DES
 

Hello everybody, I have a question and/or comment to do. To those of you who have run any DES simulation (or LES) I´m trying to visualize the "edges" of the structures using the variable that I saw in the Menter's DES paper available in CFX's website. The thing is that they describe it as (hard to type it here)

(partial ui/partial xj)/(partial uj/partial xi)

This notation is confussing and, for me, it´s a tensor ratio which makes no sense. So, then, my first question is whether any of you have used this formulation before and, if so, could explain it more in detail.

In addition many of the variables that the solver generates to the post are not clear, ie vel.trnavg or WVTRNRVX (yes, I'm serious) nor explained anywhere.

Whatever comments will be wellcome. Thanks. pi

Re: DES
 

Hello Pi, do you use 5.6 or 5.7? As far as I know DES is a full feature since 5.7. In the old version there are somethings looking like construction what is not unusual for a beta feature.

vel perhabs stands for velocity? trn for transient and avg for average.

Thats' not much but I hope it helps mstein

Re: DES
 

Hi,
As far as I know, if you're using 5.6 you can get that nice iso surfaces by enabling the VORTICITY variable under cel:
LIBRARY:

• VARIABLE: vorticity
  • Output to Postprocessor = Yes
  END

END
Go to POST and generate a new variable using the following equation:
Vorticity^2 - sstrnr^2
And you'll get the pretty vortex structures. You have to adjust the magnitude of your isosurface to see what is relevant or not.
Regards, G. G.

Re: DES
 

There are actually many papers devoted just for different techniques for representing vortices. Vorticity is the simplest one, the one G.G. mentioned is actually the Poisson equation, except that I know it as (0.5*vorticity^2 - Shear Strain Rate^2).

They will actually show the turbulent structures quite differently, so you have to remember that there is no one correct technique. Also, the "edge" of the structure is quite an arbitrary thing, it really depends on how you define it!

As for all the variables in Post, I can't remember seeing it look like that, but its the running average of some variable. CFX5's LES/DES is nice enough to calculate the rms or mean values of any variable you choose, something which was much more difficult in CFX4. The CFX5 LES is very very good in my opinion.

PS change the slip walls to no-slip and see how it goes.

Re: DES
 

Hi mstein, GG and James,

thanks for your help. We're still using 5.6 till we receive the new version (soon hope).

One interesting point. Anyone know why the isosurfaces for that variable, omega^2-ssr^i2 are almost constant whatever value you set? I guess that it has been a long time searched property.

One last question. What is the physical meaning of that variable?

James, if you use to run LES take a look to NIST's FDS code. It looks very fast, easy to set and efficient...and it's shared code.

Thanks again guys

pi