Multi-Section Surface Problem

[philyherb1]

I have a transition surface that fits between a circular shaft surface at the bottom and a constant section surface at the top.
There are 16 planer section curves and 2 guide curves (one on each side). Analysis of each section curve and the guide curves show a (generally) gradual curvature to the curves. Using “Generative Shape Design” and “Multi Sections Surface”, the result is a surface that sinks between each section curve.

The surface should show a constant color change starting on one side red, orange, yellow, green, blue and then the reverse order
back to red.

How can I prevent the issue I am having?

Thanks,
Herb....
CATIA V5 R20

 

[itsmyjob]

I usually have better curvature analysis result with less section than more section.

I also notice that giving a spine to the multi-section surface usually increase quality

Eric N.
indocti discant et ament meminisse periti ​

 

[jackk]

I agree with Eric - too many cross sections.

I suggest:
1. add another guide curve in the middle of the cross-sections.
2. add a spine
3. add a surface at the bottom to impose tangency
4. use the first and last section curve, plus 2 or 3 in between.
5. use the unused sections to verify the CATIA surface is "close enough"

How smooth are the section curves? I was taught "good, smooth curves will produce good, smooth surfaces."

 

[philyherb1]

Thanks for your reply Eric and Jack.
I have tried this surface with less sections and the deviations are quite large. (.007 - .025)
1) I tried adding another guide curve and it showed better surface curvature.
I am going to add a couple more to see if that helps.
2) I am not sure where to place the spine. The sections are all planer to the XY plane.
3) There is a surface at the bottom and the top to impose tangency.
4) Again, deviations become too large when I start to leave some sections out.

My section curves are pretty good. Their analysis shows pretty good curvature and when the curvature
changes direction, it is not abrupt.

Thanks again,
Herb....

 

[jackk]

If all the sections are in XY planes, use the Z-Axis for the spine curve.

 

[jackk]

If you get desperate; another approach would be to sweep a surface in the other direction. (looks like the first sweep goes from the bottom-to-top, so try from left-to-right.)

- create a set of equidistant planes based on the half-circle at the bottom
- intersect 5 or 6 of the existing section curves with the planes
- create new section curves passing through the intersection points
- sweep the surface through these new curves, using some of the original section curves as guides and the circle as the spine.

 

[philyherb1]

Jack,
I tried the spine but there was no change to the analysis.
I did add 5 guide curves.
What a change!

http://files.engineering.com/downlo...f-4745-8123-e1e363d55539&file=Surf1-Side2.png

I will try the sweep just to try it out but I think that my issue has been resolved.
This looks cleaner now!

Thanks,
Herb....