Ribs on cylindrical surface 4

[Egor K]

Cylindrical surfaces have two lines that, in display modes:
"Shading with Edges" - visible
"Shading with Edges without Smooth Edges" - not visible
when importing model (no difference between CATPart or STP) in Ansys, cylindrical surfaces are divided into two and, for example, there is a problem to set preload (Bolt Pretension), direction of which is supposed to be set by one cylindrical surface. If you have SpaceClaim, you can remove these extra lines and merge two surfaces into one, but I don't have it.

Why are there ribs on cylindrical surface at all? Is it possible to remove these ribs with Catia settings?

 

[3DDave]

One of the fundamental problems for managing surfaces in CAD is they often need to be able to have a coordinate system - not the typical XYZ version, but one that conforms to the contour of the surface. This is referred to as the uv coordinates - and they scale along opposite edges of the surface from 0 to 1 where 0,0 will be one corner of the surface and the (1,1) will be the opposite corner. Interpolation is done by setting the coordinate along opposing edges and then along the matching curve.

So, for a square flat surface it can be visualized like a checkerboard or graph paper. For a cylinder surface, roll that graph paper into a tube with lines going vertically from cap to cap and wrapped around horizontally.

In the case of a cylinder with sides and end capping surfaces one coordinate is aligned with the axis and the other is aligned with the curvature. The one that is aligned with the curvature is driven by the edge you see. Now, you may think, that could be a single split like on a rolled food can - and for a surface all alone that would work.

But look at the cap surface. It would have only one edge curve and so one would not have anything to interpolate against. In the data, the two edges shared with the two side surfaces become the basis for interpolation. The cap likely still has 4 edge entries, but the two at the shared vertex are flagged as zero length or some other special notation.

In other schemes one would have difficulty generating a simple, general scheme for arbitrary surfaces that would allow generating a coordinate for ever point on the surface. Some software hides this, but I expect most CAD programs do this surface division behind the scenes. The only ones I know don't are Constructive Solid Geometry modelers that use equation driven solid descriptions and don't generate interactive elements like edges and vertex points.

This is mostly done in order to subdivide the surfaces into triangles so the graphics card can manage the shading, but it also makes for general purpose intersection algorithms rather than having one for sphere to sphere, another for sphere to cylinder, yet another for cylinder to cylinder, and then one each of them to plane.

 

[JohnRBaker]

Shapes like cylinders, spheres, tori. cones, etc are what's referred to having 'periodic' faces, or ones that wrap around onto themselves. Now the more capable systems like Catia, UG/NX and some others, they have no problems representing shapes like that as a single face. However, there are a large number of lessor systems which cannot represent a periodic face as a single object, and therefore they represent them as two or more faces with boundary edges. Now sometimes it's not the system the the model was exported from, but rather the scheme used to translate the model from one format to another, which is the culprit. Sometimes that's where those extra faces and edges get created. In that case, there's really nothing that can be done unless there's some kind of option or post-process which will automatically remove the 'smooth' edges and return the faces back to periodic. That's how I ever did it when I was working on models in UG/NX which had been translated from another system. After opening the model for the first time, I would simply run the 'Merge Face' function which would remove any edges which were part of what then became a true periodic face. Note that there were other situations where you might find a 'smooth' edge which could be 'Merged', that was not part of a periodic face, but at least with models translated from other systems, the periodic face issue was the most common source of these 'smooth' edges.

Anyway, I hope that helped explain what's going on and why, and what might be the source of this issue.

John R. Baker, P.E. (ret)
Irvine, CA
Siemens PLM:
UG/NX Museum:

The secret of life is not finding someone to live with
It's finding someone you can't live without