Quad. Volume Mesh (Abaqus) - Error regarding set definition

[jessica90]

Hi!

I am using TexGen for my Master Thesis, it's really quite helpful, so thank you. However, I ran into some troubles regarding the quadratic volume mesh export for Abaqus.

I used periodic meshing with Single Layer RVE Periodic Boundary Conditions.

When I start the job (via command line) I get an error message in the .dat file regarding the equations/boundary conditions:

Code: Select all

 ***ERROR: ALL THE NODESETS USED IN DEFINING AN EQUATION MUST CONSIST OF THE 
           SAME NUMBER OF NODES, EXCEPT THAT SINGLE-NODE NODESETS CAN BE USED 
           WHERE A NODE NUMBER WOULD BE VALID

and directly afterwards also:

Code: Select all

 ***NOTE: DUE TO AN INPUT ERROR THE ANALYSIS PRE-PROCESSOR HAS BEEN UNABLE TO 
          INTERPRET SOME DATA.  SUBSEQUENT ERRORS MAY BE CAUSED BY THIS OMISSION

Apparently the number of nodes in the corresponding surfaces do not match up. The .prt file confirms that: FACEA and FACEB have 6924 and 7013 nodes respectively for example. However, when I load the .inp file into Abaqus CAE and check the sets there, I get the same amount of nodes for both FACEA and FACEB (2827).

Has anyone ever heard of a problem like this? How come that the solver finds more nodes in the sets than Abaqus CAE? I did the same calculations (same tg3 file, same element size and same boundary conditions) with linear elements and it worked perfectly fine. Also, I do know that a voxel mesh is a good alternative (already successfully used it), but I would like to be able to compare both methods in my thesis.

Any thoughts on how to successfully solve this issue will be highly appreciated!

Best regards,
Jessica

[jessica90]

Hello again,

I did some more digging and found out that in the set definition of the .inp file some nodes (as far as I can see only middle nodes) are listed twice or even three times. That explains the discrepancies between the .prt file and Abaqus CAE. I am probably going to write a python script of my own for the set definition now, but maybe it's also possible to fix the TexGen function responsible for the issue?

Thanks and best regards,
Jessica

[louisepb]

Hi Jessica,

I will investigate this and get back to you.

Best regards,
Louise

[jessica90]

Hi Louise,

Thank you. The set definition seems to be the only issue with the quadratic volume mesh, after redefining the sets with a script everything worked fine.

Best regards,
Jessica

[louisepb]

Hi Jessica,

I've now corrected this so that the duplicate nodes are removed in the node sets. The code has been updated in the repository so if you've built TexGen yourself then you can get the change now. If you've used the release bundle then it will be available in the next release.

Thanks for flagging the error

Best regards,
Louise

[jessica90]

Hi Louise,

Since I don't have admin rights at the computer I'm working on, it's probably easiest to just keep using my own script. But still, thanks for the quick fix!

Happy Easter,
Jessica

[jessica90]

Hi Louise,

I just tried the new version, unfortunately there is still something off regarding the order of the nodes in the sets. It seems like the nodes are ordered by their ID and not their location, so the bc equations connect the nodes rather randomly. There's no rush in fixing this (since I still have my script), but I just wanted to let you know!

Best regards,
Jessica

[louisepb]

Hi Jessica,

I'll take another look at this. Thanks for letting me know.

Best regards,
Louise