[CIG-SHORT] Question about PETSc

[Brad Aagaard]

Tu Xiang,

Your geometry is CUBIT/Trelis is not created correctly. You did not cut 
the volume with your fault, so there is no interior surface 
corresponding to your fault in your volume. If you look at the volume 
hierarchy, you will see there are only 6 surfaces, corresponding to the 
boundaries of the block.

Exporting the Exodus file gives this warning:

WARNING: The model contains 1665 nodes (and their attached elements)
          that will not be written to the genesis file.
          This may indicate a problem in element block assignments.

If you look at the demo_fault and fault_edge nodesets in ParaView, you 
will see they do not match the nodesets in CUBIT/Trelis because the 
nodesets in CUBIT/Trelis for the fault are associated with your sheet 
body, not the volume.

As we mention in all of our PyLith tutorials, do not ignore warnings and 
error messages unless you know they are benign!

Regards,
Brad


On 08/23/2017 09:41 PM, tu xiang wrote:
> Thank you very much for all the replies.
> 
> Yes, I also notice that there are only 2 nodes in the fault_edge 
> nodeset. I have tested my CUBIT and Pylith with the examples, all the 
> examples can run successfully. I check the mesh generating .jou files 
> again and again and view the Exodus file with ParaView, and I convert 
> the Exodus II file to ASCII file. But I still cannot find where the 
> errors happened.
> 
> I attach all the files of the demo simulation and hope someone would run 
> it and tell me where the errors happened.
> 
> Best regards
> 
> Tu Xiang
>