Shape optimization of inlet header of micro-channel heat sink using surrogate model combined with genetic algorithm

Abstract

A lot of work has documented the significance of fluid-flow uniformity on the thermal-hydraulic performance of the micro-channel heat sink. The purpose of this work is to optimize the shape of inlet/outlet headers of a micro-channel heat sink to improve the flow distribution characteristics using the back propagation neural networks combined with the genetic algorithm as the surrogate model. The slanted edge of the inlet header is defined as the quadratic parabola instead of straight line. Meanwhile, the shape of the parabola is optimally designed for different flow rates. The 40 training sample points and six testing sample points on different geometry structures of inlet header are designed by the Latin hypercube sampling method. The 3-D CFD calculation is performed for all models. The objective function is defined as the non-uniformity of the fluid-flow. It is found that the prediction of the genetic algorithm back propagation for the fluid-flow distribution is capable of obtaining objective function values within the designed space. Through the optimizations, the non-uniformity of the optimal inlet header structure can be reduced by 52.43% to 33.17% for the inlet velocity of 0.05 m/s to 0.1 m/s, respectively, compared to that of the original design. The results demonstrate that the parabolic treatment for the slanted edge of the inlet header as well as structural optimization can greatly improve the flow uniformity of the micro-channel heat sink.