Analyzing the effect of normally distributed cooling channels on a photovoltaic thermal solar unit

Abstract

Photovoltaic thermal (PVT) systems convert solar radiation into electrical and thermal energy while keeping the PV's temperature within a permissible range. This paper presents a numerical analysis of a novel normal cooling technique for the PV module and a new AI prediction model. It considers a cooling direction normal to the PV panel using rectangular cascade channels with two different orientations. This cascading of the flow inlet channels helps keep the inflow temperature as cold as possible along the PV panel. The numerical simulations are performed based on a flow rate range of 0.00024: 0.0012 LPM, insolation of 400: 1200 W/m2, convection coefficients of 5.48 and 20.7 W/m2.K, and ambient temperatures of 298 and 308 K. The proposed PVT system achieves a decrease in the cell's average temperature by 27K while improving the electrical, thermal, and overall efficiencies by 4.8%,59.5%, and 73.33%, respectively. The numerical database is used for building an ANN prediction model based on a radial base function neural network with Gorilla troops optimizer (RBFNN-GTO). The proposed RBFNN-GTO model is developed to predict the main performance parameters of the PVT which are thermal efficiency, electrical efficiency, overall efficiency, average temperature of the PV panel, and the model temperature un-uniformity. It achieves an overall correlation coefficient of 0.9997 concerning the numerical database and showed an average MSE of 5.4313E-4.