I'm a fairly experienced SynRad/MolFlow user and wanted to share a subtle problem with options for fixing which can happen in many geometries.
Slides on the problem at: https://anl.box.com/s/qf79k5mehtlw7oqfrjz3
(email me at jacarter@anl.gov)
I have a complex geometry representing 1 sector of a storage ring with periodic boundary conditions. I use SynRad to simulate the synchrotron radiation on the chamber walls and absorbers and MolFlow to translate this into PSD outgassing and estimate the pressure profile in the full sector.
I first create the model in 3D CAD and then export to STL. I import the STL into SynRad and use the global collapse function to reduce the total facets to ~16000. I've noticed that the collapse function can inadvertently create 'seams' in the model when combining triangles into quadrilaterals. The seams are small open gaps which can lead to leaking and artificially low pressures in MolFlow.
My problem is that these gaps aren't usually revealed by SynRad due to less spatial distribution of photon trajectories. I usually do my SynRad calculations first and can forget that it's worth checking for MolFlow leaking first (the leak rate can be 1000 or more times higher in MolFlow!).
A 'non-collapsed' STL model should have a zero leak rate. The collapse function is still necessary to make a large model more efficient and I believe that some leaking is inevitable for large geometries. The question is then what is an acceptable non-zero MolFlow leak rate that minimially influences computed pressures? So far I've accepted leak rates kept below 0.1%.
I have 2 ideas to avoid this problem. For both, first do a global collapse and then run a dummy MolFlow simulation with outgassing to quickly identify the leaking hot spots. Then either (1) reimport the bare STL, deselect the hot spot facet zones, and collapse the rest of the geometry or (2) patch the seams by creating fill in facets using existing vertices if this is not a huge effort.
Thanks,
Jason Carter
jacarter@anl.gov