MSRE fails to mesh with gmsh backend

[ebknudsen]

(reported on openmsr/msre#5)
Original text:
I seem to have an issue with MSRE CAD geometry and gmsh used in the CAD_to_OpenMC.

I have followed the installation and operating instructions and I got an error when attempting to use the run.sh script supplied.
More specifically, during the CAD conversion process I receive the error:
Error : The 1D mesh seems not to be forming a closed loop

From some quick googling it could be a problem with newer versions of gmsh from about a year ago but was apparently fixed in the latest version of gmsh, see here.

I am using gmsh version 4.11.0, the latest version, which I assume includes the above bug fix, so that may not be the problem.

Note that I have tried the 7pin test and everything works fine.

From msre run.sh script I tried options 1 and 2 but they both fail at the same point during 2D surface meshing.

I will attach the terminal output log.

error.log

---

[stefan-richr]

Hi, I'm having the exact same issue but with a different package (CAD_to_dagmc). How did you fix this issue? I checked and I'm using 4.11.1

[ebknudsen]

@stefan-richr Sorry I had not seen your comment before. I will try to investigate ASAP.

[ebknudsen]

Hi again,
We have identified some problems with that MSRE-model in conjunction with CAD_to_OpenMC and the gsmh-backend

• there are some overlapping objects here and there
• Being a very large model it ends up being extremely memory hungry
• It has a lot of touching faces which are really a problem with the gmsh-backend.
  We have had better luck using the 'stl2'-backend (simply add backend='stl2' to either the call to solids_to_h5m or run, whichever you use.

That being said, we have improved upon the underlying CAD-model, which we will publish very soon (in a week or two). Using the 'stl2'-backend we have a working model of msre that consistently produces good meshes. So my advice is really - stay tuned for a few days, and you'll be up and running.