Help with surfacing 3

[SiW979]

Hello all.

I'm learning surfacing on the fly and have been messing about with a few practice parts. Please could someone give me a few pointers on the best way to finish the part I have attached to this thread. I can't figure out the best way to patch up the centre, and also even though I have been very careful in creating nicely constrained sketches with tangency and all, a couple of the surfaces near the large rounded end are bit dubious to say the least. I can handle NX5 and NX 6 files for those of you who are on later versions, but Ideally NX 4 as that's what we use ay my company.

Perhaps you would be kind enough to post your versions and give me a few pointers at the same time as to where I could have done things better. Have a nice weekend.

Best regards

Simon (NX4.0.4.2 MP9 - TCEng 9.1.3.6.c)

http://www.jcb.com

Life shouldn't be measured by the number of breaths you take, but by the number of times when it's taken away...

 

[hudson888]

Simon,

That shape is easily done and done well. I may be able to offer some guidance but firstly if you can find me via my website and give us some idea as to the functional criteria.

I would completely change the surface break up. I may have mentioned to you before to avoid coming to a point or an internal tangency at all costs and that every surface needs to have count them four proper sides.

Can we assume that the part is half of a mirrored object?

What if anything are the boundary conditions for the two open ended arms?

You also really need to build the central slab first as the three loose blends are all tangentially lead in different directions.



Best Regards

Hudson

www.jamb.com.au

Nil Desperandum illegitimi non carborundum

 

[JohnRBaker]

I would use an N-Sided Surface. And while it's technically possible to do this in NX 4 or NX 5, it's a lot easier in NX 6 since NX 6 uses the selection-intent tools for both the boundaries and faces. That way you don't need to extract and edit/trim the edge curves as you will need to do with both NX 4 and NX 5 which also makes it harder to create fully associative since I discovered that your model made it hard for me to extract and associatively trim the edges needed to create the N-Sided Surface. However in NX 6 it was a straight-forward task by simply using the 'Single Curve, Stop at Intersection' selection intent method and I was able to select the face edges without the need to perform any extractions and edit/trims.

Anyway, see the attached NX 6 example of my effort.

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.

 

[potrero]

attached is another effort, without using Studio Surfaces; basically just (1) created bridge curves at each of the 3 straight ends, constrained to lie on the face of the bounding surface, then (2) N-sided surface with tangency constraints.

 

[SiW979]

Thanks for your help guys, I can't see them until I get back into work on Monday.

Potrero

Your model isn't attached, pleae could you post it again.

Best regards

Simon (NX4.0.4.2 MP9 - TCEng 9.1.3.6.c)

http://www.jcb.com

Life shouldn't be measured by the number of breaths you take, but by the number of times when it's taken away...

 

[PhilJenkins]

Hi JCB,

This is my effort(NX5). Not perfect by any stretch, but built on top of what you already have. Hopefully it illustrates an alternative way of splitting those surfaces. If I was doing it properly, I would create some additional curves before the surfaces that control the high point created where the 3 stems come together.

As a general point: If you are after high quality Class A surfaces, you need to look at and understand your curves. Avoid using lines with tangent arcs, you need to create good quality splines (you will notice in my version of the model I replaced some of your main 'outline' curves with bridge curves). Ultimately, surfaces will only ever be as good as the curves used to create them.

With any complex surface model, I often find that it can take 2 or 3 attempts just to find the best modeling strategy.

Good luck

 

[hudson888]

Simon,

Please see attached with the revised section I showed you earlier is becomes easier to arrive at a reasonably smooth shape. Lacking the studio surfacing licence for the moment I have merely slapped together the construction.

Sorry that the curve parameters have been removed. All that you really need to know is that I use studio spline for almost everything.

I kept one of your sketches in the file but it wouldn't occur to me even remotely to ever use a sketch in a surfacing application. Normally if I wanted a smooth shape I wouldn't be caught within a mile of an arc either you'll find that degree 5 splines with at least G2 continuity are stock in trade for your A-Class stuff. This is necessarily somewhere below that standard, but the same basic construct that would usually be employed.

Best Regards

Hudson

www.jamb.com.au

Nil Desperandum illegitimi non carborundum

 

[SiW979]

Hudson

Thanks for your post, it is without doubt the best quality surface and yes considerably different than methods I used. I've stuggled to understand exactly how you managed to create some of the surfaces you did, or should I say the splines, but I think I've got the general grasp. I didn't realise the importance of using splines instead of curves, I thought if I ensured all my sketches are well constraned with tangency etc, then that would flow through onto my models surfaces, but obviously not. I've learnt a lot from all the replies in this post, so I'll keep practicing and post my latest attempt later in the week.

