[lr103476]

Hi everybody,

Inspired by the nice presentation given by Eric Paterson about waves and multi phase stuff, I tried also a nice example using interFoam, version 1.3.

It is my first experience with the interFoam solver, but I succeeded in solving the multiphase flow around a plunging cylinder. The boundary conditions are similar compared to the damBreak tutorial. So, walls left, right and lower, and on top atmosphere.

At first, the flow is nicely solved (although the mesh is coarse), which is illustrated in the following movie:

http://www.aero.lr.tudelft.nl/~frank...ngCylinder.avi

The most interesting observation is that the water level rises, which is rather strange. Has anyone any ideas how this is possible. Did I use improper boundary conditions, or is conservation not preserved when mesh motion is used in combination with interFoam. Why is the mesh motion removed from interFoam in the 1.4 release?

Regards, Frank

[liu]

Might be the BC of alpha at the moving boundary.

[lr103476]

What do you mean? gamma? I just put gamma on the moving wall to be zeroGradient, like the other walls. pd is also set to zeroGradiet on this moving wall.

Furthermore, why is mesh motion removed from interFoam in the 1.4 release?

Regards, Frank

[eugene]

I think the increase in water level is due to the waves at the zeroGradient boundary. Certainly you see the same kind of thing for a wavetank with a zeroGradient outlet alpha boundary. Again this is speculation, but I think the problem is akin to pressure floating in a domain with no fixed value pressure boundaries or cells.

[liu]

So what could be the correct BC for gamma?

[eugene]

I'm not sure. And after some reflection, I don't think the wavetank example is related to Frank's problem.

Look at the gamma equation in 1.3:

ddt(gamma) + div(phi, gamma) + div(phiIR, gamma)

If both the fluxes phi and phiIR are zero on the boundary, then the total ammount of gamma should be preserved irrespective of the gamma boundary conditions.

phiIR however is a function of the surface curvature and for the moving mesh case, phi will be non-zero on the cylinder surface. I guess the easiest experiment would be to sum these two fluxes over all boundaries to figure out where the imbalance is coming from, i.e. calculate div(phi, gamma) and div(phiIR,gamma) as a global sum.

[egp]

Frank,

I'm glad my stuff motivated you!

Concerning your problem, how did you set the top of your domain? I would use the Atmosphere b.c. (which sets the pressure), however, I would also make sure to move it much further away from the oscillating cylinder (maybe 5-10 diameters). I would set all of your side walls to noslip to make a case of a cylinder oscillating in a small tank of initially quiescent fluid.

Good luck, and keep us informed.

[yingfeng]

Frank, have you solved the increasing water level problem? I think I got the similar issue when I was trying to simulate a plane moving down towards a drop of water. The volume of the water is actually increasing even there is no wall contact with it yet. The observed gamma value can be much higher than 1.0. I guess this is related to mesh flux, or it might be that my boundary conditions are wrong somewhere. Here I enclose two pictures from paraFoam.




The version used is March foam1.3. I am interested to know your progress in the sloshingCylinder case.

Yingfeng