Hello everyone (and especially the multi-phase experts out there),

this isn't cutting edge CFD, but i thought that perhaps i can start a discussion thread on how best to model this problem. Here are the conditions:

1) Water and air multiphase. 2) Flow is through a pipeline with vents at the top where the air can escape 3) Air consitutes a small fraction (about say 2.5%) of total mass. 4) Physically, there are different air bubble sizes (maybe about 5 discrete sizes) 5) Physically, air is observed to accumulate in between vents and move (slowly) downstream to the nearest vent.

Looking for explanations as to which multiphase to use (eg. ASM, VOF, Lagrangian, MUSIG) and what to look out for when using these models (drag forces, surface tension, etc.)Any comment is greatly appreciated. Thanks!

This really depends on the flow state of your water and air.

- Is the pipe horizontal? Then water will lie on the bottom and slugs might exists. - Is the pipe vertical? Then your flow might more look like a bubble column, which can have different forms. - Perhaps you have a mixed geometry. Then you might have mixed forms.

I think Langrangian is not a good option here but need Eulerian-Eurlerian. Depending on the flow form, you can use a mixture model or a disperse particle (water) model. Use Musig in a later stage but only if you know your droplet size.

I think drag forces are most important. The question is which model you need. Perhaps you have to use a user-defined one. Surfcace tension is only important when you dimensions are small. I hope you don't need surface tension as it is implemented explicitly.

Start simple, increase complexity.

Gert-Jan

www.bunova.nl

Thank you Gert-Jan,

in response 1) yes, pipe is horizontal (diameter about 2.4m). There is actually a very slight gradient downwards of 1:50. 2) flow is expected to be in the region of plug/slug flow to bubbly flow. 3) velocities are not excessively high, 3m/s at most. CFX tech support has discouraged the use of musig because according to them musig does not converge well for the slow moving or stationary air pockets. Seems like i am answering my questions here, but i would just like to find out how academics and expert CFX users would approach this problem.

First approach would be to use Eulerian-Eulerian approach with a mixture model of two continuous phases, mixture length scale of 1 mm, and free surface sharpening.

Gert-Jan