Annular Surface Construction

[JZ_4696753]

This problem involves reconstructing an annular surface using datum points-->curves and the boundary blend surface feature.  I have created a set of circumferential curves and radial curves to be used as the two directions.

The circumferential datum curves have the ends condition set to either tangent or curvature using a smaller curve segment created from the same point set.  I do not know of any other way to generate a closed continuous/smooth curve from points than this.  When I create the boundary blend surface, the result is a quilt with a small sliver surface in the region where the ends of my datum curves exist.  How can I eliminate/avoid this sliver surface and instead generate a single continuous/smooth surface?

[kdirth]

I would use Style to create the surface between the rings using the radial curves for internal curves.

[JZ_4696753]

Hi kdirth,

Thanks for the suggestion.  These are exactly the tips I am looking for.  I will try and report back.

[tbraxton]

Your problem statement is open to interpretation.

Are the radial curves identical i.e. if you cut a planar section through the center axis of the annulus is the section the same everywhere?

It is not clear from your pictures the nature of all of the curves and surfaces. Will a single plane contain all points of the annular surface?

Can you be more complete in defining the geometry constraints? Is it possible to post the points in a file you are using to build this?

[JZ_4696753]

Hi tbraxton,

Thanks for the response.  I apologize for not mentioning that each radial or circumferential datum curve is constructed from datum point sets that originated from a CMM part inspection.  So, yes, each individual curve is unique, and no, the points do not lie in a plane.  The actual topology is quite complex at the micron level and that's what I want to capture and analyze.  I thought radial and circumferential scans would be the preferred way to do this.  I have another totally different method which uses an optical scan and and stl surface, and ultimately want to do a comparison with that and whatever surface reconstruction approach I can come up with for the CMM scans.  If you have some suggestions to try, I would be happy to take those and report back.  My less-than-ideal workaround involves creating half-annular boundary blends by trimming the circumferential curves in half.  But there are a lot more features required to get to that end result and a single surface seems to be ideal if it can be generated.

[tbraxton]

Based on this additional information I would ask you some more questions. Since you are measuring topology in units of microns it leads me to believe recreating this in Creo may not provide the fidelity that you need in the geometry.

Style only supports G0 thru G3 curve continuity and therefore G3 surface connections are the highest possible. Make sure this is not an issue in your case. Creo core modeling curves/surfaces are limited to G2 continuity maximum.

Creo can build surfaces from the curves (most probably). I suspect the surface connections will give some issues if using curves from scan data. Style is a good approach as @kdirth suggested with one big caveat. If you need explicit parametric control over the Style features, you will need to be very methodical about creating construction geometry (not Style features) that will control the Style curves/surfaces. If the annulus is axisymmetric, then this will reduce the workload. It is hard to explain concisely in writing but Creo Style features do not have a lot of options to parametrically control behavior in a way that engineers would typically define the math for curves/surfaces. Creo Style is an evolution of an industrial design application, and the user interface is more for manual manipulation of the parametric controls on these features. There are parameters available for numerical control, but they may not be what you need.

I have used style to create "engineering" surfaces, but the approach is not conventional for the workflow that Style was developed for. As a simple example I would use curves from equations (not style feature curves) to drive surfaces that are created in Style.  The curves from equation capture the engineering design intent and support direct parametric modification of the surfaces.

If you are looking to import a high-fidelity representation surface from scan data to use in Creo and will not need to manipulate the imported annulus, then I would suggest using Geomagic Design X (or similar) to build the surfaces using an algorithm to determine best fit to the point cloud from scan data. You can then import this surface into Creo and use it. The best approach for this is to laser or optical scan the surface and use an algorithm to fit surfaces to the point cloud.

[JZ_4696753]

Hi tbraxton,

This is great information.  I appreciate you taking the time to mention these important points of differentiation between style and core Creo surface features.  It seems reasonable to take a similar approach as you did with style using datum curves.  One difference here is I am using a "curve from points" definition and am accepting the error that comes with a spline curve.  The goal is to capture the overall shape using the points as a guide and would like to repeat this process with point sets representing different parts.

Also thanks for mentioning Geomagic tools.  I am trying to get by without new software, but you are another who has mentioned that specifically and may come into play eventually.  

[JZ_4696753]

Here is the pts file I am working with.

[JZ_4696753]

@kdirth @tbraxton 

I have to admit I have never used the style feature before, so I have very limited knowledge of all the ins and outs but here is some feedback from my first attempt at using it.  My only knowledge base so far is from reading documentation, your remarks, and information I've found on the internet.  Also, it is worth mentioning that I am working with Creo 4.0, but have access to 6.0 as well if there are newer features that could help this situation.

The requirements for a Style surface appear to be significantly different than Boundary Blend.  Because of this, I am limited in how much of my point-->curve data I can use. The following is the best I have come up with so far.  It uses the outermost and innermost circumferential "curve from datum" curves as the primary chain references.  It does not let me select cross curves unless the primary chains are "curve from datum" curves (I can't directly select the original datum curve features outside of my Style feature).

The 16 radial datum curves can be selected as the cross curve chains.  For these, I was allowed to select the datum curve features outside the style feature rather than create and use curve from datum features in Style.

The style feature doesn't like including the interior circumferential curves as either primary or cross curves.  Not sure why this is a limitation, but I can include them in a boundary blend.  Also, I don't know what to make of the surface topology preview in purple.  Half of it looks to have a much finer mesh than the other half.

There's also a suspicious looking portion of the resultant surface that has a small sliver portion.  Not sure why this is happening.

I would be willing to construct some sort of a composite style surface/quilt, but am not sure on how to maintain tangency/curvature with adjacent segments.  

[kdirth]

Opening your points model and quickly creating the curves, I can see that you will have some problems with creating a surface with the curves.  The circumferential and cross curves to not intersect.

Attached is a quick surface that I made earlier with style..

[tbraxton]

@kdirth has revealed an issue with curve connections using your points. If you want to use the points, this must be addressed. You can do it manually or employ a reverse engineering application to generate the curves from the point cloud.

Some very general rules for surfaces created by bounding curves (for core and Style features):

All curves used to create a given surface must intersect.

The curve connections at surface boundaries must be of continuity greater than or equal to the desired surface connection continuity. i.e. a G1 curve connection will not support a G2 surface connection.

Use 4 sided bounds and trim to desired perimeter shape. (this does not appear to be an issue for your current problem)