[Nek5000-users] Reference for LES filter

[stefanke]

Paul: you may want to supplement my answer.

To give you an idea of what we're doing here:

1. construct a low-pass filter transfer function (to preserve the  
accuracy the transfer function needs to be smooth). The particular  
shape of the transfer function depends in the application.
2. transform the Lagrange polynomials into a hierarchical modal basis  
(e.g. Legendre polynomials)
3. apply the filer
4. apply transpose of transformation (see step 2)

At the moment we support only two options for LES:

- dynamic Smagorinsky (only tensor product domains are supported for  
now)

- simple filter model
In this model we simply filter the velocities u after every timestep  
(e.g. only last 2 modes are filtered using a quadratic transfer  
function. The last mode is filtered by 5%). The filtering operation  
does not preserve div(u) however this effect is typically very little.  
I cannot tell you how much and many modes you need to filter because  
this model lacks of a physical justification. However it is in the  
spirit of every LES model. All of them try to remove energy which you  
cannot present by your grid.

hth,
Stefan


On Oct 16, 2009, at 6:46 PM, 王志成 wrote:

> Hi Stefan,
>
> I'm puzzled with the subroutine build_1d_filt in postpro.f. Could  
> you please give some papers on this problem?
>
> Regards,
>
> Zhicheng
>
> 2009-10-17
> 王志成
> _______________________________________________
> Nek5000-users mailing list
> Nek5000-users at lists.mcs.anl.gov
> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/nek5000-users/attachments/20091016/3752c8e2/attachment.html>