CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   OpenFOAM Running, Solving & CFD (http://www.cfd-online.com/Forums/openfoam-solving/)
-   -   LES dynSmagorinsly model average (http://www.cfd-online.com/Forums/openfoam-solving/60183-les-dynsmagorinsly-model-average.html)

 anne June 5, 2006 04:14

Hello, I would like t know

Hello,

I would like t know how does exactely
the "average" function operate:
I have read on a discussion forum that "average"
performs a global avarage: does it mean that
it also applies in non-homogeneous directions ?
As an example, when simulating the channel flow,
the average is performed in the three directions ?
OR only in the homogeneous ones ?

Anne

 pierre June 5, 2006 10:38

Well in the programmer's guide

Well in the programmer's guide, it says:

"Average, fvc::average produces an area weighted average of surface<type>Field face values, i.e.

Sum(Af*Xf)/Sum(Af) , and returns a volField<type>."

With Af the face area, and Xf the field face value.

Pierre

 anne June 5, 2006 11:00

Hello Pierre, Indeed there

Hello Pierre,

Indeed there is a fvc::average function, but
this is not the one I talk about.
LES modelling subroutine for
computing the dynamic constant.
usually for these models, averages
are done over homogeneous directions.

BUT, In openFoam, I read that it operates
"global" averages and I am still
without knowing to what exactely
corresponds "global".

If someone could inform me ...

Thanks

Anne

Hi
In dynamic smagorinsky model we should have homogeneous averaging, whats that mean?

thanks

 ArathoN February 23, 2014 09:25

If i remember correctly the smagorinsky model use a similar hypothesis to the boussinesq, where there is proportionality between the SGS stres and the rate-of-strain tensor, so for the point of view of the matrices it means that the "principal vectors" (sorry don't know the exact English term but it should be the base of the matrix) are aligned, and so you impose the homogeneous condition. This was the problem for the standard smagorinski model because you'll have only one variable to play for each flow (Cs) then Germano came up with the dynamic concept and there you could (auto)adapt better the model to the problem you are studying.

This is what i know unfortunately my turbulent flow course didn't give us a detailed information about LES, even thought we studied for les mainly the Germano method (he was my teacher in another course before retiring).,

 Bernhard February 24, 2014 05:45

Quote:
In addition to Arathon's answer. In, for example channel flow, you don't expect any statistical differences in the streamwise direction (homogeneous direction), therefor you can average the coefficient for C_S in planes parallel to the wall. Of course, this only works for simple geometries where you have these kind of planes. In OpenFOAM you have the homogeneousDynamicSmagorinsky model, which averages C_S over the complete domain.