… to a full 3D mesh so I can export the mesh via a .ply file to a FEA package.
I have been able to import planar surface meshes created in Rhino via .ply files and then extrude them straight quite easily after import. However, the FEA pack doesn’t easily do an extrude-to-curve for example … so I would like to mesh entirely in Rhino3D and then import a full 3D mesh.
I’ve tried numerous approaches, but have not be able to accomplish this.
Admittedly my semantics are not rigorous, but the end goal here is just to generate, with quads, wedges, or others, a network of elements inside the solid with a fine enough resolution to allow an FEA package solve that 3D geometry in terms of basic stress/strain.
Rhino creates wonderfully, controllable planar surface meshes that extrude perfectly in FEA.
I am just curious if there is a way to fill the 3D space bounded by a closed polysurface with quad elements exportable in the Stanford *.ply format.
Possibly there could be a way to do this in Grasshopper?
Possibly there is a way to export the Rhino3D geometry into another package that could create the internal 3D elements?
@blcavender needs a volumetric mesh for 3D finite element analysis, with the volume inside the solid filled with mesh elements. Those mesh elements can be cuboids, each with 8 verticies. One alternative is tetrahedrons, each with 4 vertices. Adjacent volumetric mesh elements share vertices. The volumetric meshes are fundamentally different than the surface meshes used in Rhino.
I am not aware of any methods for generating voluemtric meshes in Rhino. It possibly could be done in Grasshopper but would probably be complex and require some low level coding.
From my experience with volume meshing just minimally complex geometry for FEA the complexity and errors generated with two handfuls of element types and all the error permutations possible with odd angles, min/max element size limit guesses, aspect ratios and most annoying, starting coarse so you can repeat with a finer mesh to see if your accuracy finally is good enough. I get where all that is coming from and frankly bow to those far above my pay grade that made it possible … but I see lots of room to build on their success.
From a tool productivity perspective, if you have a customer in a hurry for a high value activity, putzing/tweaking (from their view) isn’t really knocking their socks off as I would prefer.
It’s pretty easy to have a basic geometry of 2-3 combined parts blow into several million elements after refinement and watch the memmap max ~60Gb w all threads above 95%.
To me, there seems a possibility to simplify elements down to smaller, more robust, standard cube elements that go as tiny as necessary for the curvature, thin dimensions, etc … then flip the resulting terabytes to five $200 NVIDIA Jetson Orin Nano’s for 200 TOPS of crunch and make that trip to the necessary room. :^)
