Determination of thermal efficiency of air ETSC-PCM using artificial neural network technique

Abstract

In this study, the thermal efficiency of an air-type vacuum tube combined phase-change material solar collector (air ETSC-PCM) was tested to establish a theoretical mathematical model of thermal efficiency through linear approximation. Various artificial neural network (ANN) models were used to predict the thermal efficiency of the collector and the results were compared with actual experimental data to improve the thermal efficiency of the collector. The ANN model used six parameters (irradiation intensity, inlet temperature, outlet temperature, ambient temperature, air temperature in vacuum tube, and PCM temperature) and two parameters (thermal efficiency of fluid and thermal efficiency of PCM) for the input layer and the output layer, respectively. A statistical error analysis demonstrated that the prediction result of the LM6-3-2 algorithm had a higher R2 value, the prediction errors of BR6-5-2 and RBF-2 algorithms were similar, and the accuracy of the SCG6-5-2 algorithm was the lowest. In addition, linear approximation proved to be suitable for predicting the thermal efficiency of the fluid. However, an opposite was observed while predicting the thermal efficiency of the PCM, and a maximum error of 2.13 was obtained. Additionally, the sensitivity of input variables was investigated and it was demonstrated that solar radiation had the greatest impact on thermal efficiency. Furthermore, it was observed that the thermal efficiency can be improved by reducing the thermal resistance between the vacuum tube and the air to enhance the rate of heat transfer in the aluminum shell of PCM-rod.