[deada]

Hi everyone,

my medium-term goal is to analyse the transient stirring process containing a fluid phase dispersed solids. Since this is my first analyse of both moving meshs and multiphase flows i started "step-by-step". My first little success was that i got a single phase version of the problem running. This means, using periodic domain interphases, i simulated a quarter of the geometry with a stationary tank and a rotating domain containing one blade of the four-blade impeller with just the fluid as the only phase.

In a second step i tried to add an additional eulerian phase for my dispersed solids. I do suspect my problems to arrive from the fact that the particles i try to simulate have a diameter of around 0.01 m in a tank with a diameter of 0.18 m which means the mean particle size exceeds the cell size (base size of 0.0002m) by magnitudes. In the solver modeling and solver theory guide i do not find any remarks on a limitation of particle size, but when i set the mean particle diameter to 0.1µm the simulation works.

Is there any source on how to model multiphase flow for very large particles compared to cell size? Maybe i should also mentioned that in the end i want to have several (around 5) different mean particle diameters, so i need a phase for each of thoses diameters with source terms for the exchange between those phases.

Even though I didn't find a perfectly fitting example, I start to think that I have to go the Eulerian-lagrangian way. I'm also interested in the forces on the particle from wall-particle and particle-particle collisions, is that possible in one of the two approaches? I found sources where a detour using EDEM to model discrete elements was made. Is that maybe my way to go?

I know that is quite a lot help to ask for but every little advice is appreciated! Thanks in advance!

Kind regards

[ghorrocks]

The eularian model assumes that the particle size is small relative to the mesh size, so if your particle is much bigger than the mesh the model is not appropriate. The model will still probably run, but the effect of the particle will be applied as if it is a continuum of many particles rather than a small number of large particles.

Note the lagrangian model in CFX will not help you much either. It also assumes the particle can be assumed as a point and is small compared to the mesh size. Again it will probably run but the underlying assumptions are not very suitable for your flow.

With particles as large as you describe I suspect you will have particle collisions with other particles, the walls and the impeller. CFX does not have suitable models for these collisions. Also the distortion of the flow field due to the flow passing over a particle will affect nearby particles if the volume fraction is high enough - if this happens CFX does not have good models built in for this with multiphase models.

I think your best bet is to use discrete element models like EDEM or Rocky. They are best used when you want to know collision forces in situations like this.

[deada]

Thank you for the very quick reply! It confirms my suspicions.
Maybe I should move to the Fluent forum section with this question, but I'll give it a try: I read that EDEM has a EDEM-Fluent interface. Do you by any chance know if this is can be used to accurately simulate the two-way-coupling between fluid and solid phase even for such large particles?

[ghorrocks]

I think Fluent has a simple built in DEM model. It may be adequate for your requirements - have a look at it.

I understand EDEM has couplings to both Fluent and CFX so if you chose to do it this way you can choose which CFD solver you like.

[cjw666]

Hello, everyone. When I used cfd coupled fluent to separate the hydrocyclone solid-liquid two-phase flow, the light phase particles in edem were expelled from the inner wall of the hydrocyclone to the bottom flow port (under normal circumstances, the light phase should be excluded from the overflow port), but the separation effect did not play a role. Increasing the speed in fluent and edem did not work either. What is the cause of this?

[ghorrocks]

There are so many things it could be you could write an encyclopedia on it. But step one would be to put your post on the Fluent forum as this is the CFX forum.