exhibit smaller thermal resistances within a certain range of thermal contact conductance.