Computational fluid dynamics analysis and machine learning study of heat transfer in solar air heaters with distinct ribs configuration

Solar Air Heaters (SAHs) are crucial for sustainable energy, but their efficiency requires significant enhancement for broader impact. This research pioneers SAH optimization by uniquely integrating Computational Fluid Dynamics (CFD) with machine learning (ML) to analyze and predict the performance of 15 SAH designs featuring distinct curved rib configurations. The novelty lies in identifying specific high-performance geometries and demonstrating a powerful ML-driven predictive capability. CFD simulations pinpointed an optimal rib thickness-to-height (t/h) ratio of approximately 0.25 and a thickness-to-pitch (t/p) ratio of 0.075, at which heat transfer parameters were maximized. Notably, increasing t/p from 0.025 to 0.075 improved performance, while further increases diminished efficiency. Reynolds number (Re) analysis showed enhanced convective heat transfer at higher Re, with performance gains plateauing between 15,000 and 25,000. Critically, the developed Convolutional Neural Network (CNN) model significantly outperformed Random Forest Regression and Support Vector Regression, achieving a Mean Square Error (MSE) of 0.004 and an R2 value of 0.96 in predicting heat transfer coefficients. This high predictive accuracy underscores the potential of CNNs to accelerate the design of efficient SAHs. The study's findings offer precise geometric and operational guidelines, providing a timely and impactful contribution by advancing SAH design through a potent CFD-ML synergy.