Parametric modeling and thermal optimization of novel bionic heat sinks inspired by leaf venation

Nonuniform heat transfer problems in integrated circuits and turbine blade lead to parametric optimizations of various heat sinks. But the finite parameters of the current heat sinks have limited their performance. Based on the self-grown novel heat sinks in our previous study inspired by leaf venation, this study constructs 3 parametric models (Curve A, B and C) defined by 4, 19 and 20 parameters each to preserve the key characters of novel heat sinks. To increase the efficiency of the optimization with large scale of variables, this study proposes Grouped Genetic Algorithm (GGA) which divides the whole variables into several groups and conducts Standard Genetic Algorithm (SGA) in each subspace gradually. In this study, optimizations are conducted with SGA and GGA for 3 curves in a centrally peaked boundary condition to minimize the maximum wall temperature. 1D pipe-net and 2D heat conduction methods are employed for heat transfer calculation and the optimized results are analyzed numerically. As the results show, on one hand, GGA has a better convergence with the difference of the average and minimum value within 5 K for all the curves while that of SGA over 10 K for Curve B and C. On the other hand, the maximum temperature of the optima of GGA is 10.8, 23.6 and 15.8 K lower than that of SGA on average. Fundamentally, the advantage is attributed to the proper distribution of coolant heat transfer rate relative to the cooling demand derived from the thermal boundary. The optimal channel from GGA absorbs over 1700 W within 45 % of the radial range, which eliminates central hotspot. And its circumferential distribution is more uniform than that from SGA, which avoids local overheat in regions without channel extension. The optimized heat sinks in this study can provide a skeleton for the channel system design in nonuniform boundary conditions.