The use of staggered grid in numerical solution of nanofluid convection within a rectangular chamber with semicircular fins and the use of artificial neural networks in optimizing dimensionless numbers

Abstract

The present study investigates the mixed convection of the Al2O3-water nanofluid in a rectangular cavity using artificial intelligence (AI). The top wall of the cavity was cold and lid-driven, while the bottom wall and its five semicircular fins were at a high temperature. The two lateral walls were insulated. The cavity was under a magnetic field. The entropy production (ENP), Nusselt (Nu) number, and Bejan (Be) number were studied at Hartmann (Ha) numbers of 0–100, Richardson (Ri) numbers of 0.01–100, and Rayleigh (Ra) numbers of 103-106. Also, optimization was carried out to maximize the Nu and minimize the ENP and Be in the aforementioned ranges. The simulations were performed using the SIMPLE algorithm and finite volume method (FVM). Moreover, a staggered grid was employed for the simulation. It was found that a rise in the Ha increased the Nu and reduced the ENP. An increase in the Ri diminished the ENP and Nu and raised the Be. The Nu and ENP increased as the Ra increased, while the Be declined. A maximum Nu of 3.69 was obtained, while the minimum ENP and Be were 10.94 and 0.25, respectively, predicted with 89% desirability.