reducing triangle count using different meshing method

[shimwell]

For the neutron transport through the mesh to be nice and fast the mesh shouldn't have unnecessary triangles in it.

We have currently many triangles on flat surfaces which can be accurately represented with far less triangles.
We also have lots of equilateral triangles where long thin triangles might be better.

Most meshing software is for FEM and therefore we have a slight difference in requirements

Once option is to reduce triangle count by using "better shaped" triangles
It looks like gmsh supports netgen meshing
Therefore it might be possible to reduce the mesh count and improve the overall surface representation
https://gitlab.onelab.info/gmsh/gmsh/tree/ed587d3307fdf8e677936edc0106ab7cfe49b10e/Netgen

or perhaps the gmesh mesh could be re-meshed to remove unnecessary triangles on a flat surface
perhaps a remeshing could help remove these triangles https://github.com/meshpro/optimesh

[shimwell]

We currently use a meshing algorithm that looks like the right hand side image. But it would be nice to allow meshing algorithm that looks like the left hand side image

Slide 5 of this presentation has a netgen mesh image next to the current mesh used https://files.salome-platform.org/Salome/Common/SUD2015/03_JUS2015-An_advanced_MC_modeling_and_multi_physics_coupling_system_for_fusion_applications-QIU.pdf

[shimwell]

From Yuefengs thesis the approach of meshing/faceting a geometry there makes use of less triangles 👍 and obtains smaller volume differences with the unmeshed geometry 👍.

https://publikationen.bibliothek.kit.edu/1000056066/3861514

currently gmsh makes a mesh like image (a) and it would be nice to allow meshes like image (B)