Dear all, I am trying to simulate a simple bubble column where i am using a degassing boundary condition to simulate the bubbles escaping from the liquid domain. While bubbles are escaping at this free surface, the liquid is recirculating. Simply my degassing boundary condition just make sure that bubbles are corssing the free surface, and make sure that the liquid see the free surface as a wall to allow th recirculation. What I need is to add in my programm is the turbulence boundary condition at this free surface ? According to my simulation, due to bubbles, the liquid behaves like a jet impinging on a wall. Just like in a jet case my std k-eps gives me a peak of turbulent production where there is not. Unfortunately i do not have a k-omega turbulent model for multiphase flow to base the turbulent calculation rather on vorticity.

Is someone can help to fix my turbulent boundary condition at this free surface using a tubulent model such as:

Standard k-eps : need BC for k and eps for the liquid? Algebraic model : Need BC for uu, vv, ww for the water?

Thanks in advance Thomas

Dear Thomas,

why don't you treat the freesurface as it was a symmetry plane for the carrier (fluid) phase and not a wall. The conditions for k and epsilon are than quite easy to set up.

Why should you need conditions for uu, vv and ww. You are using a two-equation model and not a second order moment closure?

The Becker column has become a benchmark for the bubble column simulation. There you can find several suggestions for your problem.

regards,

Ridwan.

Thanks for answering. I don't see how a symmetry plane for the carrier fluid can be applied the a free surface the case of a bubble column. About uu, vv, ww I am talking about an algebraic model such as the Reynolds stress model of Fluent. So you would easily understand I need in addition of k and eps, a boundary condition on uu, vv, ww. I have not find any suggestions to my problem in the Pfleger and becker publication. Same thing in other publication. So in fact i am still wondering, in the case where a range of faces represent a free surface(degassing BC), what should be the correct turbulence boundary conditions to this faces ?. Thanks thomas

Hi again Thomas,

that's interesting. I did not know that Fluent has implemented an algebraic stress model for bubbly kind of flows. Also for the Two-Fluid formulation? I don't keep track anymore on the development of commercial packages.

I have used in the past the symmetry condition for the following reason. What are the conditions for a symmetry plane? The shear stress is zero right? -> thus dv/dn = 0. It is reasonable to assume that the air above the liquid will exert an neglible shear force on the liquid surface.

Once the gas exits the free surface out of the computational domain, the bubbles leave will have a so-called "terminal" velocity (which might be close to the slip velocity --> in the order of 20-25 cm/s (for spherical bubbles in the order of 3-4 mm)

I have used the above approach for my own computation.

Check out

http://www.sciencedirect.com/science...&_coverDate=07%2F31%2F2003&_sk=999509992&view=c&wchp=dGLbVtz-zSkWz&_acct=C000024500&_version=1&_userid=499885&m d5=7901d880693d3f76fbf814324e0ac378&ie=f.pdf

for further details.

regarding uu, vv and ww: with the above assumption, i.e. symmetry conditions, you might assume that uv= vw = wu = 0 (zero shear stress) and than k = 2(uu+vv+ww)/3, where uu = vv = ww. At least you have a first approximation. You more or less would assume local isotropy.Hope this will help you to find more "structures" in your flow with the ASM.

regards,

Ridwan,

Hello, For more informations, the version 6.2.x of Fluent will contain an ASM model for multiphase flow. Regarding the way you set you free surface condition i totally agree with you for the reason i did exactly the same assumptions you did but using a velocity inlet BC. Why do I want to know if it exist some REAL turbulent condition at a free surface ? 2 reasons: In some case it is interesting to know what is the shear-stress implies by the liquid recirculation at the free surface. We can find an example in the steel industry when a thin layer over the free surface protects the steel below from oxydation. In that case i do not think the assumption we take for the turbulence at the free surface allows us to get the limits where the thin layer is gonna be destroyed. Also, in the case of a short bubble column locally aerated at the center. If we set a free surface the way we did, it appears for certain flow regime a peak of turbulence production at the free surface on the column axis. Indeed, this is due to a stagnation point by the liquid re-circulating at the free surface -> Just like the case of an impinging jet on a plane wall I do not think it is realistic. I can easily get read of that problem using an k-eps RNG or a k-w model but i do want to use a standard k-eps. That is why i wonder if accurate turbulent boundary condition could not help to have more realistic turbulent production values at the free surface. We also thought about coupling DPM and VOF model to see what is going on at this free surface. But apparently there is not a specific treatment at the free surface for turbulent flow condition using the VOF. Recently someone posted a message on CFD-online saying that the VOF of fluent is not a real VOF model -> i will investigate to see if that is relevant with my problem.

About your paper, i do not have an account to access to science direct or any scientific publisher. Is there a way for you to send it to me by e-mail, or to post it on a FTP website. I would glad to read it and know from your work.

Cheers thomas

Thomas,

Sorry for my late reply, I was away from work. But the work I was talking about you can find in the Aiche journal.

AIChe volume 7, 2003, Sensitivity Study on Interfacial Closure Laws in Two-Fluid Bubbly Flow Simulations, Pages 1621-1636 R. S. Oey, R. F. Mudde and H. E. A. van den Akker

At present I am not working in CFD anymore, but if you would like to know more about the paper contact R.F. Mudde. probably he can sent you a hard copy also.

his e-mail is: r.f.mudde@klft.tn.tudelft.nl

Good luck with it,

Ridwan.

[cincin1234]

hello, Thamos, i am doing the same kind work as you do, i am adjust standard k-epsilon model to apply to free surface turbulence. how have you implemented in your work?