The coupling of computational fluid dynamics (CFD) and the discrete element method (DEM) with the aid of coarse-graining methods to CFD-CGDEM is a relatively new efficient numerical method for the simulation of combined particle/fluid flows. Based on DEM, a freely selectable number of particles are combined into one representative parcel each. This significantly reduces the number of particles to be tracked, which shortens the computation time accordingly and makes the method interesting for the calculation of industry-relevant problems. As in DEM, the trajectories of the parcels are calculated by solving Newton's laws using appropriate contact and fluid interaction models. The particle-particle, particle-wall, and particle-fluid interaction forces must be scaled for conservation of energy and similarity theory considerations, respectively. A variety of scaling rules have been developed in the past, although a systematic comparison of these has not yet taken place.
This DFG project will compare the existing scaling approaches and investigate which of the approaches provides accurate results with increasing scaling factors, which have caused difficulties so far. First preliminary investigations have shown that the ratio of characteristic reactor dimension to parcel diameter should not fall below a critical value. With the help of an adaptive coarse-graining approach, which allows different scaling factors to be used in geometrically narrow and wide areas, an exemplary investigation is being carried out to determine the extent to which such a procedure allows CFD-CGDEM to be used in technically important reactors with geometrically narrow internals.
With DEM, there is a requirement that the grid size must be larger than the parcel diameter. Therefore, coarse-graining leads to a coarsening of the computational grid. It is examined whether methods that remove the restriction regarding the necessary grid size increase the accuracy of the CFD-CGDEM. In addition, it is investigated whether modeling the mesoscale effects by filtered drag models positively affects the quality of results at large scaling factors.
CFD-CGDEM has rarely been used to simulate polydisperse systems. A bi-disperse system will be used to test whether CFD-CGDEM can correctly reproduce the segregation behavior in both the mechanically excited and fluidized systems. At the end of the project, based on the developed methods, a complex industrial-scale use case will be used to answer the question of whether CFD-CGDEM is an efficient and accurate method for simulating industrial-scale particulate systems for both mono- and bi-disperse systems, and in what framework it can be used.