Computational Fluid Dynamics study of fixed bed adsorbers informed by 3D X-ray Computed Tomography

Abstract

A resolved 3D CFD transient multi-component solver was created, solving the 3D Navier-Stokes equations for the fluid phase and containing adsorption physics as boundary conditions within the surface of spherical particles. The geometry was reconstructed by X-ray computed tomography to be a 3D spherically packed bed and reduced to a packed cube with 11 mm long sides for a more reasonable computational cost. The mesh was created by background meshing initially with (64, 64, 60) cells in the 𝑥, 𝑦 and 𝑧 directions, respectively. The mesh of spherical particles was removed to retain the fluid mesh only and implement extra refinement levels around the spherical particles. Further smoothness was applied at the edges of the packed cube and at the distorted cells due to the imperfect removal of the mesh of the spherical particles. The steady-steady solver was used to generate a maximum air velocity magnitude of 4.5 mm/s. The transient solver was used to generate CO2 mass composition maps depicting how CO2 flow replaces N2 gradually in porous media. The adsorption physics was implemented based on Henry’s and dual-site Langmuir’s equilibrium isotherms. A linear relationship between the rate of CO2 loading was confirmed for the Henry’s isotherm, while non-linear adsorption/desorption behaviour was noticed for the dual-site Langmuir’s equilibrium isotherms. The transient simulation with the dual-site Langmuir’s equilibrium implementation was computationally convergent by a grid convergence index study and validated to have 3% error from the analytical solution.