ANN and CFD-DPM Modeling of Alumina-Water Nanofluid Heat Transfer in a Double Synthetic Jet Microchannel

Abstract

AbstractIn this paper, integrating different cooling techniques into a microchannel heat sink (MCHS) is evaluated to enhance the heat transfer rate. A parametric study is conducted using the computational fluid dynamics simulation consisting of the discrete phase model (CFD-DPM) to understand the effect of operational and geometrical parameters on cooling performance. The CFD-DPM simulations for varying different influential parameters are conducted to obtain the local temperature over the silicon wafer wall, creating a large set of samples. The artificial neural network (ANN) method is then employed to discover an accurate model to predict the local temperature. The ANN model includes two hidden layers and 24 neurons in each layer, showing a precise estimation of temperature with an overall mean square error (MSE) value and correlation coefficient (R) of 6.974810\({}^{-7}\) and 0.9952, respectively. The temperature distributions along the silicon wall predicted by both CFD-DPM and ANN models verify that smaller sizes of alumina nanoparticles improve heat transfer remarkably compared to larger particles.KeywordsComputational fluid dynamics (CFD)Artificial neural network (ANN)Discrete phase model (DPM)Nanofluid heat transferSynthetic jet (SJ)