twoPhaseEulerFoam - sudden enlargement of circular pipe validation case
 

Hi all,

In Rusche's thesis, he validated the two-fluid method with the sudden enlargement of circular pipe case. Could anyone send me a copy of this original experimental paper? I can't find it through google or our library.

Bel Fdhila R, Simonin O. Eulerian prediction of a turbulent bubbly ow downstream of a sudden pipe expansion. In Proceedings of the 6th Workshop on Two Phase Flow Predictions, Sommerfeld M (ed), University of Erlangen, 1992)

I followed whatever is available in Rusche's thesis, but I got a high gas fraction zone right after the enlargement (see attachment).

Thanks,
yanxiang

Are you using the latest 2.1.x code with MULES?

Nope. This is OF211. Do you think it would be different if I I use 2.1.x code? Nevertheless, I think, OF211 shouldn't behave like that either.

The MULES version of twoPhaseEulerFoam should do a much better job at ensuring that alpha is bounded, which is less robust in the old version. Maybe give it a try :-)

Absolutely. I gave it a try and the result is attached. Gas still tends to go to the corner. The velocity field looks fine though. I have also attached the case tar ball for your review.

Thanks,
yanxiang

Sorry for hijacking this thread.But I really need you help,Prof.Alberto.
Much details has been depicted here
http://www.cfd-online.com/Forums/ope...ured-mesh.html

Look forward to you replay.Thanks in advance.

I don't see problems in the setup of your case. The accumulation of gas in the corner does not seem too different from Rusche's results (Fig. 3.3 c, page 131).

It is possible you have less numerical diffusion than in those results (it seems so from your picture). You should run with the same schemes to have an apple to apple comparison.

Best,

I commented on the other thread.

Hi Alberto,

I tried the same schemes as used by Rusche, but the results stay pretty much unchanged, with a high gas zone in the corner, and this is not there in Rusche's results, or in Oliveira's (Int. J. Numer. Meth. Fluids 2003; 43:1177–1198). I played around with the BC's, but still couldn't get rid of that. Also, sadly, I was not able to find any of the references (except for Oliveira's) used by Rusche in that particular test case section where others validated their models with it. Do you have any of those papers?

Thanks,
Yanxiang

P.S: Happy New Year!!!!

How did you decide the BC's for the turbulent quantities? They aren't provided in the article...

I ran the case you attached turning turbulence off, just to see the effect, and segregation is much lower, as expected.

Anyways, I am working on a validation case for twoPhaseEulerFoam/multiphaseEulerFoam for bubbly flows... we'll see :-)

Alberto,

So after reading your question about the turbulent quantities, I made some changes to my very original test case at work, and ran the OF211 version of tPEF on it. Weirdly, it worked :-$. By that I mean, I got similar results to those in Rusche's and Oliveira's works. I will try to reproduce that with OF21x version when I get home (although a diff on the two cases didn't give me any hints why the results would differ). Anyways, I attached the screenshots and case for OF211.

Best,
yanxiang

There is a significant difference in the two algorithms used to solve for alpha in twoPhaseEulerFoam before and after the introduction of MULES, which might explain the differences. I am running some validation cases exactly to see if accuracy was maintained (for now, it is, at least in the case of a simple bubble column).

I am eager to see validation cases for twoPhaseEulerFoam on tet mesh~:rolleyes:

I am not considering tetrahedral grids, because my geometry does not require them (it is a box with rectangular section!). I consider this test-case because it is the precursor for new model developments.

If you want to perform a verification and validation study on tetrahedral meshes, however, it should not be that hard. You should consider a simple geometry with well-defined results (there are many on bubble columns in the literature, pick a case where simulation results are also available, and check the model is the same), do a grid-independence study, and then do the experimental validation on the converged grid.

P.S. Have you tried to use more appropriate numerical schemes, as I suggested in the other thread, with a good quality mesh?

I am sorry about that question. because it looks like there is no substantial change to the result.but I dont know if its still because of the mesh.


So I refine the mesh, now the cell is half than that cell in the other thread.so the total cell number reachs to 1,79 million,its too many, so I shorten the height, the other thing is not changed.the total cells number is about 55000.

I attached my image.I think that dont need to depict it.My primary language is not english so~.

btw,I turn to the other thread,I am sorry for that.
http://www.cfd-online.com/Forums/ope...ured-mesh.html

Thanks for you consistent attention.Alberto

Hmmm... interesting. I tried the OF21x version on the same case again, and as you would expect, the results just look like what we had previously. So does that mean the new solver with the MULES method is not solving the alpha equations correctly?

Thanks,
yanxiang

I think it's an ancient question about 'twoPhaseEulerFoam' :).
Actually, I'm studying the structure of multi-phase (liquid-gas) solver and had a chance to analyze 'twoPhaseEulerFoam' of version 211 and 21x.

There are two conclusions:
1) The code structure of version 21x follows that of version 22* series (ex. 220, 221, ..), so the notation of phases (phase 'a' ->'1' , phase 'b' -> '2').

2) The difference of results for those two codes (version211 and version 21x) is derived from the "UEqns.H" (they use different value "Rc1" and "Rc2".) not from the function of 'MULES' !!!! (You can get the same results with version211 from the version21x by changing the values of Rc1 and Rc2!!!)

I hope that this answer will help the future guys :)
Thx.