[Aydar Mukhametov]

Hi everyone!
I want to move from calculations in Fluent to the OF. To do this, we compared the results obtained in both programs.

Problem: turbulent flow in a circular channel.
Fluent boundary conditions:
k-w SST turbulence model;
SIMPLE solver scheme;
Inlet velocity 10 m/s;
Turbulence intensity at inlet and outlet 4%;
Pressure outlet 0 Pa.
Geometry:
Circular cross-section channel,diameter 0.1 m, length 100D.
Symmetry on XY, ZX.

The mesh was created in Ansis meshing and transferred to OF via "Fluent3DMeshToFoam"
The first case of the mesh with y+=1 was with the warning "Failed 1 Mesh Checks" and the calculations did not coincide.
The second case with a mesh with y+=94 completely coincided with Fluent.
Mesh.jpg
log_mesh1.txt
log_mesh2.txt

OpenFoam bc:
setupOF_1.zip
k, w, vel, nut same to Fluent.

Please tell me where the error is or why the rough mesh matches the results. What boundary conditions should be set for the mesh with y+=1.

[Aydar Mukhametov]

Obtained data
omega-axis fine mesh.png
omega-axis rude mesh.png
Vel_profile_axis_r.PNG
v-axis fine mesh.png
v-axis rude mesh.png

[Tobermory]

The mesh check is telling you that the max cell aspect ratio is too high (1922) for your y+=1 mesh ... this can screw up the gradient calculations, since the wall normal gradient component of a variable is so much larger than the streamwise gradient component simply due to the cell dimensions. This means that the solution can decouple, which is maybe what you are seeing.

You can try using something like the leastSquares gradScheme approach which helps counter this, but in reality the best solution is, as ever, to just improve your mesh. If you want to minimise your cell count then just grab the near wall cells that have aspect ratios above 100 and use refineMesh to refine them in the axial direction; you'll need several nested refinements to get the near wall cell ARs down to around 100. That should then help.

Good luck!