Analyzing geometric parameters in inclined enclosures filled with magnetic nanofluid using artificial neural networks

In this article, natural alumina/water nanofluid (NF) convection in an isosceles equilateral rhombus-shaped enclosure was simulated using the Simplex algorithm and the control volume method. The enclosure under study had two insulation walls, i.e., a cold wall and a warm wall. Two blades were installed on the warm wall with a temperature equal to that of the warm wall. There was also a fin in the center of the enclosure with a temperature equal to that of the warm wall. The enclosure was horizontally under a magnetic field at Hartmann number (Ha) of 20. The average Nusselt number (Nu), entropy production, Bejan number (Be), and flow and temperature contours were studied while altering the length and thickness of the blades from 0.1 to 0.8 and 0.05 to 0.15, respectively, and the aspect ratio (AR) of the fin from 0.1 to 0.4. The obtained results were then optimized to catch the best results. The two-phase method was used to simulate nanofluid flow. By altering the width and length of the blades and the fin AR, the average Nu varies from 6.52 to 8.31. According to the results, within the range of the above variables, Nu, entropy production, and Be varied from 5.62 to 8.31, 7.55 to 12.36, and 0.48 to 0.6, respectively.