Machine learning insights and performance assessments into nanofluid-enhanced PV-T solar collector

Abstract

Photovoltaic-thermal (PV-T) systems combine solar thermal absorbers with PV cells, enhancing solar energy capture by producing both electrical and thermal energy. This hybrid setup improves PV cell efficiency and supports heating applications. Artificial Neural Network (ANN) models, effective in handling complex, non-linear interactions, are used to predict PV-T performance under various conditions. In this study, a photovoltaic/thermal (PV-T) collector employing nanofluids as the cooling medium is thoroughly investigated. Numerical models are developed to analyze the influence of key operating parameters, such as nanofluids type and their concentration, solar irradiation, and mass flow rate, on performance indicators of PV-T collectors. This research introduces a novel approach by integrating an artificial neural network (ANN) model to performance prediction of the PV-T collector. The ANN model is validated against numerical data and provides a tool that aids in both optimizing operating conditions of the PV-T collector and rapidly designing new experiments. Key findings reveal that increasing nanofluid concentration enhances convective heat transfer, reducing absorber plate temperatures. For Al₂O₃-Water at a flow rate of 3 L/min, maximum absorber plate temperatures drop to 313 K, 312 K, and 310 K for concentrations of 1 %, 2 %, and 3 %, respectively—reductions of up to 6 K compared to water. Similarly, CuO-Water achieves reductions of up to 7 K under the same conditions. The ANN model achieves R² values exceeding 0.97 for all performance metrics, with prediction errors below 0.1 for Al₂O₃-Water and 0.05 for CuO-Water electrical efficiency.