Numerical investigation and data-driven prediction of flow boiling heat transfer in manifold microchannels with slopes

Abstract

The manifold microchannels (MMC) with high heat transfer coefficient and low pressure drop has received much attention. Flow boiling heat transfer inside MMC is promising for dissipating heat of high-power devices. In this study, flow boiling heat transfer inside MMC is further numerically studied with slopes added in the upstream and downstream corners of MMC. The results show that downstream slope leads to lower local temperature and thermal resistance by removing the upstream vapor more easily, and upstream slope leads to higher heat flux corresponding to the turning point in the flowing boiling curve by restraining the local high temperature at upstream corner. The combination of the upstream and downstream slopes can reduce the thermal resistance by 5.45% and increase the heat flux related to the turning point by 14.28% compared to the MMC without slopes. A new flow boiling heat transfer correlation considering the slopes is developed with prediction accuracy of mean absolute deviation (MAD) of about 0.05%. An artificial neural network (ANN) is further built to predict the vapor saturation inside the MMC based on the outside temperature at the MMC bottom surface, which presents high accuracy of 97.03%.