Physics-Informed Neural Network Investigation of Thermal Transport in Rotating Anisotropic Porous Media

Abstract This study explores the complex thermo-fluid behavior of anisotropic, fluid-saturated porous media in a rotating channel using Physics-Informed Neural Networks (PINNs). The nonlinear Darcy–Brinkman–Forchheimer equations, formulated in a rotating frame of reference, are solved directly using a mesh-free, data-efficient PINN framework. This approach enables the accurate capture of intricate multiphysics interactions that arise from the interplay of anisotropy, rotation, and viscous effects. The velocity field exhibits a primary axial component aligned with the pressure gradient and a secondary transverse component generated by Coriolis forces. Results reveal that variations in rotation rate, anisotropy, and permeability orientation significantly influence both flow and heat transfer characteristics. Axial flowrates fluctuate by up to 96%, while Nusselt numbers vary by over 30%, indicating substantial sensitivity to these parameters. Viscous dissipation contributes to asymmetric thermal behavior across the channel walls: enhanced heat transfer occurs at the bottom wall, whereas the top wall can experience a reduction or even reversal in heat flux under certain regimes. These asymmetries are particularly pronounced in highly anisotropic configurations, where the direction-dependent permeability modulates the influence of rotational forces on the flow field. Overall, the study highlights the effectiveness of PINNs in resolving coupled, nonlinear phenomena in rotating porous systems without the need for traditional meshing. The findings provide valuable insights for the design and optimization of thermal systems in engineering applications such as geothermal energy extraction, aerospace thermal protection, and advanced cooling technologies.