Multiobjective optimization of natural convection and entropy generation in a porous wavy cavity with a heated star-shaped obstacle using a hybrid nanofluid.

Convectional heat transfer continues to receive extensive research attention due to its importance in vital applications such as solar energy collectors and cooling of nuclear reactors. This research investigates natural convection and associated entropy generation within a porous wavy enclosure with a star-shaped hot cylinder. The fluid saturated in the porous medium is assumed to be a water-Cu-Al2O3 hybrid nanofluid. The energy balance equation accounts for thermal non-equilibrium between the hybrid nanofluid and the local structure of the porous medium (LTNE). The effects of the number of lobes of the star-shaped cylinder (N), the Rayliegh number (Ra), the Darcy number (Da), the porosity of the medium (ε), and the volume fraction of the different nanoparticles (φAl2O3 and φCu) are inspected using numerical FEM analysis and optimized by the aid of artificial neural network (ANN). The results show that it is not possible to increase the Nusselt number without increasing entropy, and it is not possible to decrease entropy without decreasing heat transfer. The objective function (OBF), defined as the ratio of the Nusselt number to the total entropy generation, indicates that the best design is achieved with low obstacle waviness and low nanoparticle loading. The number of lobes and the nanoparticle volume fraction often increase entropy faster than they improve heat transfer. The maximum OBF = 3.877 (Nuavg = 38.705, ST = 9.9824) occurs at N = 1, φAl2O3 = 0, φCu = 0, Ra = 103, Da = 10-4, ε = 0.1. This study demonstrates the advantages of using artificial neural networks (ANN) to optimize the design of heat exchangers filled with nanofluids. This approach minimizes losses caused by thermal and flow irreversibility, thereby contributing to energy savings.