Set normal vectors on surface boundaries 3

[jones274]

Hello,
I'm struggling to solve this problem but I need some help because I don't know how to go on.
The problem regards the construction of a couple of surfaces, sharing one edge (in form of a spline curve).
Is it possible to impose the normal vectors of the two surfaces along their boundaries (in particular along the shared edge) on some arbitrary points near the edge?
Thanks in advance!

jones274

 

[Xwheelguy]

To add to John's comments, Siemens has to be loving their position in the auto industry, particularly in the US. All of the big 3 are using Teamcenter now. That had to be a huge win. Plus, it might put them in a great position to keep pecking away at Ford to leverage them off of CATIA. I would imagine that it's not an easy task getting those companies to let go of what they've used for decades. However, they are learning that if they cannot embrace change, then they will probably end up on the losing end at some point.

Not only is Alias being taught to the IDs, there is probably not as many NX users out there that can fully grasp the Class A tools available in NX unless they've come from an ID background. It's not something a typical user can sit down and fully teach themselves like they might be able to do with the more often used modeling tools outside of Shape Studio and Realize Shape. You've obviously been exposed to Class A modeling and the workflows that are sometimes used, so you can imagine the difficulty in trying reteach yourself those dark arts in a different software. Probably not an easy task.

Software choices in many industries is a great and interesting topic, especially when you're able to watch how things have changed over the years. Who would have thought that the former UG would "merge" with SDRC (in terms of software); Autodesk acquiring Alias; CATIA gaining ICEM, etc.

Tim Flater
NX Designer
NX 9.0.3.4 Win7 Pro x64 SP1
Intel Xeon 2.53 GHz 6GB RAM
NVIDIA Quadro 4000 2GB

 

[BubbaK]

In most auto design departments, there are two kinds of surfacing activity: the artsy stuff, and the production stuff.

The artsy folks are creating concepts. These concepts don't necessarily need to be realistic, and sometimes they are deliberatly unrealistic -- the stylist creates a "caricature" to emphasize some aspect of the design. The geometry created in this phase is not subject to any feasibility constraints; it just has to look nice. So, it doesn't matter too much if the shapes are too expensive to manufacture, or the engine won't fit under the hood, or the A-pillars are too thin to support the roof. The geometry is not used for much, beyond the concept evaluation phase. NX can be used to do this sort of work, but the guys doing it don't like to think of themselves as "CAD users", and they probably learned Alias in art school, so NX isn't used much. The market is small, so Siemens probably doesn't care very much about the artsy crew. Though I guess you could say that NX "Realize Shape" is an artsy tool.

Production "class A" surfacing (or "class 1" surfacing, at Ford) is an entirely different game. The geometry is real, it has to satisfy numerous different constraints (in addition to looking nice), and it is used to produce dies and structural body parts. For a long time, ICEM/Surf was the tool of choice in this area. A few companies use Alias, a few use Catia. Toyota and Honda both use internally-developed systems. There are numerous auto OEMs who use NX in this area. I can think of at least two that have switched from ICEM/Surf to NX. As one of the other replies mentioned, there is significant benefit to using the same system in production class A surfacing and in downstream engineering/manufacturing, so the companies who use NX for all of this are typically happy.

The "surfacing" software business does not appear to be a very good one. Lots of systems/companies have gone belly-up. Remember STRIM, and CDRS, anyone? The ICEM/Surf group has been bought and sold umpteen times, and I don't expect Think3 to last very long. Alias got bought, too. Rhino seems to be doing nicely, but it will die if SolidWorks ever gets decent built-in free-form functionality. The problem is that the market is small, but Alias has to have all the same basic infrastructure as NX or Catia -- graphics, UI, saving and opening files, translators, print/plot, etc, etc, etc. So, high fixed costs, and small volumes. The only way out is to charge high prices, which people are increasingly unwilling to pay.

 

[Toost]

Nice thread.

