[Zmur]

Hello.
I failed at setting up the following problem:
The inner volume of a cylindrical tank is separated into two parts by another cylinder. It can move up and down freely (for now I'm not taking friction into account) it also is weightless.
In other words, I have two isolated volumes, filled with air that are separated by
a moving wall.
There is no inlet or outlet.
No remeshing is used, just mesh deformation.
I use an expert parameter to override isolated fluid regions check.
The mesh is swept and consists only of quadrilaterals. About 10 000 cells, maybe less.
Time step is 0.01s Total time is 1s.
https://dl.dropbox.com/u/29706890/Cy...l%20domain.JPG (link will be active for a month or so).
The inner cylinder should start to move upwards when the temperature of the air below it raises. But it doesn't.

At first I tried to create a "rigid body" and use "rigid body solution" in the wall BC. This resulted in small cylinder "jumping" outside of the large cylinder after some calculation time.

After that I wrote an equation in CEL for the movement but it also didn't work (according to Newton's second law). The small cylinder still won't move.

I tried to separate both volumes into two domains and set a higher starting temperature to the lower one. Also I tried putting an Opening BC on the top wall of the big cylinder.

How can I setup this problem correctly?