Membranes in fluidized beds

Fluidized bed membrane reactors combine the excellent separation properties of membranes with the advantages of uidized bed reactors, such as circumventing equilibrium limitations and achieving higher product yields without ine??ciencies due to hot spot formation or concentration polarization. These advantages have clearly led to an increasing number of applications of gas-solid uidized bed membrane reactors being proposed and investigated worldwide. Nevertheless, detailed understanding of the e??ect of immersed membranes on the behavior of uidized beds is still largely lacking. However, good closure relations are crucial for the predictive capabilities of phenomenological models used for the design of membrane-assisted uidized bed reactors, and these require detailed knowledge about the uidized bed hydrodynamics and the e??ect of the presence of - and permeation through - membranes. To bridge the gap between theory and application, this study focuses on the hydrodynamic aspects of uidized beds with immersed membranes. In particular, attention is paid to the solids circulation patterns as well as the bubble size distribution. Closely related to these important topics are solids mixing, energy dissipation and the porosity distribution; all these aspects determine the reactor performance to a large extent. In this study, these aspects have been investigated in detail from both an experimental as well as a numerical perspective. Two di??erent membrane con??gurations have been considered: uidized beds with vertical ( at) membranes con??ning the uidized gas-solid emulsion and uidized beds with immersed horizontal membrane tubes. The unique combination of Particle Image Velocimetry (PIV) and Digital Image Analysis (DIA) allows simultaneous and non-invasive experimental determination of the solids ux pro??les as well as the bubble properties, which are strongly correlated. For this purpose, an improved DIA algorithm, based on arti??cial images created from a Discrete Particle Model (DPM) simulation, thus correlating 2D intensity and 3D 6 Summary - Membranes in Fluidized Beds porosity, has been developed, thoroughly tested and applied. Compared to the conventional DIA, the error in the solids uxes is decreased from 18.7 % to 12.3 %. The DPM is an Euler-Lagrange model that solves the gas ow ??eld and pressure distribution on a Cartesian grid, while the particles are tracked individually using Newton's second law of motion while accounting for collisions. To allow round objects - such as membrane tubes - to be included, an explicit Immersed Boundary Method was integrated into the existing in-house DPM code. For su??ciently small time steps and realistic packing fractions the predictions of the new model are in line with literature. Counter-intuitively, adding gas via vertical membranes in the left and right wall to the uidized bed decreases the average bubble size, because the additional voids near the membranes cause bubbles to split up and move to either side. This results in an inverted solids circulation pattern, where particles move downward via the center. On the other hand, during gas extraction, semi-stagnant zones near the walls force both particles and gas towards the center of the bed, resulting in slightly larger bubbles. Despite some shortcomings of both DPM and TFM (the latter in particular with respect to the predicted porosity distribution), in all cases qualitative agreement between simulations and experiments is found for di??erent system sizes. With respect to vertical at membranes it can be concluded that by tuning the permeation rate through the membranes, the bed properties can be optimized. Immersed horizontal membrane tubes show a very di??erent e??ect on the hydrodynamics than vertical membranes; merely the presence of the membranes causes the bed height to decrease and thus creates a more homogeneous distribution of solids throughout the entire bed. Moreover, the bubble size is signi??cantly decreased. The solids circulation pattern remains unchanged, although is decreased in magnitude. Despite the signi??cant changes caused by the presence of the membrane tubes, the permeation rate, membrane tube con??guration, tube diameter and number of tubes hardly show any in uence on solids motion and bubble properties. DPM simulations qualitatively con??rm all these results. Furthermore, the simulation results reveal that both energy dis7 sipation as well as gas back-mixing are decreased in systems containing immersed membrane tubes, while solids mixing times are increased. For the design of future uidized bed membrane reactors, these results imply that horizontal membrane tubes may be preferred over vertical membranes because of the enhanced bubble-to-emulsion phase mass transfer resulting from the decrease in the average bubble size.