Hello,

I am simulating a baffled mixing tank in which silica (dispersed solid phase) and water (continuous liquid phase) are mixing. Two domains are set up and one is for stationary domain (containing baffles) and one for rotatioinary domain (containing six vertical blades). Fluid-Fluid Frozen Rotator GGI is used for the domain interface.

When I check the vf and velocity of silica and water, there is a sharp interface between two domains, just like a wall. This is not expected. Anybody can give me some suggestion about how to imporve that?

Thanks in advance.

Jeniffer

Hi Jeniffer,

I guess the first thing to check is if you really have a solution. Is it well converged in all respects? I also assume these are velocities in the same frame where you see the discontinuity. There will be a discontinuity if you look at rotating frame velocity in the rotating piece. Do you have a nice grid, what advection scheme are you using?

There are lots of unkowns here?

Regards,

Bak_Flow

Hi Jeniffer

Make sure you are looking at the variable Velocity in Stationary Frame, since the default Velocities are in the local frames (one of which will be rotating).

Johnson

Hi Bak_flow and Johnson,

Thank you for your response. I checked the velocity using velocity at Stn Frame and still got a sharp interface between two domains. Also, if I check the volume fraction of both phases, I got the same thing as velocity.

The grid I used is relatively coarse to save CPU and memory. After 1000 steps steady-state simulation, I got RMS Res= 3E-3 for Mass of both liquid and solid phases. Max Res = 5.0E-2.

Imbalance: Mass-Silica in Baffle domain =49.3770 %;

Imbalance: Mass-Water in Baffle domain = -48.0324 %;

Imbalance: P-Vol-Baffle = 49.5208 %

Imbalance: Mass-Silica-Impeller = 2.2178 %

Imbalance: Mass-water-Impeller = -2.4741 %

Imbalance: P-Vol in Impeller is 1.6421 %;

I know the solution is not converged completely. But I am suspecting something else is the cause of this but I can not tell what that is. Should I refine the grid near the domain interface just like a wall?

Thanks,

Jeniffer

Hi Jenniffer,

thanks for more detailed information. Unfortunatly this is not anywhere near converged. I have pretty tough standards but like to see Max resiusals below 1E-4. Imballances below .01%!

You will have to run your solution longer and possibly with a different time step to really pound this in!

Keep working at it and let us know.

Regards,

Bak_flow

Hello,

Something wrong with my simulation. I run 3000 timesteps and the RMS curves remains in the range of 10^-2 to 10^-3.

Jeniffer