[Waldfreund]

Hi,
I am trying to model droplet deposition on a flat surface with a constant / dynamic contact angle. For the simulation I used and modified the dam break case. When I use a kinematic viscosity of 1e-05 m2/s or higher everything is working fine. But when I use the kinematic viscosity of water, 1e-06 m2/s, the droplet reacts very unrealistic and jumps back and forth. I attached everything as a zip-file.
I would be grateful if someone could help me solving this problem.
I am using OpenFOAM 3.0.1.



Thanks a lot


Phil


PS: The mesh is so large because I want to vary the size of the droplet

[Waldfreund]

Does nobody has an idea ?

[Tobi]

Dear Philipp,

I take a few minutes and checked your case. First I run the case without modifications to see what you are trying to do. After that I changed it to a 2d case to investigate further with faster calculation. After I saw what happen in more detail by using a finer mesh, I got the problem. It is related to the interface behavior. It was already discussed here and I made a few checks as mentioned here: http://www.cfd-online.com/Forums/ope...tml#post607243. The problem is in the nature of numerics. I had a discussion with one colleague who is a genius in this topics. He told me that the discrete interface pushes forth and back and make the simulation non stable. If the mesh gets smaller the numerical interface motion gets so strong that it will totally deform the interface and lead to non-physical results (as you already experienced). As you can see in my post, the velocities of the interface where round about 0.1mm to 3mm . Your mesh dimensions are 5mm and hence you get that problem. You will amplify the problem if you reduce the viscosity of your fluid because the viscosity is nothing more than a diffusion coefficient.

You can compare it with the Gauss Upwind scheme that is of 1nd order accuracy. If you derive the truncation error you will figure out that it is similar to a diffusion term. Thats the reason why we say that the Gauss Upwind scheme is diffusive and smears out everything. The good thing is that we do not get really sharp gradients that is numerically nice and hence it is stable. This is similar to the diffusion coefficient you set. The lower it is, the sharper the interface and the more problems you get with that interface motion (maybe I am not 100% correct here because it should be only valid for one phase. I can not say in which way the interface is treated; never went into that area). But I could think that it is related to that. You can make simple tests. Checking the high viscosities and the behavior of the interface and compare it with the low viscosity setting). By the way ... as you also see in the thread that I mentioned above, this error occurs after version 2.2.x. So you can try to solve your case with version 2.2.x. If you do so, it would be very great to share your results. You also can make a bug-report if you figure out that in 2.2.x everything is fine and in newer versions not.

From 2.2.x to 2.3.x they introduced the new MULES Implicit stuff. I already checked a few things and Arjun too but as you can imagine - for me there is no time for investigating into that (even if I want but at the moment I have my AFC project that I want finish and share with the community. And it is still in progress after 2 years -.-, but I am not always working on the project). If you will try OpenFOAM-2.2.x, please replay. If it is really like that make a bug report and Henry can check it out again or give a answer that you can share here in the forum (but as in the last post in the link above, Arjun already disclaimed that everything should be okay by a statement from Henry).


Good luck and hopefully it is helpful.

PS: To check if it really related to the interface motion, use a low viscosity and scale everything by a factor of 1000. Try it and if it is working as expected, the problem is probably related to the interface and the small mesh dimensions. If not, I don't know the answer (:

PPS: The above stuff is just a guess. I am not so familiar with the VOF method, especially with the mathematical background (how MULES is working etc.).

PPPS: what I did not show in the other thread, my interface (the sphere) was starting to move and was getting deformed and jumping like in your case.