velocity contour of code_saturne

[yany]

Hello everyone.
The velocity contour form vertical axis wind turbine simulation is not match with that from ansys-fluent. I think the result from code_saturne is not right. Could you please give me some suggestion about that?

All best best, Yan.

[Yvan Fournier]

Hello,

Without any details on computation meshes and paramaters, I wpuld not trust any CFD computation.

So if you do not provide more details I would assume your input is wrong (and this may include your FLUENT computation).

Regards,

Yvan

[yany]

Hello,

Without any details on computation meshes and paramaters, I wpuld not trust any CFD computation.

So if you do not provide more details I would assume your input is wrong (and this may include your FLUENT computation).

Regards,

Yvan

Hello Yvan,

I attached the mesh and configuration of the H-type axis wind turbine. It is an unsteady RANS calculation. I used the k-omega SST model. I have one rotational zone in the center and one stationary zone at the outside. The SIMPLIC method is used.

I think the mesh quality is OK. The y plus on the wall is around 1. I used the same mesh for both fluent and code_saturne.

The reason why I have doubts for code_saturne is that the trend of the torque is not right, let alone the magnitude. Whereas, the fluent gives the right torque as I expected.

The normal stress of wall surface on one blade is also attached. This is also very wired, but I suspect that this is the problem with output.

Also I attached the setup.xml
All the best, Yan.

[yany]

Hello,

Without any details on computation meshes and paramaters, I wpuld not trust any CFD computation.

So if you do not provide more details I would assume your input is wrong (and this may include your FLUENT computation).

Regards,

Yvan

Hello Yvan.
The setup.xml is also attached.

All the best, Yan.

[Yvan Fournier]

Hello,

The mesh seems OK. Is it extruded over 1 cell thickness ?

Could you try a k-epsilon or RSM model ? Also, k-omega has options for different wall laws (advanced options for turbulence). Did you try alternative options ?

Also, did you use the "extract block" filter in ParaView when viewing the stress ? If not, you may have a viewing artefact. If you did use it, then is seems there may be a numbering issue in the mesh output. In this case, if you ran with multiple OpenMP threads, could you try with only 1, to avoid forcing renumbering ?

Which bug-fix release of the code did you use ?

Finally, what range of CFL numbers do you have ? Could you post your "listing" file ?

Depending on the answers, we may see what to adjust to try to solve the issue.

Best regards,

Yvan