NX3, converting spline segments to bezier curve?

[morans]

Hello, a customer is asking us if we can simplify a design we're currently working on for them. We're designing 2 scroll-like objects with extremely tight tolerances (to compress gas) using a custom grip program that generates the appropriate curves to create said surfaces.

They're asking us if we can take those curves and generate a parametric bezier type of spline to replace the hundreds of surfaces that are generated by the grip program.

has anyone delt with something like this before? can UG/NX even handle something like that? I can't recall that I've ever used a bezier type of spline outside of illustration software. I don't believe this type of solution would be easy to machine either (but who knows)

thanks in advance

 

[JohnRBaker]

Now by Bezier I assume you mean a single-segment spline of whatever degree is needed to accurately reproduce the shape of the string of individual NURB (multi-segment of a fixed degree) curves. Note that NX can support up to a degree 24 single-segment spline (which is generally HIGHER than any other CAD system on the market which can cause it's own concerns when having to convert these curves for use in other CAD systems), which means that it will contain a maximum of 25 points/poles. Therefore, when converting a series of lower-degree (generally cubic or quintic) into a single-segment high-degree spline, some approximation may occur and thus the need to 'fit' the resulting curve, but if the number of curves in the original 'string' was small and their degree low (cubic or no more than quintic) you may be able to get away with this without having a big impact on the shape of the resulting curve.

Anyway, here is what you do. Go to...

Insert -> Curve from Curves -> Join...

...and select the set of preferably at a minimum tangent curves, hit OK and when the dialog comes up asking for 'Resulting Curve Type', select the 'Advanced Refit' option and make sure that you have the 'Associate Output' option toggled ON as this will allow you to edit the resulting curve which you may need to do to get acceptable results (this is much better in NX 6 as all of the fine tuning options are included on the creation dialog rather than on the edit dialog as they are in NX 3). After the curve is created, if you received some kind of warning that some approximation was needed, you will probably need to edit the Join curve feature and tweak some of the options and tolerances. Note that if these new curves will have to be exported (translated) to another CAD system you may wish to inquire as the highest degree curve that they can handle and limit you final edit to that degree otherwise when these curves are converted they will just be broken back up into a series of lower-degree segments.

Anyway, I hope this helps.

John R. Baker, P.E.
Product 'Evangelist'
Product Design Solutions
Siemens PLM Software Inc.
Cypress, CA

http://www.siemens.com/plm

http://www.plmworld.org/museum/

To an Engineer, the glass is twice as big as it needs to be.

 

[hudson888]

There's probably more to what the task entails than meets the eye. You'll probably find that the joined curves splines complain of having included corners. If so you'll need to smooth and approximate the shape within a given tolerance.

There are probably areas where the multi faceted faces indicate a continuous surface, but if it describes a body with volume then there are probably also sides and corners. In resurfacing anything you often need to pick the slab surfaces from the blends extending the slab surfaces you create to intersections and applying the blends last. It is a whole process of working with the geometry seldom so easily answered as to find the one tool that is suited to your needs.

Without seeing such a model I couldn't say whether it is particularly difficult to fix, but just to let you know that the sort of thing you described sounds as if you will need to do more than just be able to join a few curves.

Best Regards

Hudson

www.jamb.com.au

Nil Desperandum illegitimi non carborundum

 

[morans]

thank you very much for the explanations and taking the time to help, it's very much appreciated. Yes, I do mean something more like a single segment line and sounds like it will work perfect for what we're looking to do. However, Hudson brings up a good point. Guess I'll find out tomorrow.

thanks again for all the help
steve