I'm testing on the settlement of a single sphere using cfdemSolverPiso. The terminal velocity predicted by analytical model should be about 0.04 m/s. However, I got a velocity more than 0.1 m/s. I don't know what's wrong with my settings. Coud you please help me? I've attached my files in this post.
The voidfractionModel is bigParticle. The diameter and density of the sphere is 15 mm and 1120 Kg/m^3. The density and dynamic viscosity of the fluid is 970 Kg/m^3 and 373e-3 Ns/m^3.
| Attachment | Size |
|---|---|
 test_settlement.tar_.gz | 97.22 KB |
Nucleophobe | Wed, 01/28/2015 - 18:24
DEM vs. Resolved DEM
Weizhang,
I believe your problem is that your mesh is actually too fine.
DEM requires that the particles be much smaller than your cell size. In the case you attached, the particle is about 4 cells in diameter. Thus, while performing the empirical drag modelling, the relative velocity is calculated erroneously (I think), which causes the particle to effectively drive it's own motion.
By coarsening your mesh, I was able to get a solution that matches the analytical Stokes drag solution. Try changing your blockMeshDict resolution from:
hex (0 1 2 3 4 5 6 7) (20 20 32) simpleGrading (1 1 1) // regular grid
to:
hex (0 1 2 3 4 5 6 7) (5 5 8) simpleGrading (1 1 1) // regular grid
If you would like to simulate the motion of particles that are larger than your mesh cell size, you might try the "resolved" DEM solver, cfdemSolverIB, which uses the ficticious domain method. There are more details on the differences in Dr. Hager's presentation from the 7th OF workshop:
ftp://ftp.heanet.ie/disk1/sourceforge/o/op/openfoam-extend/OpenFOAM_Work...
Good luck! By the way, I am doing this same check in cfdemSolverIB right now. Let me know how things go.
-Nuc