I want to create a 2D face that has a hole in it (topologically equivalent to an annulus). So I create FACE1 having the outer edges as its perimeter, and FACE2 having the inner edges as its perimeter, and then subtract FACE2 from FACE1.

However, any mesh nodes I placed on the edges that went into FACE2 disappear, and the edges aren't available anymore to mesh them, and I can't see how to use edge meshing to dictate how the face meshing will eventually go.

I played with "retain" as an option for FACE1 and/or FACE2 but it doesn't change this.

Is this the wrong way to approach making a 2d face that's got a hole in it? Or, what am I not getting here?

Thanks for any help!!!

It sounds like you're doing it right.

How are you creating the faces? Are you importing the geometry for the faces? If so, you may have to heal the geometry to get it to work properly.

After you subtract Face2 from Face1, do you double check to see how many faces there are in the model? Just to make sure it's not creating odd faces. Also, shade the model to make sure you're getting a hole... I've seen it happen where Gambit decides that Face2 isn't perfect, so the cut simply creates an edge on Face1, instead of a hole.

Another option would be to split Face1 with Face2. Make sure you turn off retain and connected. This will create Face3. You should now have 2 faces remaining... Face1 and Face3... one of which is the part you want to keep, and the other is the same as Face2 was. Delete the part you don't want, and you should be able to mesh it without any problems.

Hope this helps, and goodluck Jason

Many thanks, Jason.

I am doing everything including creating the geometry by writing a journal file in a text editor. This is for a parametric study where I need hundreds of combinations of a few simple dimensions, but there's just one combination for now while I debug.

Now I find I have a bigger problem. The edges from which I built FACE2 get renamed; the names I gave them are no longer valid. Boundary conditions I create on the edges become invalid too, perhaps because the names change. Gambit reports "A default (unspecified) boundary entity was created." and "WARN: Boundary Entity INLET does not contain any valid entity and is not written." I can experiment and find out what Gambit renames them to at the moment, but will get incorrect results without knowing it if anything changes about this - who knows?

There is no import of geometry, no little gaps or tolerances to worry about.

After I subtract FACE2 from FACE1, I still have FACE1 and FACE2. I can mesh FACE1 and the mesh does not fill FACE2, it goes around it, as I wish.

I also tried splitting FACE1 on FACE2 and get what look like the same FACE1 and FACE2 as a result. Still the edges that compose FACE2 are renamed according to an unknown scheme, and my boundary entities become invalid.

Maybe I would rephrase my question, "How do I preserve the identities of edges (and associated boundary conditions etc) on the inner perimeter of a 2D face w/ hole?"

Is the general approach to making a 2D face w/ hole, to subtract or split using 2 faces?

actually gambit creates a new face there ..the old face is just imbibed into the face substraction ... and hence if the face do not exist ..it does not make sense that you should have its edge distribution too ... the only hting you can do is .. to understand this rule of substraction .. that is how the edges get absorbed into new face and .. remesh them with new names thus generated .. that is if you have youredge.1 youredge.2 youredge.3 .. after face creation gambit shall make ... face.1 (assuming no other face exist till now) ... and edge.1 and edge.2 and edge.3 (i guess this renumber sall based on the order you chose edges..you can check this i am not sure).. so in this case you have to add more commands to mesh these edges again as olg ones ...

i guess this is the way it shall work

I learned more generally that my problem is Gambit renames the edges used to construct a face subtracted from a bigger face, or the edges in a face used to split a bigger face. Of course I don't control or know these names.

Tech support pointed out the loc2ent(e,x,y,z) function which takes entity type (example "t_ed" for "edge") and coordinates, and returns the name of the entity at that location. Since I know where I created my original edge, I can specify a point on its renamed replacement and so can get that edge's new name.

This seems sort of goofy - but maybe it wouldn't if I appreciated some reason why it's a good idea for Gambit to rename things. In any case, it would work in any situation I can think of.