(110a) Experimental and Numerical Study of Structure and Hydrodynamics in Packed Beds of Spherical Particles
One of the most common chemical reactors is the packed bed, in which the fluid phase flows through a fixed bed of solid particles. In many applications the solid particles are catalyst pellets. This reactor is usually operated with a wide range of conditions, and its performance is strictly related to the flow fields of the fluid. Before the introduction of tomography and MRI, only overall properties of a packed bed could be determined, such as the fluid flow rates, pressure drop, liquid holdup, wetting efficiency, axial dispersion coefficient and residence time distribution. Even though global properties of the hydrodynamics are useful, they are unable to provide a complete overview of the performance of the reactor, since the inhomogeneity of the packing structure causes significant local variations in holdup, wetting and fluid velocities.
Apart from imaging the fluid distribution in the bed, MRI allows non-invasive measurements of the local fluid velocities. These data can be used to check for the accuracy and appropriateness of assumptions used in numerical modelling of packed beds.
In the present investigation, experimental results obtained with this technique, applied to the study of structure and hydrodynamics in packed beds of spherical particles are shown and compared with CFD simulations performed with an in-house numerical code based on an Immersed Boundary Model-Direct Numerical Simulation (IBM-DNS) approach.
Pressure drop and radial profiles (averaged along the axial direction) of porosity and axial velocity of the fluid were determined, by comparing experiments with simulations, for packed beds of spherical particles with different sizes.
Particular attention was paid to the values of experimental results when directly encoding velocity by means of MRI, since the overall material balance, which relates the average bed porosity with superficial and axial velocity of the fluid, must always be satisfied.
Structure, in terms of phase fractions and, therefore, local porosity, was reconstructed through an algorithm able to detect spheres in a stack of slices. Experimental 3-D datasets of the packed beds have been obtained with static MRI measurements, whereas the axial velocity of water flowing through the particles was obtained by directly encoding velocity values.
From the axial velocity map, a PDF of experimental velocity values was also obtained for each sphere size and flow condition, showing good agreement with the overall material balance involving the relationship among average porosity, superficial and average axial velocity of the fluid.
Since the IBM-DNS method was developed and used to perform CFD simulations of transport phenomena in packed beds only in recent years, this work tried to fill the gap related to validation of the numerical approach, by means of a one-to-one comparison with detailed 3-D datasets obtained through MRI.