[Paraview] EnSight Reader/Writer Question

[Brad King]

RAVINDRANATH LANKA wrote:
> This morning I was conducting a couple of tests with paraview. I Created 
> an ensight case file for a simple channel flow problem (462 
> Points/nodes).  It had an interior, 3 OSFs and 3 SBCs (rather 7 
> parts/element sets/cell sets).  Then I loaded the case file into 
> paraview.  I did "extract parts" and this displays a list of parts in 
> the list box, "interior", "OSF: inflow", "OSF: outflow", "OSF: wall", 
> "SBD:inflow", "SBC: wall" and "SBC: outflow".
>  When I go to the information tab, It shows that there are "3243" points 
> and not "462".  This is nothing but 7*462.  Is paraview storing these 
> points 7 times?.

No.  It is storing 7 data sets that are grouped together.  When read by 
the EnSight reader, a single array of points is shared amongst all the 
data sets.  As far as paraview is concerned, it sees 7 data sets each of 
which could have its own points, so it adds up the total.

> Then I saved the data.  There is only one option, which is to save as 
> *.pvd file. I saved the pvd file.  Then I opened the pvd file.  When I 
> clicked on "Filter->extract parts", i didn't see the part names,  But a 
> list like "Unstructured Grid: 200 cells, Unstructured Grid: 10 cells" 
> etc.  pvd is not saving the part names "or" the pvd reader doesn't read 
> the part names.
> 
> When the pvd file is saved, It creates a directory containing a bunch of 
> "vtu" files corresponding to each part. Paraview can independently load 
> these vtu files.  When I loaded each one of these vtu file and tried to 
> display the number of cells and points in each of them, this is what I got.
> 
>  200 cells and 462 points
>  10 cells and 462 points
>  10 cells and 462 points
>  40 cells and 462 points
>  10 cells and 462 points
>  10 cells and 462 points
>  40 cells and 462 points
> 
> It saved all the points in each of the part files.
> Is it how it is supposed to work? or am i missing something.  From 
> efficiency standpoint, is there a better way to do it?

This is expected behavior.  The PVD format is meant to collect multiple 
unrelated data sets into a single file.  Each part can have different 
points.  For example, create a sphere and a cone and then use group 
parts to put them together.  Then save the data.  You will see that each 
.vtp file has different points.

ParaView does not natively have the notion of shared geometry amongst 
several data sets.  Instead a single data set stores a geometry and 
several arrays of data on the points and cells of the geometry.  If you 
can reorganize your data to look like this then ParaView can store it 
efficiently.

-Brad