Prediction of the Hitec Molten Salt Convective Heat Transfer Performance Using Artificial Neural Networks

Abstract

Hitec molten salt is a ternary eutectic mixture salt that is used as an energy storage medium in concentrated solar power plants to improve the system performance and reduce the operational cost. Thus, the heat transfer performance represented in Nusselt number has been investigated numerically under different inlet temperature and velocity conditions with constant uniform side heat flux. Also, friction factor and mass flow rate are studied numerically. CFD input/output data with 40 studied cases are used as a training dataset of a 2-layer Neural Network for thermo-hydro fields’ accelerated results predictions. Bayesian regularized Neural Network showed a satisfactory agreement for thermo-hydro fields’ predictions compared to the CFD results for the testing dataset not included in the training of the network.