aerospaceweeb wrote: ↑Thu Jul 29, 2021 4:37 am
If the warping check...
yepp you are absolutely right, the simple check I did for need to flip does not work (clearly had not thought that one through properly).
so I did a new one which checks the lines from the 4 points,
theoretically that one should be watertight (I think
),
however fc continuously throws curved balls, so in practice it fails in fc (seems like there can be misshaps in reported start/end points vs actual orientation), can't help but think that this is one of the things that at some point leads to other type of hard to find bugs in geometry creations.
this time around the selection is a bit more elaborate,
pretty happy with how that works, more specifically the possibility to easily go through a chain of curves with minimal amount of clicks.
have added a test-file, it generally works, but it has geometry in it giving strange results when using certain edge combinations.
anyone interested in bug hunting for core geometry could possibly use that as starting point of the hunt.
although this for now is mainly for play, it is not very hard to see a path to make this quite useful in building smaller type meshes from scratch from within fc where one has easy control over the outcome. it could be that for example fc has trouble importing a step because of some malformed definition of spline profile on a surface (has happened to me at least), then this could turn out to be very useful in creating a mesh from those edges in a simple way (since it just works from coordinates). if fc has trouble reading a surface, so does gmsh since they are using the same things for surface repr under the hood as far as I can gather. when this happens I have never been able to make it work decently neither in gmsh nor netgen.
or another example could be, having a body which is full of holes, and instead of defeature, one makes a shell mesh directly based on edges, that shell mesh is then shipped to gmsh or netgen to make a volume mesh, or keeping it as shells (welded structure?).
or if one for example have an airplane wing, only the outer profile but want to make some quick internal support webs. start with ruled surface mesh like this, and from that make the webs from generated nodes.
these type of workflows allows for full control over element size/position on all edges from fc.
obviously the "femmesh 2 mesh" creates a swiss cheese mesh due to the quads on the ruled surfaces (but it works), otherwise that would have been pretty handy to take a ruled mesh and make a matching geometry with "mesh 2 shape" & refine (if the auto-geometry creation fails - which it for sure will from time to time).
one can of course use "fill holes" and "add faces" to make a proper tri-mesh.
or one could use "Mesh Remodel" add-on, fc seems to allow for selection and use of femmesh nodes to create lines, wires and faces from nodes (although there is no highlighting of selected, but that can be tracked in the "selection view").
my driver for doing this has been to learn a bit about the basics in making a wb, just for learning purpose - that target is now fulfilled - so don't really see myself doing much more on this from now on.
on the multiple mesh thing, bernd will of course know it all (and has answered above - after me authoring this answer offline), but I think...
fc mesh is really intended as an easy to use pre/post, and the first thing to get done was that it is possible to mesh one single solid and show results for that. this I think was the original milestone to get done, and that is probably still largely reflected in the codebase.
as with everything in fc, it however evolves over time, since people are coding small pieces here and there to increase functionality.
one simple example of a missing basic functionality afaik is, there is no way to do a proper torque load from within fc gui, one has to resort modifying the input deck by hand to get it done.
looking forward to see how this evolves.
the download is the "add-on".
edit: removed attachment, look at later posts