Best regards

Simon (NX4.0.4.2 MP9 - TCEng 9.1.3.6.c)

http://www.jcb.com

Life shouldn't be measured by the number of breaths you take, but by the number of times when it's taken away...

 

[hudson888]

Simon,

What I've done with the curves is to make associative construction splines. I used your curves from the sketches in part and appended a couple of points outboard wherever necessary to work with. After than you construct the spline associatively, turn on the curvature comb analysis and then move two end points at the same time using transform so that until the curve takes on the desired shape. You're looking for curves that follow the original fairly closely in the centre and then provide an over sized construction slab out to where the corner would notionally occur.

Missing from the file simply because I did a quick and dirty cleanup were some other construction curves that I probably have left behind. The trimming curves were all constructed on the symmetry plane at first and then projected straight up along the Z-axis. Always project along a vector wherever possible when casting curves onto a surface because then they'll remain at least G1 (tangent) across the meeting edges of the surfaces (presuming the surfaces meet with similar or better continuity).

The rest is straightforward. You can use studio surface if you wish and could possibly even chase G2 continuity. The trick to the peripheral blend from one segment to the next is to break it up into discrete segments and have means to adjust the section curves until the continuity comes in looking smooth using surface analysis with the black lines (zebra stripes) display. Getting anywhere near close to that circumstance might actually approach A-Class.

Best Regards

Hudson

www.jamb.com.au

Nil Desperandum illegitimi non carborundum

 

[SiW979]

Hudson

I've spent a good while looking at your model, and what I can't quite grasp is how you created the top most control surface and also when you have the other overbuilt surface, how you have created the splines to constrain it. Could you shed any light please I would be most grateful.

Best regards

Simon (NX4.0.4.2 MP9 - TCEng 9.1.3.6.c)

http://www.jcb.com

Life shouldn't be measured by the number of breaths you take, but by the number of times when it's taken away...

 

[hudson888]

Simon,

This uses the parametric curve and I went over it adding lines that give order to the construction. I don't use the sketch splines because you cannot tweak the pole and maintain continuity while doing so. The splines with curvature combs have all be manipulated to improve the shape.

This is not so good as to be curvature (G2) continuous but it is a reasonably well built model with a few examples of where it can be improved by using better curves.

Have fun!

Best Regards

Hudson

www.jamb.com.au

Nil Desperandum illegitimi non carborundum

 

[SiW979]

I have taken what you said on board and have had another attempt. I have used a control sketch which has made the model nice and editable by changing some of the angles etc in the control sketch. I hvae constructed everything else using splines even, the straight lines. I'm very pleased with the result, however, whilst it is a considerable impovment on my intial attempt, there are still a few areas where I feel I need to improve. Have a look anyway and let me know what you think.

I don't think the Formula 1 designers need worry......yet!

Best regards

Simon (NX4.0.4.2 MP9 - TCEng 9.1.3.6.c)

http://www.jcb.com

Life shouldn't be measured by the number of breaths you take, but by the number of times when it's taken away...

 

[hudson888]

Simon,

If that's your design intent then I can't really fault the construction. Had it been my design intent I would have used different construction methods aimed at getting the surfaces as smooth as possible.

Elements that allow for progressive an adjustable shape change are included in my model. If you want to see something cool adjust the lengths of lines 48 and 49 with the black and white striped surface analysis turned on.

You can try and make surfaces parametric but there are limits to what can readily achieved. That's not to say that you wouldn't occasionally tie elements together with a parametric construct in order to get an adjustable set of curves that you can iterate towards a smoother solution.

You're better to ditch the Bridge curves for studio splines. I explained why earlier. I actually adjust the poles manually to improve shapes.

If you have access to the studio licence use it.

Unless there is some totally essential mechanical requirement to use an arc in any design I'll create a spline even if it is relatively similar in apparent shape to the arc. Try to create meshes of curves where possible to be single segment splines of the same degree. This feeds into the direct control that you'll have over surfaces that you build.

I would not be surprised if Formula one car surfaces aren't too different from what you and I are looking at in this example. Their main concern is functional and aerodynamic. Automotive A-Class however is the principal discipline where a higher standard of smoothness is sought.

Best Regards

Hudson

www.jamb.com.au

Nil Desperandum illegitimi non carborundum