Numerical study and multilayer perceptron-based prediction of melting process in the latent heat thermal energy storage system with a finned elliptical inner cylinder

Abstract

Abstract This study numerically investigated the melting phenomena of a phase change material (PCM) in a latent heat thermal storage system with a finned elliptical inner cylinder. The enthalpy–porosity model was used to simulate the phase change process of the PCM. A parametric study was performed by varying the length of the fin in the range of 5, 6.5, and 8 mm and at temperatures of 56.15, 69.16, and 92.22°C, considering three different geometric types. The distributions of the liquid fraction, temperature, and velocity vector were numerically investigated to analyze the effects of these variables on the melting performance of the latent thermal energy storage system. The results indicate that the heat and mass transfer characteristics can be significantly affected by changes in the geometric types, length of the fin, and temperature of the inner cylinder. Based on the results from the numerical simulations, the multilayer perceptron neural network model was trained to estimate the melting time of the PCM with respect to the changes in the main variables considered. It was found that a multi-layer perceptron neural network model can estimate the melting time more accurately than the least square regression method.