[shuangqingx]

Hi, everybody,
Can anyone help me with the LES quality assessment by means of the subgrid viscosity ratio? I mean, is there any particular requirement about the subgrid viscosity ratio when using LES? I found little relevant information in this forum (http://www.cfd-online.com/Forums/cfx...ratio-les.html) but was still confused.

[FMDenaro]

Originally Posted by shuangqingx

Hi, everybody,
Can anyone help me with the LES quality assessment by means of the subgrid viscosity ratio? I mean, is there any particular requirement about the subgrid viscosity ratio when using LES? I found little relevant information in this forum (http://www.cfd-online.com/Forums/cfx...ratio-les.html) but was still confused.

The quality of an LES solution can not be assessed by the SGS viscosity. Other more relevant quantity must be considered (RMS, spectra, etc.).
Furthermore, LES can be performed with many other SGS models other than an eddy viscosity model.

[shuangqingx]

Originally Posted by FMDenaro

The quality of an LES solution can not be assessed by the SGS viscosity. Other more relevant quantity must be considered (RMS, spectra, etc.).
Furthermore, LES can be performed with many other SGS models other than an eddy viscosity model.

Thanks for your reply.

I'm sorry for not telling the context. I've done a LES with the standard Smagorinsky-Lilly SGS model and was asked for the quality assessment for that. The viscosity ratio method is suggested as one of the standard methods (other methods including the fraction of resolved TKE, a two-point correlation). In the "fraction of resolved TKE" method, it's reported that a good LES should resolve at least 80% of the TKE.

So I wonder if there is any similar criterion for the "viscosity ratio" method. Hope this makes the question more clear. Thank you.

Shuangqingx

[FMDenaro]

Originally Posted by shuangqingx

Thanks for your reply.

I'm sorry for not telling the context. I've done a LES with the standard Smagorinsky-Lilly SGS model and was asked for the quality assessment for that. The viscosity ratio method is suggested as one of the standard methods (other methods including the fraction of resolved TKE, a two-point correlation). In the "fraction of resolved TKE" method, it's reported that a good LES should resolve at least 80% of the TKE.

So I wonder if there is any similar criterion for the "viscosity ratio" method. Hope this makes the question more clear. Thank you.

Shuangqingx

I suggest to check for RMS, spectra and zero order statistics as average profiles of velocity. TKE can be also an indication. The SGS viscosity is dependent on the arbitrary choice of the value of the Smagorinsky constant. If you would use the dynamic Smagorinsky model then SGS viscosity would be more useful to check

[flotus1]

If you are absolutely determined to use the SGS viscosity for the Quality assessment of a LES, here it is:



The definition can be found here for example:
Klein, Meyers and Geurts. Assessment of les quality measures using the error landscape approach.
In: Meyers - Quality and Reliability of Large Eddy Simulations, pages 131-142. Springer Science+Business Media B.V., 2008

Having applied this to some LES, I can copy what FMDenaro said.
It is useless to assess the quality of a LES so use the methods he mentioned instead.

One reason might be that the coefficients in the formula are calibrated for isotropic turbulence.

[FMDenaro]

Originally Posted by flotus1

If you are absolutely determined to use the SGS viscosity for the Quality assessment of a LES, here it is:



The definition can be found here for example:
Klein, Meyers and Geurts. Assessment of les quality measures using the error landscape approach.
In: Meyers - Quality and Reliability of Large Eddy Simulations, pages 131-142. Springer Science+Business Media B.V., 2008

Having applied this to some LES, I can copy what FMDenaro said.
It is useless to assess the quality of a LES so use the methods he mentioned instead.

Yes, I partecipated at that conference, the full study was published here:

http://people.mech.kuleuven.be/~jmey..._22_125106.pdf

However, I did not suggested it as I was quite critical in using this methodology for many reasons...

[shuangqingx]

Originally Posted by FMDenaro

Yes, I partecipated at that conference, the full study was published here:

http://people.mech.kuleuven.be/~jmey..._22_125106.pdf

However, I did not suggested it as I was quite critical in using this methodology for many reasons...

Thank you for your kind help, FMDenaro. I've got some idea now.

One more question, senior.
Hartmann et al (Chemical Engineering Science, 2004, 59, 2419-2432. see http://www.sciencedirect.com/science...09250904001575) have some discussions about the eddy viscosity ratio and the grid resolution of their LES (specifically in Section 5.1, the lower right quarter of page 2423). What would you comment on that?

Thanks.
Shuangqingx

[FMDenaro]

Originally Posted by shuangqingx

Thank you for your kind help, FMDenaro. I've got some idea now.

One more question, senior.
Hartmann et al (Chemical Engineering Science, 2004, 59, 2419-2432. see http://www.sciencedirect.com/science...09250904001575) have some discussions about the eddy viscosity ratio and the grid resolution of their LES (specifically in Section 5.1, the lower right quarter of page 2423). What would you comment on that?

Thanks.
Shuangqingx

In principle, if the SGS viscosity vanishes everywhere then it is plausible to assume the LES solution as a DNS one. However, you probably know that the static Smagorinsky model is arbitrarily modulated close to a wall in the sub-layer region (for example using the Van Driest function) to drive the SGS viscosity to zero. Conversely the dynamic Smagorinsky model will automatically detect a laminar region and produce a vanishing viscosity.

In any case, I do not agree that the capturing of the Kolmogorov scale by an LES can be assessed by the profiles of SGS viscosity.

[shuangqingx]

Originally Posted by FMDenaro

In principle, if the SGS viscosity vanishes everywhere then it is plausible to assume the LES solution as a DNS one. However, you probably know that the static Smagorinsky model is arbitrarily modulated close to a wall in the sub-layer region (for example using the Van Driest function) to drive the SGS viscosity to zero. Conversely the dynamic Smagorinsky model will automatically detect a laminar region and produce a vanishing viscosity.

In any case, I do not agree that the capturing of the Kolmogorov scale by an LES can be assessed by the profiles of SGS viscosity.

Thanks for your analysis and suggestion. It helps a lot.

[FMDenaro]

Originally Posted by shuangqingx

Thanks for your analysis and suggestion. It helps a lot.

good
I have some papers published wherein the SGS viscosity profiles were analyzed but I used the dynamic Smagorinsky and the analysis does not imply the LES assessment in terms of capturing of the Kolmogorov scale.

[shuangqingx]

Originally Posted by FMDenaro

good
I have some papers published wherein the SGS viscosity profiles were analyzed but I used the dynamic Smagorinsky and the analysis does not imply the LES assessment in terms of capturing of the Kolmogorov scale.

Would you mind showing me some information about the publications to this point?

[FMDenaro]

see for example

http://link.springer.com/article/10.1007%2Fs00162-010-0202-x#page-1

http://www.sciencedirect.com/science...45793012004537

[shuangqingx]

Originally Posted by FMDenaro

see for example

http://link.springer.com/article/10.1007%2Fs00162-010-0202-x#page-1

http://www.sciencedirect.com/science...45793012004537

Thanks a lot

[shuangqingx]

Originally Posted by flotus1

If you are absolutely determined to use the SGS viscosity for the Quality assessment of a LES, here it is:



The definition can be found here for example:
Klein, Meyers and Geurts. Assessment of les quality measures using the error landscape approach.
In: Meyers - Quality and Reliability of Large Eddy Simulations, pages 131-142. Springer Science+Business Media B.V., 2008

Having applied this to some LES, I can copy what FMDenaro said.
It is useless to assess the quality of a LES so use the methods he mentioned instead.

One reason might be that the coefficients in the formula are calibrated for isotropic turbulence.

Thank you for your advice and for the specific reference.