[vicder]

Hi guys,

Recently, I'm working on my research about cyclone separator and trying to use multiphase model to simulate the flow of air and sand within cyclone separator.

In my simulation, I define new material called "sand fully coupled". The option of morphology is "particle transport solid" and the specific diameter is equal to 0.0001m. Both of the type of boundary condition at inlet and outlet is "Opening" and the value of opening pres. and dirn. is equal to 0 Pa and -1000Pa. Besides, the value of particle mass flow at inlet is equal to 0.01kg/s. However, the simulation still can't converge. How can I solve the problem?

The attachment is CCL file and converge curve. Thanks for help.

[ghorrocks]

FAQ: https://www.cfd-online.com/Wiki/Ansy...gence_criteria

There is a lot of literature on modelling cyclones. Depending on what turbulence model you are using, it is highly likely it does not have a steady state answer and therefore requires a transient model to converge.

[juliom]

My Master degree thesis was on vertical separator and I realized that this problem are quite challenging. Firstly I see your problem is not well posed. You are imposing pressure at inlet and outlet. That won't work out. Since this is incompressible you must prescribe velocity or mass flow somewhere. Also, you have a third nozzle that you must treat wisely. On the other hand the turbulence model. In principle, only RSM can deal with the physics due to the rotation that the delator imposed on the continuum medium. Secondly, the system is too stiff and you might use a transient approach. Finally the mesh. Ohhhhh
The mesh...
well I hope this helps. To conclude, I use Eulerian-lagrangian mode. I guess you did too.

[juliom]

My Master degree thesis was on vertical separator and I realized that this problem are quite challenging. Firstly I see your problem is not well posed. You are imposing pressure at inlet and outlet. That won't work out. Since this is incompressible you must prescribe velocity or mass flow somewhere. Also, you have a third nozzle that you must treat wisely. On the other hand the turbulence model. In principle, only RSM can deal with the physics due to the rotation that the delator imposed on the continuum medium. Secondly, the system is too stiff and you might use a transient approach. Finally the mesh. Ohhhhh
The mesh...
well I hope this helps. To conclude, I use Eulerian-lagrangian mode. I guess you did too.

[vicder]

To ghorrocks,

Thanks for your reply. I'll read the data. Thanks.

To juliom,

Thanks for your reply. I use Eulerian-lagrangian model, too. Besides, I have tried to use mass flow at inlet and pressure at outlet. The convergence curve is really more stable than previous one. However, it doesn't converge, either. Well, I'll keep going on it.

[juliom]

The fact that is "more stable" doesn't mean is the right solution. Don't let the solution mislead you...

[vicder]

To juliom,

OK, I see. Thanks for your kind reminder.