resolved vs. unresolved CFDEM

Submitted by evansmuts on Tue, 06/12/2012 - 16:01

Hi

I was wondering what the difference is between "resolved" and "unresolved" CFD-DEM? If I am correct, the solver "cfdemSolverIB" uses resolved CFD-DEM with the "immersed boundary" method.

One obvious benefit is that a finer CFD mesh can be used (smaller than the particle). This would give more accurate flow calculations, but with greater computational expense I am guessing. Is there extra computational expense, other than the extra CFD cells?

What other advantages/disadvantages are there of resolved CFD-DEM? How much smaller than the particle do the cells have to be? Would the coupling interval have to be very small (1-1 even)?

Is there a good book/paper that explains the immersed boundary method?

Any ideas/comments would be appreciated. Thanks.
Evan

Forums:

alice's picture

alice | Tue, 06/12/2012 - 16:48

Hallo Evan,
you are right: cfdemSolverIB is a resolved CFD-DEM solver. It uses an immersed boundary or fictitious domain method. This class of methods was initially designed for depicting the interaction of fluid flow and elastic boundaries, and was lateron generalized for problems involving rigid bodies as well. Literature provides a number of papers dealing with fictitious domain and immersed boundary methods (e.g. Haeri, Shrimpton, 2012, "On the application of immersed boundary, fictitious domain and body-conformal mesh methods to many particle multiphase flows", just to name one).
The bodies located in the domain are assumed to be "large", as you already pointed out. In our case this means that at least 8 CFD cells per particle diameter are necessary for getting accurate results. The method is well suited for tackling problems involving few large particles. For these cases dynamic local mesh refinement speeds up the computations.
So usually you will not be confronted with the question of whether to use resolved or unresolved CFD-DEM, their field of application differs...
The coupling interval depends on the very problem you are looking at, usually the DEM step-width is finer than the CFD-step width, which might leave you with invervals of e.g. 10 - 1.
Cheers,
Alice

evansmuts | Thu, 06/21/2012 - 15:58

Hi Alice

Thanks for all that information, it was interesting to read. It looks like this approach will not be suitable for my current problem, but I will definitely do some more reading on the method.

Thanks
Evan