Abstract
To cope with the increasing heat generation resulting from increased power in micro-scale devices, an optimum design of a tree-like microchannel heat sink is suggested utilizing Artificial Neural Networks (ANN) and Response Surface Methodology (RSM). Based on the numerical finite volume model, both ANN and RSM models were trained, and their performances were assessed using four distinct perfor-mance criteria. The results showed that both of the models are reliable for the purpose of this study, showing a significantly high adjusted R2 for both performance control variables pressure drop (Delta P) and Nusselt number (Nu). The adjusted R2 values observed for the RSM were 0.987 and 0.997, respectively, for the Nu and Delta P. These values were calculated as high as 0.997 for Nu prediction and 0.997 for Delta P prediction using the ANN. Finally, utilizing the calibrated RSM and ANN models, three optimum design configurations for three different purposes given the priority requirements of the heat sink imple-mentations were suggested. The minimum Delta P configuration is suitable for the longevity of the heat sink, and the maximum Nu configuration is suitable for applications where heat transfer performance is more critical. For a balanced optimum design, the efficiency index was maximized.