[Tarak]

Hii,

I'm trying to run the circular cylinder case for Re=3900 using unstructured grid (prism/tetras) generated by ICEM. I'm using a hybrid DES/LES model. Although I'm being able to run the case using Upwinding scheme, the solution blows off after some time as I switch to CD scheme. Does it have to do anything with the mesh quality, given the checkMesh output as:

Checking geometry...
Overall domain bounding box (0 0 0) (0.579999987 0.4199999906 0.07999999821)
Mesh (non-empty, non-wedge) directions (1 1 1)
Mesh (non-empty) directions (1 1 1)
Boundary openness (2.796323021e-17 1.133420052e-17 -2.412554847e-16) OK.
Max cell openness = 2.744236741e-16 OK.
Max aspect ratio = 5.985448358 OK.
Minumum face area = 1.860038895e-07. Maximum face area = 0.0001631631359. Face area magnitudes OK.
Min volume = 7.890352088e-11. Max volume = 7.346122294e-07. Total volume = 0.01946299762. Cell volumes OK.
Mesh non-orthogonality Max: 63.07437883 average: 7.798812186
Non-orthogonality check OK.
Face pyramids OK.
Max skewness = 0.8103598145 OK.

Mesh OK.


Are the parameters such as aspect ratio, skewness ok for OF to handle tetra domainated meshes?

Can someone advice any other settings (for schemes etc) which he/she would prefer for this case.

Thanks,
Tarak

[stevenvanharen]

Hi Tarak,

as I understand it unstructered grids increase the unboundedness of schemes.

I have had some succes using the blended scheme for the convection term, however, in my case (low-Re DNS) this gave so much numerical diffusion that the solution turned laminair (for 5 percent upwind).

You can also try to use really low CFL (<0.5).

Hope this helps.

[Tarak]

Hii Steve,

Thanks for the reply. Can you please tell me how to use blending of Upwind and CD schemes in OF.

Thanks,
Tarak

[stevenvanharen]

this is for 5 percent upwind:


div(phi,U) Gauss blended 0.95;