High-Fidelity CFD Modeling of Particle-to-Fluid Heat Transfer in Packed Bed Reactors
Multi-scale and/or multi-disciplinary approach to process-product innovation
CFD & Chemical Engineering- I (T3-4a)
Keywords: high-fidelity CFD modeling, packed bed reactor, heat transfer, DEM
A better understanding of packed beds can be gained by developing high fidelity models that take into consideration the flow of heat, mass and momentum through and around the particles that constitute the packed bed. This has been prohibitively expensive because of the enormous size of the problem when computing the flow around each of the many particles in a packed bed. However, the continuing advances in computer performance, such as distributed parallel computation, easier CFD mesh generation and more efficient CFD solvers are opening the way for more complete descriptions of packed beds. Packed beds are modeled by allowing spherical particles to fall randomly under gravity into a cylinder. In this study the commercial code EDEM (DEM Solutions Ltd) has been utilized. However, the automatic meshing of a large number of particles with low mesh skewness creates a very fine mesh, and therefore a large mesh, especially in the narrow gaps between particles. In this approach cylindrical shaped solid geometries are used to bridge spherical particles in contact. The diameter and the physical properties of the bridges can be adjusted. Alternative approaches involve making the particles larger so that they overlap slightly. However, compared with this latter approach, the method described in this paper keeps the mesh voidage practically the same. Moreover, the bridges reduce the requirement for very small elements by eliminating narrow gaps between particles thus reducing the overall mesh size. In the present study, the predication accuracy of particle-to-fluid heat transfer is examined in terms of the relation between Nusselt (Nu) and Reynolds (Re) numbers. 220 of spherical particles are randomly packed in a cylindrical tube whose diameter and length are 4 and 50 times larger than the particle diameter, respectively. The upwards inlet velocity of air with 300 K is varied in the particle Re range of 0.1 to 1000. The temperatures of tube and particle are specified as 300 K and 400 K, respectively. The Nu is evaluated based on heat flux through the particle surface and bulk-averaged temperature in the bed region. All the CFD processes are performed by a commercial code based on finite volume method (FLUENT6, Fluent Inc.). The predicted Nu is compared with a widely accepted correlation, experimental results and model predications. In the smaller Re range, mesh-independent solutions are easily obtained, which well agree with the literature. In the higher range, the applicable Re range within 15 % deviation from the correlation varies from 200 to 700 depending on the examined mesh density. This model retains the true geometry of the bed as it does not treat the packed bed as a porous media. This novel methodology is a potential tool for process intensification since the local fraction and size distribution of catalytic and absorbent / inert particles can be taken into consideration when necessary.
See the full pdf manuscript of the abstract.
Presented Tuesday 18, 11:40 to 12:00, in session CFD & Chemical Engineering- I (T3-4a).
