[maike151]

Hello everyone,


I am trying to simulate an air flow through a curved pipe (you can find the files here: https://www.dropbox.com/s/ahuf08jhwh...ulent.zip?dl=0). I already did a laminar simulation and it worked fine and returned the expected values.


Now I want to do the same simulation with turbulence. It converges, but when I visualize the stream lines in Paraview, they are parallel and laminated just like in the laminar simulation (picture attached). I'd expect some vortices. I already increased the Reynolds-number by increasing the velocity at the inlet to 100 (nu=3*10e-5 and L=1)! Still no turbulence. I refined the mesh and ran the case with 3.5 million points. Still no turbulence. I decreased the convergence criteria to 1e-4. The simulation did not converge (with the fine mesh). Does anyone have an idea what else I could try? I'd appreciate any help.



I am using simpleFoam and the turbulence model is RAS (kEpsilon).


Thank you
Maike

[cryabroad]

RAS models typically give you a smooth field, because it is the averaged field that it is calculating. To "see" vortices you need to do LES.

[maike151]

Hello cryabroad,


thank you for your answer!


I thought about that, too. I agree to that as far as the value of the velocity is averaged. But don't the streamlines represent the direction of the velocity which should not be affected by the averaging of the value of velocity?

[cryabroad]

The velocity used to construct the streamlines are also averaged, which is the time averaged velocity. I do think that in curved pipes you should get secondary flow though, but that depends on the curvature and flow rate and other parameters.