Cooling performance optimization of liquid alloys GaIny in microchannel heat sinks based on back-propagation artificial neural network

Abstract

Microchannel heat sinks using liquid alloys GaIny as coolant may be a promising solution to the heat dissipation of micro-electro-mechanical systems with an increasing power density. However, the cooling performance depends on the mass fractions of the components, which makes sense of the optimization by finding out the appropriate mass fraction of Indium. To accomplish this task with insufficient data of thermophysical properties, we developed an artificial neural network based on which more refined data of thermal conductivity and specific heat of liquid GaIny were obtained. Meanwhile, the viscosities and densities of various GaIny were calculated using classical theoretical models. Then, we compared the cooling performances of microchannel heat sinks using these liquid GaIny based on an analytical model reported in the literature. The appropriate mass fractions of Indium that can harvest a best cooling performance were determined to be around 5.2%, 3.9% and 3.4% at three different inlet temperatures of 298K, 303K and 310K, respectively. Any of other cases can also be performed in the same way. This work demonstrates that the artificial neural network is a useful tool for thermal engineering applications.