Simulation of subcooled flow boiling with an SVR based interphase mass transfer model

Numerical simulation has been becoming more and more important for the investigation of subcooled flow boiling in practical engineering. A precise description of interphase mass transfer between the liquid phase and the vapor phase is quite crucial for simulations regarding phase change. In this paper, Lee's interphase mass transfer model was improved to make the simulation of subcooled flow boiling more accurate and robust. With the experimental data collected by Liu et al., the trial and error method was adopted to develop a support vector regression (SVR) based prediction model of time relaxation parameter. The proposed model could successfully predict the time relaxation parameter with a MSE of 0.56% on the training data and 2.0% on the test data. Numerical validation suggests that current SVR based prediction model could accurately give the value of time relaxation parameter in different flow conditions. Besides, simulations performed with the improved interphase mass transfer model indicate that the inlet velocity has much more powerful influence on transferred heat flux than the inlet liquid subcooling and the system pressure in subcooled flow boiling.