Odd meshing of a truncated 1/4 sphere
I have two problems when meshing the attached (in the next post) sphere in 6.6 or 6.5 (I haven't tried other versions.)
The first problem is that when partitioning the sphere I receive two edges on the top line instead of 1 ( error 'Face must have 4 sides but not 6') If I partition the sphere in half along the midline I get a very odd shaped pattern of quads. See the attached image.
My questions are
1) is there a way to tell the 2d algorithim that these 2 edges should be treated as 1?
2) Why is the mesh so oddly shaped when partitioned?
Hello,
About the first problem:
You have this problem because the sphere is actually a surface of revolution created from a half-circle. You can see this arc by exploding your sphere into edges (New Entity -> Explode, then choose "Edge" as Sub-shapes type).
Doing this, you will see three new edges in the study tree inside the Sphere_1 object. The first two are the result of the revolution of the boundary vertexes of the revolution edge (and so are "void"). The last one is the revolution edge itself.
When you cut your sphere, you did it so as to cross this revolution edge. And this led to additional edges in your final object. You can see it by exploding it into edges. You can see that one edge is split in its middle.
Two solutions:
- - Cut your sphere in another position to avoid the revolution edge.
- - Rebuild the faces which are impacted by this revolution edge after the cut.
Enclosed is a dump study in which I rebuilt the two faces impacted by the revolution edge (solution 2).
About the second problem:
It is possible that sometimes the way of construction of the shapes in the Geometry module impact the way they are meshed in the mesh module (especially for bent or twisted surfaces).
One time again, the best way to fix it is to rebuild the faces in a very proper way, so as to "orient" them like in the mesh. For example, by trying to have their iso-curves (the curves that are visible inside the faces when displayed in Wireframe mode) the more parallel possible to the desired mesh.
This problem is also solved in the previous dump study since I rebuilt the revolution surface of your geometry in this way.
Best regards,
William
Hi Peter
I suppose you are interested in getting rather a result than answers to your questions 
so I just propose a workaround: to rotate your sphere by 45 degrees (see a fixed script).
You can get an answer to your 2) question by reading this thread http://www.salome-platform.org/forum/forum_10/984341982
St.Michael
Actually, I'm looking for both! If I just have a fix then I'll just end up doing it again.
Thanks
Previously Saint Michael wrote:
Hi Peter
I suppose you are interested in getting rather a result than answers to your questions
You can get an answer to your 2) question by reading this thread http://www.salome-platform.org/forum/forum_10/984341982
St.Michael
Hello
Only one question seems remaining unanswered.
1) is there a way to tell the 2d algorithim that these 2 edges should be treated as 1?
There is no such a mean currently; but the algorithm does it automatically if it can; unfortunately this case, where two continuous edges are separated by a degenerated edge, is not treated so far.
I will keep this case in mind and will fix it at a favorable occasion.
Best regards
Edward
