Numerical strategies for simulating flow through thin isotropic porous media: benchmark between Ansys Fluent and OpenFOAM

Abstract

The present work performs a benchmark between the softwares Ansys Fluent and OpenFOAM for the simulation of flow in thin isotropic porous media. The single-phase, turbulent, and steady-state simulations were conducted based on experimental data from the literature, namely, pressure drop data by filtration velocity resulting from flow in a filtration box, which allowed adjusting the parameters of viscous and inertial losses for the macroscale modeling of the flow in a filter. The results showed agreement between the adjusted and simulated pressure drop obtained by the pressure averages at the porous medium inlet and outlet. The analysis of velocity and static pressure contours proved to be coherent with the flow dynamics in both softwares with deviations ranging between 1.43% and 12.37% for maximum pressure and between 0.97% and 4.18% for maximum velocity. The study revealed critical issues for modeling flow in porous media through the adjustment of pressure drop data by filtration velocity: the viscous and inertial resistance coefficients calibrated by the regression of experimental data induced pressure drop errors of 10%, 4.6%, and 1%, for filtration velocities of 5, 10, and 15 cm/s, respectively; the approach used for data validation based on the average pressures at the porous medium inlet and outlet proved to be robust due to the capture of the pressure distribution over the entire filter surface; the modeling of the inlet velocity based on the mass balance between the porous medium and inlet, resulted in numerically higher filtration velocities than the theoretical ones, indicating the need for an iterative adjustment in this condition to ensure that the required filtration velocity is properly reproduced in the porous medium; the high-pressure gradients, boundary layer detachment, and recirculation zones evidenced the need for using more robust turbulence models for the capture of the involved phenomena; and finally, the boundary condition approach using porousBafflePressure presented unsolved convergence challenges for the tested scenarios, preventing its viability.