Net‐based thermal‐fluid model and hybrid optimization of cooling channels

Abstract This study aims to propose a hybrid optimization method for cooling channels based on the improved net‐based thermal‐fluid model. The channel topology was optimized through the genetic algorithm (GA), whereas the channel diameter and node position were optimized using the moving asymptote method (MMA). The linear combination of the average temperature and pressure drop was defined as the objective function. Three‐dimensional computational fluid dynamics (CFD) simulations were performed to verify the physical fields of the optimized results. The optimized results indicated that, as the weighting factor ( w ) for the average temperature increased, the flow distribution was more reasonable and uniform, effectively decreasing the average temperature at the cost of higher pressure drops. In comparison with the optimization entirely based on GA, the hybrid optimization could yield a similar and efficient cooling performance with lower time‐consuming. Moreover, the optimized results obtained under the suitable objective could achieve the better thermal performance at a lower pressure drop in comparison with the conventional serpentine‐shaped channel layout. Finally, the hybrid optimization method used in this article was compared with the conventional density approach, and the former could obtain the clear and complex channel distribution with greatly reduced computational time.