[bretttully]

Hi guys,

I am relatively new to OpenFOAM and am trying to assess if it is a suitable tool to use for my DPhil.

I have started by trying to design a simple case to test features that I will need to use.
I have a 1/4 circle extruded to make a "wedge" cylinder. The flat surfaces are symmetry conditions, and the circular face is a wall (essentially forming an pipe), the ends are inlet (fixed velocity) and outlet (fixed pressure).

I can run this with a blockMesh generated mesh with no problems, however, it will be important for me eventually to be able to use unstructured tet meshes. For this, I have written a *.geo file and fed it to gmsh2ToFoam and as far as I can tell the generated mesh is decent (viewed in paraFoam, and run checkMesh). Yet, if I pass the same control/initial conditions from the blockMesh case to this one, the first few iterations _converge_ but it eventually blows up. Looking at the results from the first few iterations with paraFoam shows that these solutions are not at all correct.

Can anyone suggest where I can look to see what is causing the trouble. It is important to be able to use unstructured grids, as my project will require some very complex meshes.

Cheers,
Brett.

[wese]

Hi Brett,
I've also just started using openFOAM.
So I also tried to simulate the flow through a circular pipe.
First I've tried icoFoam, but my results also were divergent.
So I've switched to simpleFoam for the steady state solution but even now I received strange results.
After a while I figured out, that the "double used" points (as described on p. 148 of the users manual) in the middle of the tube were the problem (but untill now I have no idea why).
An image of the calculated velocity profile is in the attached case (t1_u.jpg).
After I've put then on one of the sides everything worked fine (see images the attached case):
tube.tar.gz

Greetings
Christian