Investigation into the velocity distribution through an annular packed bed
Reyneke, Hendrik Jacobus
MetadataShow full item record
The purpose of this study was to investigate the velocity distribution through an annular bed packed randomly with equal sized spheres. Extensive research has been conducted on the velocity distribution inside packed beds packed with equal sized spheres, different sized spheres, deformed spheres, cylinders and Raschig-rings. A majority of these experimental and numerical studies focused on the cylindrical packed bed. These studies and numerical models are all confined to the velocity profile once the fluid flow is fully developed. The development of the velocity through the inlet region of the bed and the fluid flow redistribution in the outlet of the bed is thus neglected. The experimental investigation into the velocity distribution down stream of the annular packed bed of the HTTU indicated that the velocity profile was independent of the mass flow rate for a particle Reynolds number range of 439 £ Re £ 3453 . These velocity profiles did not represent the distribution of the axial velocity due to shortcomings associated with the single sensor hot wire anemometry system used to measure the velocity distribution. A numerical investigation, using the RANS CFD code STAR-CCM+®, into the velocity distribution downstream of an explicitly modelled bed of spheres indicated that the axial velocity distribution could be extracted from the experimental velocity profiles by using an adjustment factor of 0.801. This adjusted velocity profile was used in the verification of the implicit bed simulation model. The implicit bed simulation model was developed in STAR-CCM+®. The resistance of the spheres was modelled using the KTA (1981) pressure drop correlation and the structure of the bed was modelled using the porosity correlation proposed by Martin (1978), while the effective viscosity model of Giese et al. (1998), adjusted by a factor of 0.8, was used to model the velocity distribution in the near wall region. It was found that the structure in the inlet region of the bed, where two walls disturb the packing structure, can be modelled as the weighted average of the radial and axial porosity while the structure in the outlet regions can be modelled by letting the radial porosity increase linearly to unity. The basic shape of the velocity profile is established immediately when the fluid enters the bed. The amplitude of the velocity peaks however increase in magnitude until the velocity profile is fully developed at a distance approximately of five sphere diameters from the bed inlet. The profile remains constant throughout the bed until the outlet region of the bed is reached. In the outlet region a significant amount of fluid redistribution is observed. The amplitude of the velocity peaks is reduced and the position of the velocity peaks is shifted inwards towards the centre of the annular region. The fully developed velocity profile, predicted by the simulation model is in good agreement with profiles presented by amongst others Giese et al. (1998). The current model however also offers insight into the development of the profile through the inlet of the bed and the fluid redistribution, which occurs in the outlet region of the bed.
- ETD@PUK