Analysis of natural convection heat transfer in a rectangular cavity with discrete heat flux: Implications for building thermal management using artificial neural networks

Abstract

This study numerically investigates free convection within a rectangular air-filled cavity, simulating real-romm conditions. The top, bottom, and one sidewall are at constant temperatures, while the opposite sidewall has a constant discrete heat flux, akin to heater appliances. The impact of heating intensity, length, and position on temperature distribution is explored. Artificial Neural Networks (ANN) are utilized to correlate the average Nusselt number, providing a model for engineering applications in building thermal management. The dataset includes 2436 simulation runs with varying parameter: Rayleigh number (103 to 106), aspect ratio (0.5 to 2), heating surface length (0.1 to 1), and elevation (0.05 to 0.95). Results show increased Rayleigh numbers intensify the stream function and promote uniform temperature distribution. The elevation of the heating surface influences temperature distribution, with placement closer to the floor or ceiling optimizing heat transfer. ANN modeling predicts the average Nusselt number with high precision (±3%).