[einatlev]

Hello!

In preparation to building a more complex model, I am running interFOAM in two viscous flow cases for which there are analytical solutions: dam break, and point source on inclined plane. Unfortunately, I noticed that interFOAM doesn't agree with the analytical solution.

Case 1: dam break within a walled, rectangular channel. Analytical solution (by Balmforth et al. (2007)) states that at long time, the length of the flow should scale with time to the power of 0.2. So a plot of log(distance) versus log(time) should be a straight line with a slope of 0.5. OpenFOAM gives a straight line, but with a slope of 0.3. I ran the same model with Flow3D and got a slope of 0.2, as should be.



Case 2:
A point source with constant volumetric flux rate, at the top of a flat inclined plane with a constant dip. Analytical solution (Lister 1992) states that the flow length at long times should scale as time to the power 0.78 (7/9). In my OpenFOAM simulation, the flow's length scales as time to the power 0.6. I have not tested this case in Flow3D since I don't have a license anymore.



I placed the input files for both models here:
http://www.ldeo.columbia.edu/~einatlev/BM1/

http://www.ldeo.columbia.edu/~einatlev/BM2/

I am pretty sure that I set up my models correctly, but perhaps I missed something... my suspicion is on boundary conditions...

I would greatly appreciate any advice!

Thanks,
Einat

[vatavuk]

Hi,
I suggest that you also include the 0 folder so we can check the boundary conditions. I am interested in the papers that you cited, can you give the full references?
Best Regards,
Paulo

[vonboett]

well, the difference in viscosity between the two phases is quite extreme.
May I ask what liquid that is?
Have you tried out nOuterCorrectors 2; in the fvSolution?
I currently validate interFoam by a 40m long and 1.1 m wide turbulent clear water open channel flow with 50 l/s discharge and by a 32 m long and 3 m wide rough mountain torrent channel with 1000 l/s flow, both with URANS-LES, and I hit the measured average surface position within 2.5 % accuracy, and about the same deviation I get with vertical velocity profiles except the closest 3 cm to the wall were the URANS causes some greater deviations...

[einatlev]

thanks to those who replied!

vonboett -- sure, the viscosity difference is huge. That's because I am simulating lava flows (usual viscosity anywhere between 400 Pa s to 4e8 Pa s), flowing under air. I thought interFOAM was the best tool in Open FOAM to simulate such a free-surface flow, but I will be happy to switch to another tool if such exists.

vatavuk -- I uploaded the "0" directories for both cases to the website. Good catch.
The full references for the papers I cited are:

Balmforth, N. J., R. V. Craster, P. Perona, A. C. Rust, and R. Sassi (2007), Viscoplastic dam breaks and the bostwick consistometer, Journal of Non-Newtonian Fluid Mechanics, 142(1–3), 63–78, doi:10.1016/j.jnnfm. 2006.06.005.

Lister, J. R. (1992), Viscous flows down an inclined plane from point and line sources, Journal of Fluid Me- chanics, 242, 631–653, doi:10.1017/S0022112092002520.

Looking forward to further suggestions,
Einat

[demichie]

Hi Einat,
I don't know why but I am not able to download your cases. Can I ask you if you have used a 2D or fully 3G geometry? What happen at lower viscosities?

Ciao
Mattia

[einatlev]

Mattia -- the models are fully 3D. I am trying a lower viscosity now. Will see what happens :-)

[duongquaphim]

Hello,

Did you check if the initial residual of p is low enough? Can you post part of the log file here? I remembered running case with large viscosity ratio and I even have to increase the number of correction to 10 or 15 to ensure the good convergence of pressure equation at each time step.

Duong