[vtkusers] What is the most current pattern/reference for using VTK for finite element post-processing?

[scotsman60]

Hello,

I'm using VTK to post-process finite element analysis results.

For "simple" cases - where results data is associated with Nodes or Elements
the standard VTK classes work well.

But finite element results processing can also requires more advanced cases
to be supported. For example "Element Nodal" results and "Integration point"
results.

Element Nodal data is characterized by having multiple result values
calculated per element - one at each of the element nodes. When a node is
referenced by more than one element (the elements are connected) this gives
rise to multiple result values per node. In some cases, it is possible to
simply average the nodal data and reduce the Element Nodal data to simple
Nodal data, and this can be handled directly by the VTK classes -
UnstructuredGrids and PointData.

But in some cases, it is desirable NOT to average the data, and this cases
seems not to fit well with the standard VTK classes. 

One approach would be to create an UnstructuredGrid where each cell has it's
own independent Points. This would require some mapping from the FE data
structures to the VTK data structures and of course would increase the
number of Points in the UnstructuredGrid - increasing memory requirements an
likely negatively impacting performance. Although I believe this WOULD work,
it does seem clumsy......

So my question is this. Is there an established pattern and/or reference
material on the best way to support this type of Element Nodal data nu VTK?  

I did search this forum and found some rather old references to the
"AdaptorFramework" - Is this a good direction to go or is there something
more recent that would make a better starting point?

Thanks in advance,

Doug

 







--
Sent from: http://vtk.1045678.n5.nabble.com/VTK-Users-f1224199.html