Artificial neural networks based computational and experimental evaluation of thermal and drying performance of partially covered PVT solar dryer

This study proposes a mixed-mode dryer with a semi-transparent photovoltaic thermal (PVT) collector for the assessment of drying and thermal performance using computational and experimental findings. The thermal behavior and fluid flow characteristics have been analyzed to optimize the air flow rate in the PVT solar dryer by considering three different inlet velocities of 0.048 m/s (Case 1), 0.096 m/s (Case 2), and 0.144 m/s (Case 3). The temperature distribution is obtained more uniformly for the PVT collector and dryer cabin in Case 2. The results of the investigation show that Case 3 has a positive impact on the PVT solar dryer performance. In numerical and experimental methods, the enhanced thermal efficiency is attained as 30.78% and 29.78% for Case 2, and 33.20% and 31.14% for Case 3, respectively, in comparison to Case 1. Case 3 has improved Reynolds and Nussselt numbers by 3.06 and 2.45 times, respectively compared to Case 1. Experimental results varied by 2.24 to 4.90% from simulated outcomes obtained from CFD. The machine learning approach of ANN has been implemented with different hidden layers network models to choose the best drying conditions by predicting the drying performance parameters.