[hester]

Hello everyone,

I'm struggling with simulating a bubble column with twoPhaseEulerFoam in OF 2.3.0.

Air is injected into water, after Deen [1,2]. The dimensions are 0,15m x 0,15m x 0,45m. The simulation was done after Renze [3]. I didn’t include the liquid surface in my simulation. The continuous phase was modelled as turbulent using k-epsilon modell, the dispersed phase was modelled as laminar. But I didn’t get the wanted results. There is now oscillation of the bubble plume. The velocity for air is too big and the velocity of water is too small compared to the values from Deen (at height 0,25m). I think, it looks like there is a problem with the momentum transport between the phases. Drag, virtual mass and lift force were taken into account.

I also tried the mixtureKEpsilon modell, the description looks promising [4]. The results showed the expected bubble plume oscillation. But the results were worse, there’s strange behaviour at the top of my simulation regime (see attached pictures). Is it even possible to use this turbulence modell without including the liquid surface? Nevertheless, I think mixtureKEpsilon is the modell of choice here, I just haven't figured out how to make it work. Or would the standard k-epsilon modell suffice?

I included my case file (with kEpsilon modell). The setup with mixtureKEpsilon is the same. I hope someone could take a look and give me some advice as to where my error is. This is my first multiphase simulation and I've been struggling for a while now.

Thank your for your advice in advance!

hester

[1] N. G. Deen, T. Solberg, and B. H. Hjertager. Numerical simulation of the gas-liquid flow in a sqaure cross-sectioned bubble column. CHISA 14th international congress of chemical and process engineering, Praha, Czech Republic, 2000.
[2] N. G. Deen, T. Solberg, and B. H. Hjertager. Large eddy simulation of the gas-liquid flow in a square cross-sectioned bubble column. Chemical Engineering Science, 56:6341–6349, 2001.
[3] P. Renze, A. Buffo D.L. Marchisio, and M. Vanni. Simulation of coalesence, breakup, and mass transfer in polydisperse multiphase flows. Chemie Ingenieur Technik, 86:1088-1098, 2014.
[4] A. Behzadi, R.I. Issa, and H. Rusche. Modelling of dispersed bubble and droplet flow at high phase fractions. Chemical Engineering Science, 59:759-770, 2003.

[Romain ARNAUD]

Hi Hester,

I'm currently trying to simulate the same column but with a LES model with the latest version of OF. I'm taking drag (model Ishii&Zuber), virtual mass (constant coefficient) and lift (constant coefficient) into account.
I also have the same problem, e.g compared to Deen experimental results, the liquid velocity is to low and the velocity for the air is to high.
Have you find any solution to your problem ?

Romain

[hester]

Hello Romain,

the main issue with my model was the turbulence modeling. Using k--model for water and no turbulence for air didn't work at all. I am now using the mixture model implemented in twoPhaseEulerFoam and it looks good. At the outlet region I used zeroGradient instead of pressureInletOutlet for U.air. The results were better but are still not perfect. Also the simulation presented in this thread used SchillerNaumann for drag modeling which yields very low drag coefficients and therefore the velocity for air is predicted too high and for water to low.

I found that Openfoam 2.3.x yields strange results for this simulation setup that is why I use 2.3.0. Maybe you want to check that out too.

How did you implement the drag model after Ishii and Zuber? Did you take the phase fraction into account? I think the problem for difference in velocity is most likely the drag coefficient.

Regards,
hester

[Romain ARNAUD]

Hello Hester,

For the outlet condition as I had some weird things happening with my previous case (slip condition for U.water and pressureInletOutlet for U.air), I decided to also simulate the interface between air and water but this add a lot of cells in 3D ....

For the drag coefficient, I choose a constant coefficient as in Deen's article. Indeed Ishii & Zuber gave a formula for the drag coeffcient for distorded bubble only depending of the Eotvos numbe and OF doesn't take into account multiple bubble size so constant coefficient should do it. This gives Cd = 0.5 for 4mm diameter bubbles.

Moreover I used a blending model for the air/water mixture so the drag force is calculated taking into account the phase fraction.

Did you manage to get closer results to experiments using the 2.3.0 version ?

Regards,
Romain

[hester]

Hello Romain,

using 2.3.0 I got a meandering bubble plume, as expected. With 2.3.x the bubble plume got stuck in a corner for some reason, that's why I decided to use 2.3.0.

When using the Ishii-Zuber drag model I also experienced problems at the outlet region (I didn't include the water surface) as described here:
http://www.cfd-online.com/Forums/ope...et-region.html
How did you fix your outlet problem?

I am very much interested in your case setup (although I'm not using LES). If you manage to obtain the expected results I like to hear about it.

Regards,
hester

[Romain ARNAUD]

Hi Hester,

Using LES model for turbulence gives me a meandering plume as expected.
When not using a free surface I have some kind of bubbles appearing and quickly dissipating. And when using a free surface the case crashes because of the free surface.

So I didn't fix yet the outlet problem, but I'm not sure it's responsible for the bad correlation with experiment.

Romain

[masoudsh]

I'm new Foamer,and i wnt to use twophaseeulerfoam
i want to simulation bubble column
where should i start it?
thanks