Suppose that the geometry is not simple enough to be represented by regular geometry primitives.
Then, there are two chooses:
1- use of parametric surfaces (possible as external function)
2- use of import geomerty (i.e. STL file which is externally created)
use of parametric surfaces are in my case relatively slow. So I concentrated on STL import.
I externally create STL file (a surface defined by triangles). I can convert it to a solid in comsol. But I cannot use this solid in difference or union operation later on in a stable manner. I almost always get a domain decomposition error.
In general, even for standard models, one needs to play with the relative repair tolerances to get to desired final geometry (which is not practical it is an automated model). If the geometry creation fail, all automated model fails.
Did anyone of you had similar problems? Any solution suggestions?
Does comsol have any plan to improve the stability of geometry creation?
regards,
dr.ing.- yalcin kaymak
Then, there are two chooses:
1- use of parametric surfaces (possible as external function)
2- use of import geomerty (i.e. STL file which is externally created)
use of parametric surfaces are in my case relatively slow. So I concentrated on STL import.
I externally create STL file (a surface defined by triangles). I can convert it to a solid in comsol. But I cannot use this solid in difference or union operation later on in a stable manner. I almost always get a domain decomposition error.
In general, even for standard models, one needs to play with the relative repair tolerances to get to desired final geometry (which is not practical it is an automated model). If the geometry creation fail, all automated model fails.
Did anyone of you had similar problems? Any solution suggestions?
Does comsol have any plan to improve the stability of geometry creation?
regards,
dr.ing.- yalcin kaymak