Sculpt meshing fails when decreasing the element size

Dear all,

I am trying to mesh a complex model using Sculpt. When I use element size, say, 10 I get some of the detached portions and a lot of negative jacobian elements. Therefore, I try reducing element size gradually. When I set element size 8, the mesh is much better but there are still some detached portions and bad elements. However, when I try element size 5, Sculpt fails to run with the following error message:
Primary job terminated normally, but 1 process returned

a non-zero exit code. Per user-direction, the job has been aborted.



mpiexec noticed that process rank 2 with PID 0 on node sage exited on signal 9 (Killed).


Sculpt Unexpectedly Aborted

Sculpt log file written: sculpt_parallel.log

Begin execution of external EPU process…

/opt/Coreform-Cubit-2021.11/bin/././epu -p 8 sculpt_parallel.diatom_result

epu – E Pluribus Unum

(Out of Many One – see http://www.greatseal.com/mottoes/unum.html)

ExodusII Parallel Unification Program

(Version: 4.11) Modified: 2013/08/29

Cannot open file ‘./sculpt_parallel.diatom_result.e.8.0’

ERROR: (EPU) Problem initializing input and/or output files.

End execution of external EPU process

Sculpt log file written: sculpt_parallel.log

Importing exodus mesh: sculpt_parallel.diatom_result.e

Import terminated abnormally

ERROR: Command Failed.

I believe I have enough memory and storage. Any suggestion/advice would be greatly appreciated.

Best,
Hom

Hi Hom,

Here is the model that I created as a proxy for your complex model.
comb

I don’t generate discontinuous elements, but I have “broken” teeth on the comb.
comb_mesh

I did find that turning off adaptivity and just using a finer grid gave me good results.

The tips of the model are about .04 units in the smallest direction. I used a grid size of .035 units which gave me an initial grid size of ~1M cells.

sculpt_gui

The final model has 111 925 hexes.

In summary, I think you might be better served by using a finer grid to start with rather than trying to adapt a coarse grid.

Karl

Thank you so much, Karl!
That was also my observation. I was able to generate the attached mesh decreasing the element size. When I try to further decrease the size, I got the error message:

mpiexec noticed that process rank 0 with PID 0 on node thesage exited on signal 9 (Killed).

I tried doubling the RAM of my machine to check if this is due to the memory but it did not resolve the problem. Therefore, I was trying other options such as adaptivity, pillowing, and so on but I was unable to improve the mesh.

Should I try installing the Cubit on a cluster? :slight_smile:

Your help is greatly appreciated!

Best,
Hom

Forgot to attach the image.

Hi Hom,

I got that same error on my machine. I have 24 GB on my Linux machine and I could watch the memory being used. You may need to use a cluster.

Karl

Thanks, Karl for the suggestion. I think I ran out of the total number of allowed deactivation/reactivation of the license due to several reinstallations of my system. Is it possible to increase that limit? I want to try on a big cluster.

Best,
Hom

Go ahead and try again, I’ve added more deactivations to your account.

Thank you so very much, Scot! I will try. Greatly appreciated.

Hi Hom,

Here is one other thing you might try. Under the Advanced Mesh tab, try turning on Stair-step Mesh and select the Interior (2) option. It won’t be smooth on the surface, but that may model your final product better.

I was able to create the mesh for my sample with a grid size of 1.2M elements in about two minutes on my machine.

Interesting idea, Karl! I will give that a try.

Best,
Hom

Dear Karl, Scot, and Greg!
Thank you so much for your help! Finally, I was able to create a quality mesh for my complex model!
Due to the node-locked license, I was able to run the psculpt only on a single node, which is enough for now. To take full advantage of our powerful cluster, I think I need a floating license? How can I change one of my node-locked licenses to floating? There is no rush for this.

I wish you all happy holidays and Happy New Year 2022!

Best,
Hom

Hi Hom,

When the project is finished, would you be willing to collaborate on a white paper describing your process and showing the results? I think many people would find this very interesting.

Thanks,
Karl

Hi Karl,

Absolutely!

Thanks,
Hom