We are trying to simulate gas solid flow in a certain geometry. We are having serious problems with convergence.

As a result, we reverted to trying a very simple case of solid and gas flowing down a vertical pipe. Still convergence is not easy at all. We have tried a combination of boundary conditions, velocity inelt/outflow; velocity inlet/pressure condition at the outflow, pressure/pressure B.C. We managed to get convergence at solids fraction of 10%. However, such a simple problem cannot converge at higher solids volume fraction.

We are using Eulerian-Eulerian model.

Is there anyone solving gas-solid flow problems? Can some one shed some light on this problem using FLUENT?

Thank you

I'm working on the Eulerian granualar model too. I obtained satisfactory results on some risers with a solid volume fraction of 0.2 (20%).

I used velocity inlet and outflow boundary conditions.

I also met some convergence difficulties at the beginning, but these were related to solver settings.

I need more details of your problem to give precise information (sorry for all questions):

- Are you using the kinetic theory to model the granular flow? If yes, which correlation for viscosity?

- Which drag factor are you using?

- Which values of under-relaxation factor are you using?

- What kind of grid and what discretization scheme?

- If you model turbulence, how do you do it?

Hi

ap

-Yes we are using kinetic theory; gidaspow correlation is used for viscosity.

-We are trying all the available correlation.

-0.1 is the under-relaxation factor for all .

-Tetrahedral mesh is used

-The flow is laminar.

Our case is a steady state, solid and gas are both coming to the pipe inlet and leaving from the outlet.

I'm wondering if the gas/solid flow in steady state conditions is possible to be solved by available codes.

Thank's

Gas-solid flows are not steady. The presence of the gas phase causes fluctuations in the flow, and particles form clusters. So, you should simulate your system using unsteady calculation. In the literature usually a time step between 5x10^-4 and 10^-5 is used (Gidaspow (1992) - Arastoopour (2000)).

- Under-relaxation factors seem too low to me: try 0.6 for pressure, 0.4 for momentum and leave 0.2 (or try to increase to 0.4) granular temperature and volume fraction.

- Make sure your grid is fine enough (If possibile, switch to hexa). Use the second order upwind (or higher) discretization scheme for convective terms. Do not use symmetry boundary conditions: it has been demostrated that these conditions cause unphysical cluster formation.

- I obtained results in better agreement with experimental data using Syamlal O'Brien correlation for both drag and viscosity.

Hope this helps

Hi

ap

ap

Thanks for your inputs. I'm using them and every thing is working fine,up to now.

I have concern regarding the computational time, from your experience, in general how long it will take to get some results.

Well, it's strictly related to the dimension of your computational domain and on your computational resources.

However, on a P4, you should obtain results in a few days.

In one of my cases I simulated a system with a grid formed by about 35.000 nodes (hexa grid), for 60 seconds in about a week on a 2.5 GHz P4 workstation.

Hi

ap

ap,

I really appreciate your help.

AAK

You're welcome

ap

[pandaba]

Hi ap,

I am also working on gas solid flow. Your post helped me in convergence . I am getting the right profile for gas velocity variation.But for solids , the velocity plot shows zero value .


I am not able to find out the cause. Please help me.

Regards
Pandab