[Trilinos-Users] Sundance boundary conditions

[Kevin Long]

Hi Andrea,

Shota is correct: there are no sidesets present in the exodus file. I 
ran ncdump and got

netcdf cylinder2D {
dimensions:
     len_string = 33 ;
     len_line = 81 ;
     four = 4 ;
     time_step = UNLIMITED ; // (0 currently)
     num_dim = 2 ;
     num_nodes = 22134 ;
     num_elem = 43786 ;
     num_el_blk = 1 ;
     num_el_in_blk1 = 43786 ;
     num_nod_per_el1 = 3 ;
...

Were there any sidesets present, there would be lines such as these
     num_side_sets = 4 ;
     num_side_ss1 = 148 ;
     num_df_ss1 = 444 ;
(taken from one of my meshes).

I don't know how ices-cfd works, but in Cubit, solid-model surface 
labels don't automatically get assigned to sidesets; you have to 
explicitly assign sideset labels to surfaces in the solid model.

If your mesher creates string names for each surface, the Exodus file 
will store it by both name and number. Sundance doesn't use the string 
names.


Kevin

On 12/22/2011 11:59 AM, Shota Soga wrote:
> HI Andrea,
>
>    I think your exodus file does not have any sidesets.
>
>    Could you make sure that your software properly output sidesets?
>
>
>
> Shota
>
> PS: I edited your exodus file, so I will send it to you via different 
> email.
>
> On Thu, Dec 22, 2011 at 11:18 AM, Andrea Penza <pepe.aero at alice.it 
> <mailto:pepe.aero at alice.it>> wrote:
>
>     Hi Kevin,
>
>     there are no error messages, but simply at the end of the
>     execution the
>     result is zero everywhere (if I start from a null initial guess),
>     as the
>     boundary conditions are not enforced.
>
>     The problem is that I have no Cubit license, and this mesh was made by
>     Icem Cfd. Icem Cdf can put labels, but just as string while Sundance
>     accept only int number as argument of:
>
>     CellFilter left = edges.labeledSubset(1);
>
>     Is it possible to put label and call it in Sundance without using
>     Cubit
>     for making the mesh?
>
>     Ps: Kevin, I can reach your mailbox just from the trilinos-users
>     mailing
>     list because of some spam controll, so sorry people for spamming your
>     mailboxes with my attached file now.
>
>     Andrea
>
>
>     On gio, 2011-12-22 at 09:45 -0600, Kevin Long wrote:
>     > Hi Andrea,
>     >
>     > This is odd. Identifying boundaries by label is the most common use
>     > case, so it's tested thoroughly.
>     >
>     > A few questions/suggestions:
>     >
>     > (1) What, precisely, goes wrong? Does it fail to find any labeled
>     > cells?
>     > Crash? Get cells mixed up?
>     >
>     > (2) You might use ncdump to write the exodus file to text format so
>     > you
>     > can verify that the cells are labeled correctly.
>     >
>     > (3) Could you email me the exodus file so I can test it?
>     >
>     > Regards,
>     >
>     > Kevin
>
>     _______________________________________________
>     Trilinos-Users mailing list
>     Trilinos-Users at software.sandia.gov
>     <mailto:Trilinos-Users at software.sandia.gov>
>     http://software.sandia.gov/mailman/listinfo/trilinos-users
>
>
>
>
> -- 
> Shota Soga
> Graduate Assistant at North Carolina State University
>
> Email: ssoga at ncsu.edu <mailto:ssoga at ncsu.edu>
> URL:http://abdelk.ne.ncsu.edu/ 
> <http://abdelk.ne.ncsu.edu/wiki/index.php/Shota_Soga>w/
>
>


-- 
Kevin Long
Associate Professor
Department of Mathematics and Statistics
Texas Tech University

"The end of fear is the beginning of wisdom"
- Bertrand Russell

-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://software.sandia.gov/pipermail/trilinos-users/attachments/20111222/16df6acb/attachment.html