Enhancing the performance of a photovoltaic thermal system with phase change materials: Predictive modelling and evaluation using neural networks

Abstract

The study aims to develop a Photovoltaic thermal which generates both electrical and thermal energy as well as addresses the electrical power drop as temperature rises. For PV system, absorber tubes with rectangular spiral design were glued to the backside of a multi-crystalline PV panel for transferring the heat from the PV panel. For phase change material (PCM) integrated PV/T, a novel biochar-based PCM was also developed using biochar procured from water hyacinth and pure PCM(OM35) which was inserted in the gap in between the PV and back cover. The average value of electrical power output, electrical efficiency and thermal efficiency for PCM integrated PV/T 65.93 W, 12.54% and 61%, respectively, whereas the average value of electrical power output and electrical efficiency for the PV system is 54.513 W and 11.15%, respectively. The data obtained from field tests were used to develop a neural network model using Multilayer Perceptron for forecasting the performance of the same photovoltaic system. The proposed model accurately forecast the systems output performance as evident from the coefficient of determination (R) and mean error (MSE) values. During the training phase, the predictive model obtains an exceptional R-value of 0.9982 and a good MSE value of 1.1328.