The present investigation was aimed at studying the hydrodynamic behaviour of fluidization in an unconventional fluidized bed geometry consisting of a vertically orientated annular space. It was hypothesized that by using multiple gas injectors, orientated tangentially to the annulus walls, induced rotating fluid bed behaviour would occur in the annulus. Annular fluidized beds with induced rotating fluid bed behaviour would ensure complete lateral mixing of solids and gas-solids, and offer the additional advantage that the feed inlet to the fluidized bed can be positioned at a single location.
A physical model, constructed of two Perspex tubes, was used to study the behaviour of a fluidized bed in an annulus with various bed materials and gas distributor designs. In order to study the hydrodynamics of the annular fluidized bed in more detail, representative 2-dimensional computational fluid dynamic (CFD) models were also simulated.
In experiments conducted with multiple tangential air injectors, induced rotating fluid bed behaviour was not observed throughout the entire range of superficial gas velocities tested with any bed material in the annulus. Induced rotating fluid bed behaviour was not observed even after a reduction in the air injector diameter and the addition of a secondary blower. It was concluded that the centrifugal forces were significantly less than initially anticipated with this air distributor design. In addition, it was also concluded that regions of stationary bed material located directly behind the air injectors significantly impeded the momentum transfer between the moving air and bed material, preventing any bed rotation in the annulus.
In the experiments conducted using the overlapping metal leaves air distributor design, stationary regions of bed material were also observed on top of the overlapping metal leaves. These regions were significantly smaller than the regions observed directly behind the tangential air injectors. Unfortunately, induced rotating fluid bed behaviour was not observed with any of the bed materials throughout the entire range of superficial gas velocities tested. It was therefore concluded that the vertical and tangential components (?????????? and ??????????) of the air velocity flowing through the overlapping metal leaves were insufficient to result in uniform fluidization or the desired induced rotating bed behaviour in the annulus.
In order to overcome the problems experienced with the overlapping metal leaves air distributor design, a new air distributor was designed and printed 3-dimensionally to have double the number of slits for the fluidizing air to flow out of. Although induced rotating fluid bed behaviour was once again not observed, uniform bubbling fluidization was apparent during experiments conducted with river sand and poppy seeds as the bed material in the annulus. This observation implied that the vertical component (??????????) of the air velocity flowing through the slits was sufficient to achieve uniform fluidization in the annulus, significantly increasing the mixing effects of the bed material. However, the tangential component (??????????) of the velocity was still insufficient to induce any rotation of the fluid bed.
Based on the sensitivity studies performed, a minimum percentage open area of 1.6% was recommended to ensure uniform fluidization of the bed material in the annulus. It was hypothesized that once uniform fluidization is achieved, induced rotating fluid bed behaviour is likely to occur with sufficient fluidizing air flowing in the tangential direction, since a uniformly fluidized bed experiences little resistance to flow and/or movement. In order to test this hypothesis, it was recommended that future experiments should be conducted using slit (or hole) angles below 45° and with blowers able to achieve higher superficial gas velocities (> 1 m/s) in the annulus.
The Euler-Euler Model Laminar Flow Interface in COMSOL Multiphysics® software was used to study the hydrodynamic behaviour of the annular fluidized bed in more detail. The simulation results proved beneficial since the CFD models could illustrate several hydrodynamic properties not visibly obvious in the experiments conducted with the Perspex model. However, it is recommended that more accurate CFD models of the annular fluidized bed should be developed that model the dispersed phase as a particle size distribution and model the annulus in 3D, and that a turbulent CFD model, capable of simulating high Reynolds numbers, be used. Such a CFD model can then be used to test porous plate-type distributors and variations in slit or hole angles with high superficial gas velocities in the annulus, as previously recommended.
If induced rotating fluid bed behaviour is successfully modelled using CFD, the model can then be used to check whether sufficient contact time is available for the pyrolysis reactions to occur at the recommended higher superficial gas velocities. If not, the design concept of having an induced rotating fluid bed in the annular pyrolysis chamber of the new fluidized bed fast pyrolyser could be impractical, and in such an event, a rotating gas distributor plate should be considered instead.
Dissertation (MEng)--University of Pretoria, 2015.