Integrated intelligent neuro computing technique for mixed convective flow and heat transfer in heterogeneous permeability media

Abstract

This article examines the flow and heat transfer characteristics of variable permeability such as linear, exponential and quadratic permeability functions with Brinkmann–Forchheimer extended Darcy model by utilizing the intelligent computing paradigm via artificial Levenberg–Marquardt back propagated neural networks (ALM – BPNNs). The vertical porous channel walls are considered as electrically conducting and non-conducting. The variable permeability and induced magnetic field effects are significant on flow and heat transfer. To solve the obtained governing equations, finite element method is adopted. It is observed that for the three cases of permeability functions, increase in Hartmann number drops the velocity profile and is accompanied with an increase in the fluid thermal state level. The significance of the Brinkmann number on the fluid temperature in all the cases of permeability indicates a rise in temperature due to increased heat generation. But, the reverse effect is observed when the mixed convection parameter rises, and the boundary layer thickness reduces due to which temperature drops for distinct cases of permeability under consideration. The heat transfer characteristics are analyzed through an artificial neural network and multiple linear regression models. The present work finds its applications in studying packed bed heat exchangers and fixed-bed catalytic reactors.