Empirical and Machine Learning Approaches for Turbulent Thermal Convection in Rectangular Enclosures Tilted at Acute Angles

The present study proposes both an empirical and a machine learning approach to predict Nusselt numbers for turbulent thermal convection in rectangular enclosures inclined at acute angles for a given range of Rayleigh numbers (1.85 × 106–1.04 × 1011) and aspect ratios ( \(1, 3, 6, 10\) ). Nusselt numbers predicted using the empirical correlation and the machine learning model are found to yield a mean absolute percentage error of \(5\) and \(7\%\) , respectively, when compared with the corresponding experimental values. Nusselt numbers predicted using both the approaches are also compared with the corresponding reported values from the literature and found to agree reasonably well.