code_saturne User's Forum

Posted: **Thu Jun 12, 2025 10:25 am**

Hello everyone,

I am recently trying to conduct the numerical simulations of the outer flow of an axisymmetric model. The version I used is v8.0.3. Here is some basic information of my calculation settings.

Firstly, I used kw-SST turbulence model with default wall function to conduct a RANS calculation. The time step option is ‘steady (local time step)’. The RANS results are in good agreement with experimental results.
Then, I tried to change the turbulence settings to LES (WALE or Smagorinsky, the time step option is ‘constant’) in GUI in the same case ‘DATA’ folder, and set the ‘Restart checkpoint directory’ pointing to the ‘checkpoint’ folder in the RANS calculation’s ‘RESU’ folder.

In my calculation cases, I have ensured that the time step keeps the maximum value of Courant number always below 1.0. Besides, the wall y+ of my mesh is around 80~100.

However, despite the fact that RANS calculations can proceed normally, the LES calculations always diverge (field velocity exceeds limits).
My question is that, are my setup procedures for the LES calculation based on RANS ‘setup.xml’ file correct? Are there any other important points that I need to pay attention to?
Besides, could you please kindly give me some advice on how to improve the convergence of LES (WALE or Smagorinsky) simulations?

Thank you very much in advance and best regards,
WangYW

Posted: **Thu Jun 12, 2025 12:24 pm**

Hello,

What type of meshes do you use ? LES in code_saturne can be quite sensitive to mesh quality and regularity. In a similar way, though centered convection schemes are needed for the conservation of energy, a very small portion of upwind (2%) might help stabilize the computation.

Best regards,

Yvan

Posted: **Fri Jun 13, 2025 10:57 am**

Hello Yvan,

Thanks a lot for your reply.

The mesh I am using is a full hexahedral mesh created by ICEM CFD, and the format is .cgns.

During my calculation, the listing file indicated that there are some mesh quality issue in my case, but the number of bad cells is very small compared to the total amount of cells.

Code: Select all

  Criterion 1: Orthogonality:
    Number of bad cells detected: 286 -->   0 %

  Criterion 2: Offset:
    Number of bad cells detected: 107073 -->   0 %

  Criterion 3: Least-Squares Gradient Quality:
    Number of bad cells detected: 37443 -->   0 %

  Criterion 4: Cells Volume Ratio:
    Number of bad cells detected: 5428 -->   0 %

  Criterion 5: Guilt by Association:
    Number of bad cells detected: 9138 -->   0 %

 Warning:
 --------
    Mesh quality issue has been detected

    The mesh should be re-considered using the listed criteria.

    The calculation will run but the solution quality may be degraded...

Should I need to further improve my mesh quality until all of the criteria above have zero bad cell before my calculation can work normally?

Thank you and best regards,
WangYW

Posted: **Fri Jun 13, 2025 6:30 pm**

Hello,

Sometimes, just a few pinched cells can cause issues, but It is hard to tell without looking at your mesh.

Would it be feasible an LES on a simpler mesh, or a truncated version of this one in which you have only the "good" cells, to see if this works better, or if the issue is related to other computational options.

Best regards,

Yvan

code_saturne User's Forum

Question about the convergence of LES calculations

Question about the convergence of LES calculations

Re: Question about the convergence of LES calculations

Re: Question about the convergence of LES calculations

Re: Question about the convergence of LES calculations