Optimization of nanofluid heat transfer in a microchannel heat sink with multiple synthetic jets based on CFD-DPM and MLA

Abstract

This study uses a combination of computational fluid dynamics consisting of the discrete phase model (CFD-DPM) coupled with support vector regression-particle swarm optimization (SVR-PSO) techniques to maximize the cooling performance of Al2O3water nanofluids in a microchannel heat sink (MCHS) with multiple synthetic jets (SJs). First, a CFD parametric study is carried out to understand the effect of influential parameters, including orifice spacing, particle volume fraction, orifice height, diaphragm length, phase actuation of jets, frequency and amplitude of oscillating diaphragms. A hybrid SVR-PSO algorithm is then adopted to predict the optimum values of the influential parameters for the maximum homogeneous heat transfer. The results predicted by the machine learning algorithm (MLA) are also compared to CFD findings. 0.15% and 3.5% deviations are found between predicted and actual values for the minimum average temperature and temperature uniformity, respectively. The parametric study reveals that heat transfer increases when the two jets are placed apart from each other. Based on the parametric study, the average temperature drops by 4.3 K as the membrane length increases from 0.95 mm to 1.96 mm. The highest heat transfer is obtained at a particle volume fraction of 5% in both jet arrangements. It is found that increasing the amplitude and frequency of the membranes results in better cooling performance. The results also confirm that larger orifice heights allow the creation of longer and stronger flows in the orifices that propagate the furthest in the microchannel. Overall, in-phase jet configurations show more uniform and lower temperatures (e.g., better heat transfer) at higher particle concentrations compared to the 180° out-of-phase jet arrangements.