(11d) Investigation of Layer-Inversion and Mixing in Binary-Solid Liquid-Fluidized Beds Using a Combined Continuum and Discrete Model

It is widely recognised that binary systems consisting of two distinct particulate components, differing only in size or density, have a tendency to segregate when fluidized. The extent to which segregation occurs depends on the difference between the two species' particle diameters or densities ? if this is sufficiently large the bed can form two distinct layers; alternatively, if the size/density difference is small a mixing zone will exist between the monocomponent layers.

However, for the special case of a binary mixture where the two species differ in both particle size and density such that the larger particles are also the less dense, it is possible for the system to exhibit a very interesting phenomenon ? layer inversion. This occurs as follows: when a suitable system is fluidized at low liquid velocities, the bed is segregated with the larger, less dense particles forming a layer above the smaller, denser species. If the liquid velocity is increased through a critical value the two layers will invert, such that the larger, less dense particles occupy the lower part of the bed.

As well as being of academic interest, the layer inversion phenomenon also has practical significance, since the critical inversion velocity corresponds to the point at which the system is completely mixed. While such binary fluidized beds are relatively simple systems, their study can nonetheless provide useful information about the mixing and segregation behaviour of the more complex polydisperse systems that are operated in industrial fluidized bed units.

Previous studies have examined the layer inversion problem via experimental methods or using simulation techniques that are based on a continuum-model approach, and various semi-empirical models for predicting the critical inversion velocity have been proposed, but to gain a fundamental understanding of the mechanisms governing this phenomenon it will be necessary to understand the complex microscale behaviour of the particles and the fluid in these systems. Although such information is difficult to obtain via experimental methods, the increasing availability of relatively inexpensive computing power makes it possible to investigate the layer inversion phenomenon using computer simulation techniques.

In this work we examine the layer inversion phenomenon in liquid fluidized beds using the Combined Continuum and Discrete Model (CCDM), a computational technique for the simulation of multiphase particle-flow systems. CCDM employs the Discrete Element Model (DEM) to simulate the motion of the individual particles, while the continuum fluid flow is solved via a Computational Fluid Dynamics (CFD) approach. The key feature of CCDM that ensures a rigorous solution to the multiphase flow problem is that the DEM and CFD models are coupled via careful application of Newton's 3rd law of motion to the fluid-particle interaction forces, which in the liquid medium include added-mass, spin-lift, and pressure-gradient forces in addition to the steady drag force. Application of the technique to liquid-solid systems has also required modification of the traditional DEM contact mechanics to take account of the lubrication effect of the viscous interstitial fluid on particle collisions.

The ability of CCDM to track the motion of individual particles makes it a powerful technique for examining problems such as mixing and layer inversion in fluidized beds. A suitable bidisperse system of particles was selected based on information from the literature. By simulating the fluidization of this system at a range of superficial liquid velocities it has been possible to observe varying degrees of segregation and mixing, and to demonstrate that the system can be operated stably above and below the inversion point.