[GBarraza]

Hi everyone,

I've been facing a weird behavior in the Skin friction coefficient (Cf) of the NACA 0012 validation case from the OpenFOAM validation cases repertoire (https://www.openfoam.com/documentati...irfoil-2d.html ).

Instead of using the SA model, I use kOmegaSST with a C-mesh created with blockMesh.

I have run 3 cases with same number of elements, but changing the grading in the z direction to achieve different values of y+ (35, 7, and 5). While the 3 of them give me fair values of surface pressure coefficient (Cp graph attached), the Cf deviates more from the reference solution with lower values of y+ (Cf graph attached). I would have expected the solution close to the leading edge would get better by decreasing the y+ values.

Is there anything obvious that could come to our mind by looking to the graphs and the mesh?

I also attach the case files of one of the runs so you can check my setup.

Best regards,
Guillermo

[HPE]

Hi,

- I think your results are pretty good within a typical engineering accuracy expectations.
- Having said that, I think the predictions for this particular case is highly dependent on the wall functions/boundary conditions on wall for nut, and omega.
- Would you try to use "nutLowReWallFunction" for nut, e.g. for the y1+ is within the viscous sublayer?
- May be using fixedValue omega instead of omegaWallFunction may change things as well, but I am not sure if it would be stable enough.
- Might help to tune the turbulence model coefficients to this specific case since kOmegaSST coefficients might be calibrated with other canonical flows. Have a go for a literature review if possible.

Hope these help.

[GBarraza]

Hi HPE,

Thanks for the reply. I have tried the following:

1. I used nutLowReWallFunction, but it yielded the same result as above.
2. I tried fixedValue for omega, but the solution diverges. What value should I impose at the wall for this case? Would it be necessary to change the fvSchemes according to the modifications?

-Guillermo

[HPE]

Hi,

- I was suspecting that using `fixedValue` would cause numerical stability issues, and this happened. I think you can safely use `omegaWallFunction`, since for low-Re type meshes, there is no modelling - but resolving by using the analytical relations within the viscous sublayer.
- For theoretical boundary conditions and their values being used in kOmegaSST, you can refer to the relevant NASA website.
- Some of my previous comments do still apply.
- Yet.. Could you please attach your `fvSchemes`, `fvOptions`?
- Have you double checked the `nu` value - and are you sure that you are replicating the flow in terms of non-dimensional numbers, e.g. Re, given in the description?

Hope this helps.