Q-criterion based on RANS?
 

Dear all,

is it possible/logical to generate Q-criterion isosurfaces with RANS?

My experience is with LES, but I'm curious.

Thanks.

My opinion is that there is no physical meaning in RANS.

But Q-criterion for the mean flow simulated with LES does have meaning, right?

Yes that's right

Let me disagree on this... or, at least, with the fact that Q-criterion has significance for the MEAN flow from LES but not from RANS.

I don't dispute the significance of the Q criterion for the MEAN flow in general (you can think both ways). But, it either is significant for any mean flow or it isn't for any.

Note also that the mean flow itself may be more significant in RANS than LES, as LES is not a guarantee of anything in terms of results' significance.

Of course, I sustain the significance for the INSTANTANEOUS flow field in LES. I'm not sure instead about its use for the instantaneous flow field in URANS/DES (but some people use it).

Paolo I was focusing on RANS, therefore a statistically steady solution. In such case there is nothing that must be detected, you can use the mean variables.
Conversely I agree about URANS/DES, in such formulations you can adopt Q- or any other structure detection.
Consider that often the goal is to detect choerent structures. That means 1) identify the structure in an instantaneous field 2) Associate the character of coerence to the detected structures.

I know that you know what we're talking about. Couldn't be otherwise.

And I know that you know that I know what we're talking about.

But he asked about the MEAN LES flow, the one after averaging the instantaneous ones.

Why this mean flow is different from the RANS one in terms of Q criterion meaningfulness?

The proccess should be differently intepretated... you first detect the structures from the instantaneous field and then assume some statistical data reduction in time. Conversely, if you use directly the MEAN field this is not congruent as the choerence character cannot be detected.

To be more clear, immagine what happens if you would do a spectral analysis of the MEAN field or if you compute the MEAN of several samples of spectra.

But even in that case, even if I never tried, I don't think that averaging instantaneous Q-related stuff can produce meaningful things

The key is the average only over the characteristic turn-over time, not over a complete time interval [0, T->+Inf]. That will produce the "coherent" character.
Still considering the example of the spectra, if you have a vortical structure containing energy at some wavelength you can do the statistical averaging of several spectra, they still highilght the energy peak for that structure. But what if you wanto to do the spectra of the statistically averaged fields?;)

I see your point, but I still have doubts that this was what the original poster meant by "mean flow simulated with LES"

Quote:
Originally Posted by FMDenaro https://www.cfd-online.com/Forums/im...s/viewpost.gif
My opinion is that there is no physical meaning in RANS.

But Q-criterion for the mean flow simulated with LES does have meaning, right?


Yes Paolo, if the previous post stands for a "statistically mean" such that <> is a RANS operator, I agree. This mean field would not be useful to be analyzed by such criterion.