Numerical investigation of heat transfer and thermo-hydraulic performance of solar air heater with different ribs and their machine learning-based prediction

The utilization of artificial roughness, in the form of repeating ribs on a surface, is a proficient method for enhancing the rate of heat transmission. This study entailed a numerical analysis of heat transfer and thermo-hydraulic performance in a solar air heater, which incorporated three types of ribs on the absorber plate. Applying a consistent heat flux of 1000 W m-2 to the upper surface of the plate generated flow across 141 distinct Reynolds numbers ranging from 4000 to 18,000 and at varied relative roughness of 7.14, 10.71, and 14.29. Through numerical investigations with machine learning techniques like support vector machine (SVM), artificial neural network (ANN), and extreme gradient boosting (XGB), the Nusselt number, friction factor, and thermo-hydraulic performance parameters (THPP) were predicted. The maximum value of the friction factor on a corrugated absorber plate with a triangular section is achieved at P/e = 7.14 and a Reynolds number of 7200. It has been found that the highest Nusselt number ratio of 2.09 corresponds to a relative roughness pitch of 7.14 at a Reynolds number of 7500 within the triangular section. In addition, the maximum THPP value of 1.77 was obtained utilizing a semi-circle section, P/e = 14.29, and a Reynolds number of 18,000. The XGB model provided the most accurate estimates for the Nusselt number (R2 = 0.9993) and friction factor (R2 = 0.9985).