Defining inner surface from STL file

General discussion
1

hello,
I have been given an STL file containing a complex vacuum chamber and internal structure. It has > 200,000 facets. I would like to define just the inner surfaces (i.e. vacuum facing) or remove the outer surfaces (i.e. air facing). Is there an easy way to do this in MolFlow or should I try to persuade the CAD designer to redesign his model to only include inner surfaces ? One problem I anticipate is that as a result of the vacuum simulations (or other simulations), the design may change and then I would have to reimplement the definition of which surface is an inner surface all over again - which I guess could be quite time consuming.

Thanks in advance for any suggestions.
Chris

2

Hello Chris,

This is the most difficult step, to go from CAD models to vacuum models.

You can either do it in the CAD, see examples and seminar here:

https://molflow.web.cern.ch/content/documentation-designing-geometries-molflow

Or take the original model and use the Smart Select tool (which is very fast in 2.9.4 beta) to sleect the outer shell and remove.

That said, usually a good geometry is one that has irrelevant vacuum elements (such as scres, for example) removed.

If you get below 80-100k facets, then it can be simulated.

Good luck, Marton

3

Hello Marton,
Thanks for your help. I am looking into it.

This is probably a suggestion from an ignoramus who doesn’t know very much, but I was wondering whether the Monte Carlo technique could be used to ‘discover’ which facets are internal. If a source of gas atoms is created at a point that is known to be internal to the vacuum chamber, then any facet that a gas atom hits would be an internal surface (assuming the model was completely closed and atoms always reflect off surfaces). One would then use the Select Hit>0 option to select the internal surfaces. I imagine you have thought of this long ago, but I was just wondering what the problems with such an approach might be.

Best Wishes
Chris

4

Hello Chris,

It’s a good idea, we have tried before to “spray paint” the inner surfaces with simply letting molecules hit all relevant facets.

The problem is usually that tiny facets get missed, and deleting even a single “hole” will create a leaky geometry.

You can play around with the method you described in detail:

  1. Invert all facet normals (by default normals point to the solid part)
  2. Outgas from an internal “seed” facet
  3. Select Hit>0
  4. Invert selection
  5. Carefully check if a tiny internal facet is still selected, deselect it
  6. Delete selection (non-hit facets)
  7. Run simulation again and check for leaks

Notes:

Note1) If step 5) was smarter (for example auto-combined with the Smart Select feature to connect missed facets), this method had potential to become the preferred method, so I’ll think about it.
Note2) As you can see, this method was used quite a lot, there’s a helper function:

image
image325×504 5.15 KB

The menu option above sleect “large enough” facets with no hits, thus with a great certainty non-internal.

Many thanks for the suggestion! Marton