(249b) Coarse Discrete Element Simulations for Gas-Solids Flows

Discrete element model (DEM) coupled with the gas-phase continuum solvers are computationally very expensive to perform simulations of realistic devices. In this paper, we present an alternative approach inspired by the recent success of MP-PIC (Multiphase Particle-in-cell) methods to improve the computational efficiency by conducting DEM on particle clouds rather than individual particles. There are many open questions about strategies for good approximations for the coarse-grained systems and we will be exploring and evaluating some of those possibilities in this paper along with comparing the methods with existing continuum simulation capability.

In the coarse DEM simulation, each particle represents N number of original particles at close packing and these coarse particles are subjected to collisions etc. using soft-sphere spring-dashpot DEM model and the drag computations are based on the original particle size and density. The major assumption into this model is that the

internal collisions within the parcel do not have any first-order effects and the parcels are homogeneous. The pros of this approach are:

- Drastically reduce the computational costs

- Includes both normal and tangential forces

- Address close-packing naturally

- Account for particle rotation

- Superior solids advection through the Lagrangian tracking of the particles

- Use existing Eulerian-Lagrangian Framework

- Can be extended easily to multiple particle sizes

Some of the cons of this approach are:

- No mathematical framework to bound the error from the approximation

- There is no way A priori to judge the sampling accuracy and efficiency

In this paper we present what-if numerical experiments of a fluidized bed with a central jet using our Coarse DEM model. We find that the these simulations are able to capture the two dominant modes in the power spectrum density of pressure while the bed height variation seems to have the correct qualitative behavior. Comparing these results to comparable continuum-continuum simulations shows that the ability to convect the particles accurately might have a big influence on the ability of continuum-discrete to have better predictability. We will report on the role of the parameter space (in terms of numerical approximations) on the differences between the various simulations and make recommendations on the acceptable coarsening strategies. We hope that this work would lay down some groundwork for more rigorous coarsening strategies for discrete-continuum simulations for gas-solids flows and improve drastically the efficiency of current techniques.