CFD-driven regression models for enhancing photovoltaic performance with nano fluid Cooling: Economic and environmental analysis

Enhancing photovoltaic (PV) efficiency is crucial for sustainable energy production, as elevated temperatures can decrease PV performance. This study evaluated a 300 W PV panel under Yazd city conditions, both with and without cooling. Numerical data from ANSYS Fluent simulations are used to train various regression models, with the most accurate model selected for predicting annual energy output and minimizing costly experimental tests. The cooling performance of four water-based nanofluids Al 2 O 3 , TiO 2 , Fe 2 O 3 , and ZnO is assessed under radiation of 900 W. m −2 and a convection coefficient of 4.30 W. m −2 . K −1 at 1:00 p.m. ZnO, at 5 % concentration and an inlet velocity of 0.20 m. s −1 , exhibited the highest cooling efficiency. Thermal contour analysis evaluated a non-cooled PV, a water-cooled PV thermal (PV/T), and a PV/T using ZnO nanofluid. It was found that ZnO achieved a 6.3 % temperature reduction compared to no cooling. Support vector regression (SVR) with a POLY kernel is the most accurate for non-cooled PV, while multiple linear regression (MLR) excels for PV/T systems. Estimated annual energy from the best regression model showed that ZnO increased PV efficiency by about 16 %. The payback period for PV and PV/T is 3.67 and 4.57 years, respectively, while reducing CO 2 emissions by 24 kg in PV/T.