(691g) Comparison of the Filtered Two Fluid Model and Dense Discrete Phase Model for Large Scale Fluidized Bed Reactor Simulations

Fundamental modelling of fluidized bed reactors is generally done via the two fluid model (TFM) approach where the effects of non-resolved particle motions and collisions is modelled by the kinetic theory of granular flows (KTGF). This approach has been developed to a high level of maturity over the past three decades, but is not applicable to large-scale 3D simulations of fluidized bed reactors. TFM simulations generally require very fine grid sizes to adequately capture the dynamics of small particle clusters and/or gas bubbles that form in a fluidized bed. If these structures are not adequately captured, large prediction errors can occur because all transport phenomena in a fluidized bed reactor are directly influenced by cluster/bubble dynamics. 

One promising approach for computationally affordable simulations of large scale fluidized bed reactors is the filtered TFM (fTFM) approach. This multiscale modelling approach uses a large set of resolved TFM simulation data to derive closures for the effect of particle structures on transport phenomena in the bed. These filtered closures can then be implemented in a large-scale 3D simulation using a coarse computational grid which cannot directly resolve particle structures. All required closures for fTFM simulations of a fluidized bed reactor are available in the literature, but more work is required to verify and improve the generality of these closures. 

Another promising approach to large-scale fluidized bed simulations is the dense discrete phase model (DDPM). This approach, also known as the multiphase particle in cell (MP-PIC) method, uses a hybrid Eulerian-Lagrangian framework to track Lagrangian particle parcels through the domain and model particle interactions based on local Eulerian data. This approach still resolves particle structures, but can do so on much courser meshes than the TFM, primarily because the Lagrangian particle tracking eliminates the problem ofnumerical diffusion. The DDPM therefore does not introduce the uncertainty of additional closures in the fTFM, but will still be subject to grid dependency issues in large scale simulations. 

This work will critically compare the fTFM and DDPM approaches in order to identify the conditions where one approach is clearly favoured over the other. Since particle structures must still be resolved by the DDPM, it may be expected that accurate large-scale simulations using fine powders are still out of reach. In these situations, the fTFM approach will be the only alternative. The study will therefore present and discuss results from fluidized bed reactor simulations carried out for increasing reactor sizes using both approaches. Validation studies previously completed for both approaches will also be presented.