[dune-pdelab] Visualization of Gauss point data

[Nicolas Schwenck]

Sorry, I still don't get the problem:

If you solve the elasticity problem as a single field problem (solving 
only for displacement directly) then you don't have stress or strain in 
the weak form and also not in the local operator. So, why is it costly 
to compute the stress in the driver before passing to the vtk writer?

As Felix already pointed out, a mixed approach would directly give you 
stress and displacement (you could also include strain) as a solution 
vector.

As Christian E. pointed out the subsampling can help you if your 
visualization tool cannot handle higher order/discontinous basis functions.

Best,
Nicolas



On 07/20/2011 03:34 PM, Christian Simon wrote:
> Dear Nicolas,
>
> yes, the deformation is my primary variable. In order to compute the 
> the deformation, I need to compute the stress (it is necessary in the 
> residuum). So I have the stress only in the local operator.
>
> Thanks in advance,
> Christian
>
>
>
> -----Nicolas Schwenck <nicolas.schwenck at iws.uni-stuttgart.de> schrieb: 
> -----
>
>     An: dune-pdelab at dune-project.org, Christian Simon
>     <christian.simon at twt-gmbh.de>
>     Von: Nicolas Schwenck <nicolas.schwenck at iws.uni-stuttgart.de>
>     Datum: 20.07.2011 15:08
>     Betreff: Re: [dune-pdelab] Visualization of Gauss point data
>
>     On 07/20/2011 02:45 PM, Christian Simon wrote:
>>     Dear all,
>>
>>     I moved this question from the dune mailing list to the
>>     dune-pdelab mailing list.
>>
>>     Here is the problem:
>>     I computed stresses at the Gauss points resulting from the
>>     deformation gradients in my local operator. Now I wonder, how I
>>     can visualize these stresses via vtkwriter directly from the
>>     local operator, which is on element basis, without a second
>>     calculation of the variable in the driver *_Q2.hh or somewhere
>>     else? Or is there a way to assign these stress values to a vector
>>     container and visualize it in the driver *_Q2.hh?
>>
>>     Maybe it is possible to extrapolate these values to the nodes
>>     within the operator and assign the values to a vector container,
>>     which in turn can be visualized by the vtkwriter?
>>
>>     Is this there a solution for this problem?
>>
>
>     Dear Christian,
>
>     Just to clarify:
>
>     Do you have the deformation as primary variable for which you
>     solve and with this solution you reconstruct the stress?
>
>     If that is true I don't get why do you have the stress in the
>     local operator.
>
>     Or do you use a mixed approach, where you solve for the stress and
>     deformation? But then you still would get the solution not in the
>     operator but in the driver as a solution vector.
>
>     Best,
>     Nicolas
>
>     -- 
>     _____________________________________________________________________
>
>     Dipl.-Ing. Nicolas Schwenck
>
>     IWS, Universität Stuttgart     phone: +49 711 685 64736
>     Pfaffenwaldring 61     fax:   +49 711 685 60430
>     D-70569 Stuttgart
>
>                 email: nicolas.schwenck at iws.uni-stuttgart.de
>                     url: www.hydrosys.uni-stuttgart.de
>     _____________________________________________________________________
>


-- 
_____________________________________________________________________

Dipl.-Ing. Nicolas Schwenck
				
IWS, Universität Stuttgart 		     phone: +49 711 685 64736
Pfaffenwaldring 61			     fax:   +49 711 685 60430
D-70569 Stuttgart

             email: nicolas.schwenck at iws.uni-stuttgart.de
                 url: www.hydrosys.uni-stuttgart.de
_____________________________________________________________________


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.dune-project.org/pipermail/dune-pdelab/attachments/20110720/69c51d9b/attachment.htm>