Algorithm

Engineering

Heat transfer

Materials science

Cartography

Constructal law

Thermal resistance

Computer science

Porosity

Mechanical engineering

Optimal design

Microchannel

Thermodynamics

Sink (geography)

Physics

Nanotechnology

Genetic algorithm

Geography

Porous medium

Composite material

Heat sink

Mechanics

Machine learning

Multi-objective optimization

International Journal of Heat and Mass Transfer

Thermal resistance and pumping power are two key metrics to evaluate the performance of microchannel heat sinks. To reach their better performance, porous medium can serve as an appropriate alternative to conventional solid rib due to its unique geometric architecture. In this work, the performance of a selected porous-ribs microchannel heat sink proposed previously is significantly improved through a 3D fluid-solid conjugated model coupled with a multi-objective and multi-parameter genetic algorithm optimization method. The optimal design and operation parameters are achieved under a constant volumetric flow rate. The Pareto-optimal front presents that the MCHS can reach a minimum thermal resistance of 0.06306 K/W with pumping power of 0.38317 W, and a minimum pumping power of 0.00171 W with thermal resistance of 0.37755 K/W. The optimization finds the optimal thermal resistance and pumping power are 0.09348 K/W and 0.02888 W, respectively. It shows that not only the cooling performance is significantly improved by 14.06%, but also the pumping power is considerably reduced by 16.40% when compared with the original design. Furthermore, the underlying physics of the effect of multi-parameter on the performance is analyzed. The results reveal that the compromise mediation between the pumping power of the upper channel and the cooling performance of the lower channel is responsible for reaching the optimal performance. Based on this mechanism, the design criteria to reach performance of preference are proposed.

Optimization of a double-layered microchannel heat sink with semi-porous-ribs by multi-objective genetic algorithm