Fixed bed reactors are widely applicable in a range of chemical process industries. Their ease of use and simplified operation make them an attractive and preferred option in reactor selection, however the geometric complexities within the bed as a result of the unstructured packing has made the design of such beds historically based on pseudo-homogenous models together with correlation-based transport parameters. Low tube-to-particle diameter ratio (N) beds, in particular, are selected for highly exothermic or endothermic reactions, such as in methane steam reforming or alkane dehydrogenation. Due to the large fraction of tube to catalyst particle contact in these low N beds, wall effects induce a mass transfer boundary layer at the wall, and in the case of thermal beds, a simultaneous resistance to heat transfer. Computational Fluid Dynamics (CFD) has been shown to be an accurate tool for experimental validation and predictive analysis of packed beds, and may be used to derive more accurate design parameters for fixed bed reactors. More specifically, the elucidation of dispersion, or the transport of reactant and product within the bed due to molecular diffusion and convective flow is of fundamental interest to the design of fixed beds. Computational Fluid Dynamics was used in this research to study solute dispersion in eight beds of varying N at a range of particle Reynolds numbers in the laminar flow regime. In the first stage of research, flow development was simulated in three-dimensional packed beds of spheres. Then, the reactor wall was sectioned to include a boundary condition of pure methane, from which the solute could laterally disperse into the bed. In the second stage, a two-dimensional representation of the bed was created using the commercial Finite Element Analysis software COMSOL Multiphysics. In these models, axial velocity profiles and radial methane concentration profiles taken from the 3-D models were supplied, and a fitting procedure by use of the Levenberg-Marquardt Least-Squares optimization algorithm was completed to fit radial dispersion coefficients and near-wall mass transfer coefficients to the CFD data. These optimization runs were conducted for all N at a number of bed depths in each case. Two sub-studies were conducted in which a constant velocity profile and a local velocity profile were supplied to the 2-D model, and the optimization re-run. It was found that this two parameter model did not fully account for various mechanisms of dispersion in the bed, namely the increasing rate of dispersion from the tube wall boundary layer up to the bed center, but only accounted for a diffusive-dispersion at the wall and a constant-rate, convective-dispersion everywhere else in the bed. Length dependency of dispersion coefficients were also noted, particularly in the developing sections of the bed. Nevertheless, the combined CFD and optimization procedure proved to be an accurate and time-efficient procedure for the derivation of dispersion coefficients, which may then lend themselves to the standard design of packed bed reactors.

Fixed bed reactors are widely applicable in a range of chemical process industries. Their ease of use and simplified operation make them an attractive and preferred option in reactor selection, however the geometric complexities within the bed as a result of the unstructured packing has made the design of such beds historically based on pseudo-homogenous models together with correlation-based transport parameters. Low tube-to-particle diameter ratio (N) beds, in particular, are selected for highly exothermic or endothermic reactions, such as in methane steam reforming or alkane dehydrogenation. Due to the large fraction of tube to catalyst particle contact in these low N beds, wall effects induce a mass transfer boundary layer at the wall, and in the case of thermal beds, a simultaneous resistance to heat transfer. Computational Fluid Dynamics (CFD) has been shown to be an accurate tool for experimental validation and predictive analysis of packed beds, and may be used to derive more accurate design parameters for fixed bed reactors. More specifically, the elucidation of dispersion, or the transport of reactant and product within the bed due to molecular diffusion and convective flow is of fundamental interest to the design of fixed beds. Computational Fluid Dynamics was used in this research to study solute dispersion in eight beds of varying N at a range of particle Reynolds numbers in the laminar flow regime. In the first stage of research, flow development was simulated in three-dimensional packed beds of spheres. Then, the reactor wall was sectioned to include a boundary condition of pure methane, from which the solute could laterally disperse into the bed. In the second stage, a two-dimensional representation of the bed was created using the commercial Finite Element Analysis software COMSOL Multiphysics. In these models, axial velocity profiles and radial methane concentration profiles taken from the 3-D models were supplied, and a fitting procedure by use of the Levenberg-Marquardt Least-Squares optimization algorithm was completed to fit radial dispersion coefficients and near-wall mass transfer coefficients to the CFD data. These optimization runs were conducted for all N at a number of bed depths in each case. Two sub-studies were conducted in which a constant velocity profile and a local velocity profile were supplied to the 2-D model, and the optimization re-run. It was found that this two parameter model did not fully account for various mechanisms of dispersion in the bed, namely the increasing rate of dispersion from the tube wall boundary layer up to the bed center, but only accounted for a diffusive-dispersion at the wall and a constant-rate, convective-dispersion everywhere else in the bed. Length dependency of dispersion coefficients were also noted, particularly in the developing sections of the bed. Nevertheless, the combined CFD and optimization procedure proved to be an accurate and time-efficient procedure for the derivation of dispersion coefficients, which may then lend themselves to the standard design of packed bed reactors.