Hi all,

Currently I am studying behavior of a large air bubble (Taylor Bubble) rising in a vertical channel filled with stagnant lwater. For this, I am using CFX 10.0. I have chosen inhomogeneous model for a transient simulation. Altough bubble does rise, its boundary (air-water interface) becomes blurred. I think this issue is due numerical diffusion in volumetric fraction at interface, so an interface reconstructing tracking algorithm (not available in CFX) could be needed.

Do you have any suggestion for this problem?

Thanks for your help.

Miguel

Hi Miguel,

Actually, i want to ask u a doubt regd a similar problem as urs. I am similating water sprinkler. I have seen in cfx help manual, to consider inhomogeneous model to simulate Air bubbles in water column & water drops in Air domain. But when i take in to consideration the inhomogeneous model the solver terminates. The physics of the problem is inlet water velocity[inlet of sprinkler] & opening static pressure 0 [for air domain(cube)]. The same problem when i have given run with homogeneous model,i am able to see the stream lines coming out like a jet but the volume fraction of water in the air domain is very much less. I feel that simulating this problem with inhomogeneous model can give a better result. Pls help me if u find a solution 2 this problem.

Hi,

When you run the problem using inhomogeneous model the solver terminates, with errors?

I have read a paper titled NUMERICAL SIMULATION OF INTERNAL FLOW IN A LARGE-SCALE PRESSURE-SWIRL ATOMIZER, by Madsen, Hjertager & Solberg. I think they resolved a problem like yours, using both VOF (volume of fluid method available in Fluent, but not in CFX 10.0, I don't know if it's available in your CFX relase), and Two Fluids method (inhomogeneous)available (I know) from CFX 4.4 until last release, getting good agreetment with experimental data by both ways. This suggests that inhomogeneous model could models the problem without issues.

Are you taking in consideration surface tension model? In my bubble problem, interface becomes blurred when i check this model but doesn't when i don't. I don`t know why yet, maybe becouse continuum surface model (Brackbill) used in CFX induces numerical diffusion in volume fraction. I'm studying this.

I hope this would be useful for your work, i write you for any advance in this problem.

Regards

Miguel

Hi, For your blurred interface problem i think you can go for mesh adaption.!!!! I have done a free surface problem comparing with mesh adaption & without mesh adaption. The coclusion is interface is blurred for run without mesh adaption. Regards Sridevi