Numerical investigation and deep learning-based prediction of heat transfer characteristics and bubble dynamics of subcooled flow boiling in a vertical tube

Subcooled flow boiling presents an enormous ability of heat transfer rate, which is extremely important in the heat-dissipating systems of many industrial applications, such as power plants and internal combustion engines. Using an Euler-Euler-based three-dimensional numerical simulation of subcooled flow boiling in a vertical tube, we investigated different heat transfer quantities (average and local heat transfer coefficient, average and local vapor volume fraction, average and local wall temperature) and bubble dynamics quantities (bubble departure diameter, bubble detachment frequency, bubble detachment waiting time, and nucleation site density) under various boundary conditions (pressure, subcooled temperature, mass flux, heat flux). Numerical results show that an increase in heat flux leads to the increase in all of the physical quantities of interest but the bubble detachment frequency. An entirely opposite behavior is observed when we change the mass flux and inlet subcooled temperature. Furthermore, a rise in pressure reduces all of the target quantities but the wall temperature and bubble detachment frequency. Since numerical simulation of such multiphase flow requires significant computational resources, we also present a deep learning approach, based on artificial neural networks (ANN), to predicting the physical quantities of interest. Prediction results demonstrate that the ANN model is capable of accurately predicting the target quantities with mean absolute errors less than 2.5% and R-squared more than 0.93.