Heat transfer and entropy generation analysis of water-Fe3O4/CNT hybrid magnetic nanofluid flow in a trapezoidal wavy enclosure containing porous media with the Galerkin finite element method

The present study addresses theoretically and computationally the performance of electricallyconducting water-Fe3O4/CNT hybrid nanofluid in three-dimensional natural convective heat transferand entropy generation within a wavy-walled trapezoidal enclosure. The enclosure has two layers - ahybrid nanofluid layer and a porous medium layer. A transverse magnetic field is applied in the upwarddirection. Newtonian flow is considered and the modified Navier-Stokes equations are employed withLorentz hydromagnetic body force, Darcian and Forchheimer drag force terms. The wavy side planesare heated down while the top and vertical planes are thermally insulated. A rectangular heated fin isplaced in the lower plane and several different locations of the fin are considered. The transformed,non-dimensional system of coupled non-linear partial differential equations with associated boundaryconditions is solved numerically with the Galerkin finite element method (FEM) in the COMSOLMultiphysics software platform. The effects of Darcy number, Hartmann number, volume fraction,undulation number of the wavy wall and Rayleigh number (thermal buoyancy parameter) on thestreamlines, isotherms and Bejan number contours are studied. Extensive visualization of the thermalflow characteristics is included. With increasing Hartmann number and Rayleigh number, the averageBejan number is reduced strongly whereas average Nusselt number is only depleted significantly atvery high Rayleigh number and high Hartmann number. With increasing undulation number, there is aslight elevation in average Bejan number at intermediate Rayleigh numbers, whereas the average Nusselt number is substantially boosted, and the effect is maximized at very high Rayleigh number.Increment in Darcy number (i. e. reduction in permeability of the porous medium layer) is observed toconsiderably elevate average Nusselt number at high values of Rayleigh number, whereas the contraryresponse is computed in average Bejan number. The simulations are relevant to hybrid magneticnanofluid fuel cells and electromagnetic nano-materials processing in cavities