Some questions in the spatial derivatives parameter setting

[jushi-123]

Hello Dr. Laizet and the Incompact3d Development Team,

I have been using Xcompact3d for DNS and LES of turbulent channel flow and have some doubts in the spatial derivatives setting.
The second derivatives is usually set as hyperviscous 6th and the important parameter in input file is "nu0nu" and "cnu". When we simulate the DNS or LES of turbulent channel flow, the spatial resolution has been fixed. After that, how to set the "nu0nu" and "cnu"?Whether this two parameters should be adjusted according to the spatial resolution?
I am looking forward to receiving your valuable suggestions.

[mathrack]

Hello,

The implicit LES available in the code is based on http://dx.doi.org/10.1016/j.jcp.2011.01.040.

One can adapt the numerical dissipation in each direction. It is also possible to change the numerical dissipation used for transported scalars.

If you know (or have an estimation of) the turbulence spectrum + DNS resolution requirement, then it is possible to estimate the extra dissipation needed in the implicit LES, based on the exact grid used in the LES.

[slaizet]

To complement this, we are often relying on a test and trial approach to fine tune these parameters (range of values for cnu available in the paper from the previous answer).
Sylvain

[jushi-123]

Thanks for your valuable suggestions! So the suitable parameters should be adjusted based on the default value(4.0, 0.44)

[slaizet]

yes. Fro cnu, typical values are 0.44 or 0.22. For xnu0xnu, you can go from 1 (no added dissipation) to 10^4. xnu0xnu represents the amount of numerical dissipation to be added at the cut-off wave number. cnu represent the range of wave numbers over which you add numerical dissipation.