Data-driven modeling for boiling heat transfer: using deep neural networks and high-fidelity simulation results
Authors:
Yang Liu,
Nam Dinh,
Yohei Sato,
Bojan Niceno
Abstract:
Boiling heat transfer occurs in many situations and can be used for thermal management in various engineered systems with high energy density, from power electronics to heat exchangers in power plants and nuclear reactors. Essentially, boiling is a complex physical process that involves interactions between heating surface, liquid, and vapor. For engineering applications, the boiling heat transfer…
▽ More
Boiling heat transfer occurs in many situations and can be used for thermal management in various engineered systems with high energy density, from power electronics to heat exchangers in power plants and nuclear reactors. Essentially, boiling is a complex physical process that involves interactions between heating surface, liquid, and vapor. For engineering applications, the boiling heat transfer is usually predicted by empirical correlations or semi-empirical models, which has relatively large uncertainty. In this paper, a data-driven approach based on deep feedforward neural networks is studied. The proposed networks use near wall local features to predict the boiling heat transfer. The inputs of networks include the local momentum and energy convective transport, pressure gradients, turbulent viscosity, and surface information. The outputs of the networks are the quantities of interest of a typical boiling system, including heat transfer components, wall superheat, and near wall void fraction. The networks are trained by the high-fidelity data processed from first principle simulation of pool boiling under varying input heat fluxes. State-of-the-art algorithms are applied to prevent the overfitting issue when training the deep networks. The trained networks are tested in interpolation cases and extrapolation cases which both demonstrate good agreement with the original high-fidelity simulation results.
△ Less
Submitted 8 June, 2018;
originally announced August 2018.
