CuO‑nanofluid cooling PV/T using machine learning under Iraqi climatic conditions

Abstract

The performance of photovoltaic/thermal (PV/T) systems declines significantly under hot climatic conditions because rising cell temperatures reduce electrical conversion efficiency and limit the recovery of useful heat. This study aims to evaluate the effectiveness of CuO/water nanofluid cooling in improving PV/T performance under Iraqi climatic conditions and to assess the capability of machine-learning models to predict system efficiency. A three-dimensional CFD model was developed in ANSYS Fluent for a PV/T system operating under steady laminar flow, with CuO/water nanofluid concentrations ranging from 0.25 to 1 vol%. The numerical analysis considered temperature distribution, pressure drop, and thermal-electrical performance. At the same time, Linear Regression, Random Forest, and Decision Tree models were trained using the simulation outputs to predict system efficiencies. The results showed that increasing the CuO nanoparticle concentration improved cooling performance, with the 1 vol% nanofluid exhibiting the best overall performance. In this case, the PV surface temperature decreased from 53.4 °C under uncooled conditions to 41.15 °C, while thermal efficiency reached 68.22%. The pressure drop increased slightly with nanofluid concentration but remained within an acceptable range. The machine-learning models also showed strong predictive capability, with the best model achieving an R² value of 1.00 for overall efficiency prediction. These findings confirm that CuO/water nanofluid is a promising cooling medium for PV/T applications in hot regions and that machine-learning tools can support accurate performance prediction and system optimization.