To compliment BubbaK's post, maybe the most intriguing acquisition is the Autodesk buying Alias, considering that Autodesk already owned 3DStudio Max and Softimage, and subsequently has bought Delcam.
3DStudio and Softimage ( and Maya which was a part of the Alias acquisition) are more intended for game development / image rendering than product development, but all 4 (5?) are also capable surface modelers.
(Softimage has since been terminated)

Back to the original question.
trying to actually design something by assigning vectors ( or angles) along an edge like that is like trying to find the matches in the darkness, difficult. - It's hard to guess what angle or vector will look good.
But, NX does more or less have this function, as a part of the Law extension surface .

Attached is a version of the part Tim uploaded, with both methods included. (My version is quite different from Tim's.) NX10.
The Law extension example is eyeballed towards the spline used in the upper example. The angle is ...


Regards,
Tomas

 

[BubbaK]

Yes, going back to the original question is a good idea.

As the previous reply indicated, NX has functions like Law Extension that let you control angle variation. But I doubt that you know what the desired angles are until you see the result and confirm that it looks good. In other words, the correct angle is an output of the design process, not an input.

The ICEM crowd would create this shape by pulling control points (poles) around. In fact, they do everything by pulling poles. So ...
(1) Create a simple surface (one patch, degrees of around 2 x 2).
(2) Split it into four patches, call these p11, p12, p21, p22.
(3) Modify the poles of one patch to produce the non-tangency, while keeping the other 3 joins tangent.
The trick is that moving poles that are two or more rows away from an edge will not affect the edge normals.
(4) If you need a more complex shape, gradually increase the degrees to give yourself some editing freedom. But, the more freedom you give yourself, the more difficult it is to keep the surfaces smooth.

 

[jones274]

Hi,

@Toost: this might be EXACTLY what I'm looking for!! I'm new to NX and didn't know the Law Extension command: I'll start to study it in detail, but it seems it permits to do what I intended to!

@Bubbak: In facts, I have to develop a procedure to impose some angles before building the two connected surfaces.

However, if I understood well, the coomon practice of style designers is to build nice surfaces without bothering with angles, vectors or, in general, without following a "parametric" approach. Is this correct?

Thanks again for your help!

 

[Toost]

As BubbaK said, for "artsy stuff" designers follow their gut feeling rather than specific angles.
in most (?) cases a design starts as a sketch on a piece of paper where the proportions , the "expression" and the "style" is captured.
Then somebody will transform that into a computer model.It might be the designer or a different person.

the computer model will then be iterated into the final product and during these iterations the "parametric approach" is very valuable, but trying to describe the shape of a car fender in numbers, is... difficult.

Regards,
Tomas

 

[BubbaK]

To some people, "parametric" means "driven by numbers". I would say that even production surfacing is not driven by numbers most of the time, and artsy surfacing is hardly ever driven by numbers.

In a larger sense "parametric" might mean "driven by parent-child relationships" (though "associative" would be a better word). This sort of "parametrics" can be very useful in production surfacing. It's nice to have your fillets updated automatically when you move your base surfaces, for example. The automatic update might not give you exactly what you want, but it will still save you some time. In fact, this is one of the reasons that some people prefer NX; systems like ICEM and Alias are not very good at associative relationships.

In your example, I'm still not convinced that entering angles is a good approach. Moving poles allows much more flexibility. But, you understand your task much better than I do.

 

[IvanNX]

I worked in plane industry (wind tunnel experimental aerodynamics, optimization of model surfaces) with A class surfaces, and some of best results are combination of surfaces and solids, actually using surfaces to trim off solids. For example the image of car surfacing I would do with trimming solids with surface, that way I can always relatively fast edit surface and repair the shape if needed. This also give flexibility for surface design as there is no need to patch edges of surfaces, so closing actual solid is not an issue. Main "skill" in this method is transitions between already done surfaces, but if it is done properly there is little or no problems.