CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Main CFD Forum (http://www.cfd-online.com/Forums/main/)
-   -   A flow to verify a SGS model (http://www.cfd-online.com/Forums/main/103508-flow-verify-sgs-model.html)

crwan June 19, 2012 13:16

A flow to verify a SGS model
 
Hi guys:
I am trying to verify my implementation of a LES model, I need a simple and well studied flow case to do that.
I try to use channel flow, however, because the code use octree mesh scheme which make the total cell number unaffordable.
any suggestions?

sbaffini June 19, 2012 13:29

There are a lot of test cases available, depending on the aspect you want to check. Homogeneous isotropic turbulence (Comte-Bellot & Corrsin experiment) is a classical one. Also you can look for the Taylor-Green Vortex. They both allow you to test your LES code without the burden of the near wall issue and have a lot of reference LES/DNS solutions.

A general collection of LES test cases can be found in:

http://torroja.dmt.upm.es/turbdata/agard/

crwan June 19, 2012 13:33

Quote:

Originally Posted by sbaffini (Post 367294)
There are a lot of test cases available, depending on the aspect you want to check. Homogeneous isotropic turbulence (Comte-Bellot & Corrsin experiment) is a classical one. Also you can look for the Taylor-Green Vortex. They both allow you to test your LES code without the burden of the near wall issue and have a lot of reference LES/DNS solutions.

A general collection of LES test cases can be found in:

http://torroja.dmt.upm.es/turbdata/agard/

Thanks, but i am facing one problem, the code can only set periodic boundary condition in more than one direction. actually now i am looking for the mixing layer case, but I can not find a solid turbulence case, do you have any idea?

sbaffini June 19, 2012 13:46

If you check the agard document above (from page 129) you will find several experimental and numerical cases for the mixing layer. Moreover, Vreman (http://www.vremanresearch.nl/) has surely worked on this topic but i don't know if the database is available (probably not).

However, the HIT and TGV test cases require full 3D periodicity (3 directions are more than 1, i guess). Still, i don't get the point in having Octrees and being limited by boundary conditions. It's probably me missing it

crwan June 19, 2012 13:50

Quote:

Originally Posted by sbaffini (Post 367297)
If you check the agard document above (from page 129) you will find several experimental and numerical cases for the mixing layer. Moreover, Vreman (http://www.vremanresearch.nl/) has surely worked on this topic but i don't know if the database is available (probably not).

However, the HIT and TGV test cases require full 3D periodicity (3 directions are more than 1, i guess). Still, i don't get the point in having Octrees and being limited by boundary conditions. It's probably me missing it

well, i use a commercial code and implement new les model via user define functions, the code mainly focus on the engines, and there are some troubles with periodic bounder conditions.

FMDenaro June 19, 2012 14:45

As already suggested, isotropic/homogeneous turbulence and channel flow are a MUST for any new SGS modelling. You cannot avoid to check for performance in such test-cases

cfdnewbie June 19, 2012 15:36

Quote:

Originally Posted by FMDenaro (Post 367302)
As already suggested, isotropic/homogeneous turbulence and channel flow are a MUST for any new SGS modelling. You cannot avoid to check for performance in such test-cases


I second that, I have yet to see a SGS model that has not been tested against Taylor Green or HIT....both flows have been studied very extensively....
any chance you could fix the periodic bcs??? in fact, they are the easiest one to implement of all bcs, so maybe you can talk to the programmers / vendors...


All times are GMT -4. The time now is 13:02.