[mvoss]

Dear all,

i am facing a very strange situation here. I am trying to get an initial solution for the FSI3 case of the Turek Hron Benchmark case.

Simply spoken, this means Kárman street + a little flag (pic1) and a parabolic inlet profile with 2m/s mean velocity.
This seems quite forward, so i´ve used my ICEM-hex mesh from the simulation i´ve made with Ansys CFX and since i know about Courant numbers in OpenFoam, i´ve chosen pimpleFoam with max Co=2 and andjustableTimestep for solving my problem.
The initial setup was a resting fluid and i was using a smooth sin-function to "ramp" the velocity-profile at the inlet over time (0-2s).
Attatched you´ll find the given result after just a few timesteps (pic3), showing a very strange "vortex-core". As you might see, the center of the core is just where the bl-resolution is crossing the main flowvolume. At this point in simulation time the ramp-function is somwhere near or arround 0.001m/s thus not causing anything near the shown 1.1m/s.

So here is my question... is this supposed to be caused by a wrong solver (GAMG) for u or p OR is this because of something wrong with the schemes being used? I want to make sure not to start digging in the wrong direction.

Waiting for yout thoughts and thank you in advance,

neewbie

[Tobi]

Hey Matthias,

i tested your case the last 8 hours. I tried several BC and wanted to create a mesh with Salome to compare it but i realized that this is not as easy as i thought.

I 've the same problem's with the velocity and i looked at the vectors. It seems that your mesh is cousing the problems.

After i tested so many things i decided to create a 3D mesh with snappyHexMesh. I solved it and got better results for t=0.01 s ( see pictures). In the pictures you can see that some vectors are showing into some wrong directions. I don't understand why your mesh is cousing that problems but i think thats the reason.

I am still running the 3D mesh and i 'll give you more information tomorrow (or today )

So i think its not a problem of the schemes or solving p with GAMG.
Did you solve that problem with ANSYS?

[Tobi]

hi,

after some simulations with pimpleFoam, icoFoam and a 3D - Mesh i think your mesh is cousing some problems (Min Volume ~ e-12) but you can solve your case!

You have to set your timestep dt very low (icoFoam dt=0.000001). With that settings you get a very good solution.

Maybe you can set the Co-number to 0.01 or sth like that to solve it with the pimpleFoam.

Hope that was helpful.

tobi