Multiphase air-water filling problem
 

Hi everybody,
I'm facing some strange results in a filling problem where a cylindrical tank full of air is filled with water.
Water enters the tank flowing on the lateral walls and everything is fine with a well defined interface.
But as it reaches the bottom of the tank the interface becomes really thick and it seems some air in entrained below the thick surface as the tank is filled.
I used both homogeneous and in-homogeneous (better results but not satisfying with air as dispersed phase mean diameter 0.0001 mm) on a very refined 3d sample grid.

Any suggestion?
Thanks in advance

Hi,

This area is much improved in V12. Are you using V12? The new coupled VF solver should work well here (I did lots of beta testing on this feature and I think we ironed out all the bugs! For your information the issue was the coupled VF solver in V11 did a poor job of conserving volume fraction.)

Are you modelling surface tension? This makes a big difference.

Glenn Horrocks

Thanks Glenn for replying.
I really spent a lot of time in reading throughout the forum and I know you have gained a lot of pratice and experience in multiphase problems - really appreciated you ansewred me (i.e. you suggested in a previous thread SST instead of k-e and I agree with this choice).

I'm trying to solve this (apperantly to me) simple problem in conjunction with CFX techs with CFX 12. I'm currently taking account of surface tension with a value of 7.2e-2 [N m^-1].
To explain better what is happening: very thin film of water on the lateral walls collide on the bottom of a tank. As the filling process go on, when displaying water fraction, under the interface the color is not uniform and water is displayed as "a flame".
Buoyancy activated.
Any other suggestion is appreciated.

Thanks

Hi,

Set the volume fraction smoothing to none, use adaptive timestepping homing into 3-6 iterations per timestep.

CFX is poor at this type of analysis, it is very slow. I got a speedup of 30x by switching to Fluent for free surface simulations with surface tension.

Glenn Horrocks

Hi Glenn,
Thanks for your suggestion: I'll try and report about this.
I wouldn't expect these differences with Fluent since CFX coupled solver usually provide less (but longer) iterations to get to a converged solution.
Please correct me if I'm wrong: VOF's Fluent is equivalent to Homogeneous in CFX?
I am not really sure if I should use Homogeneous or non homogeneous for my problem...I got a really nice solution for homogenous with open bottom to use as initialized solution with a well defined film on the walls but as I close the bottom I got the "flame" under the interface; for non-homogeneous the film on the walls are not so well defined...

Thanks

Hi,

Yes, VOF in Fluent is equivalent to Homogeneous multiphase in CFX.

Use Homogeneous if you want the interface to remain sharp. Use inhomogeneous if some mixing (eg foam, bubbly regions) occur.

Glenn Horrocks