How to draw a "twisted" spline in NX 9.0 or 8.5

[peiter]

Hi Guys,

Let me introduce myself shortly first;
I'm an Dutch mechanical engineer which loves to tinker on two-stroke engines in his spare time and recently I came up
with the idea to develop my own cilinder.

I've got all the tools / connections such as a lathe, mill, a foundry which is willing to cooporate with me, etc. but it only lacks one thing currently.
At first I thought about reverse engineering the scavenging channels of the cylinder by means of 3D scanning. I am still investigating this possibility and I allready have some scanned .stl files but I think it's perhaps not the most efficient way of designing an optimal cilinder.

While browsing the internet I found some guys which have done the same and they have been able to control the shape of the channels along the swept line. When I try to do this with the command "swept", I can only conrol the surface along the guide.

I hope one of you is able to help me and give me suggestions on how to model this exactly in either NX 8.5 or 9.0

Thanks,

Peter

 

[peiter]

And hereby a second picture just to make it clear.

These guys used solidworks but I think it should be possible with NX as well?

 

[JohnRBaker]

If you sweep along a single guide curve there is an option to use 'Another Curve' to control the orientation of the sweep. Perhaps this will work for you. If not, you could always use a 'Law' which will provide all the possible options that you could ever need, if you know how to use a 'Law' of course.

John R. Baker, P.E.
Product 'Evangelist'
Product Engineering Software
Siemens PLM Software Inc.
Digital Factory
Cypress, CA
Siemens PLM:
UG/NX Museum:

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

 

[peiter]

Thank you for the response John .

Is there perhaps an example in cast that you know of which explains using a law in sweep?
I can't access cast from home right now so I can't check it but I'm really curious and I would love to be able to model this myself.

Thanks,

Peter

 

[JohnRBaker]

Laws are covered in the Help files.

John R. Baker, P.E.
Product 'Evangelist'
Product Engineering Software
Siemens PLM Software Inc.
Digital Factory
Cypress, CA
Siemens PLM:
UG/NX Museum:

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

 

[peiter]

Hi all,

Today I looked through the help function and I tried my best (as shown in the attachment) to design a channel as good as possible. The result already looks more like my desired result but I still can't use the law function, due to the fact that "Angular Law is available only for sweeps with one section string". I would like to control the bottom section and the upper section at the bore so it seems that a law function is no option?

My goal for now is to design the channel exiting horizontally along the YZ plane and following the two curves which are drawn at the exit of the port window, whilst having a smooth transition along the whole body. I am able to achieve this partially, the horizontal exit is ok but the smooth body and correct exit angles still fail.

I hope someone is able to help me or advise me on what I am doing wrong and how I should do it properly.

Thanks,

Peter

 

[JohnRBaker]

This ia about as close as I think you can get (see attached model). Note that I replaced your guide curves with a spline through points rather than a planar sketch curve.

Without any more idea of what the final shape is supposed to look like, it's about as best I can do. BTW, trying to use all three profiles to define your swept shape doesn't appear to be that practical considering how extreme the change in direction would be near the top.

John R. Baker, P.E.
Product 'Evangelist'
Product Engineering Software
Siemens PLM Software Inc.
Digital Factory
Cypress, CA
Siemens PLM:
UG/NX Museum:

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

 

[peiter]

Hi John,

Thanks for the quick response, I'm very grateful for the help you are providing!

I accidently left my laptop at work today so I can't open your modified part this evening, but I can comment on your BTW; I didn't feel comfortable with the middle profile to define my swept but I saw no other possibility to do it otherwise.

In the attachment I added 3D scanned data (from an Aprilia grandprix motorcycle) which i partially converted into surfaces. My goal is to be able to draw such such channels and control their shapes and angles. Hopefully this makes it more clear to you what I am trying to achieve. If you think, after seeing these channels, that I should try to model them in a different way in NX then please give me a suggestion because to me "swept" seems the only viable option.

Cheers,

Peter

 

[JohnRBaker]

Take a look at my example and we'll compare notes tomorrow, OK?

John R. Baker, P.E.
Product 'Evangelist'
Product Engineering Software
Siemens PLM Software Inc.
Digital Factory
Cypress, CA
Siemens PLM:
UG/NX Museum:

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

 

[peiter]

Yes I will, I'll come back to you tomorrow after I had a good look at it .

 

[Toost]

A thousand questions arises...
The scan data, do you want to replicate something existing or are you about to create something new ?

Is the channel cross section area important , such as constant cross section area or an area which is law driven ? ( increasing / decreasing etc )

When you build shapes like this, do not attempt to do it all in one feature, - the result will not be as desired.
( If you look at the first image you uploaded, you can see some weird edges on the left body, the guy who did that model did it with multiple features and in the end he/she had to
use all the tricks in the book to finish. )
If you have high expectations on the surface quality, create one at a time. let each face "do one shape", not "all at one time"
when there is radical shape changes, i.e the face starts as relatively flat but in then end it changes over to a tight bend/curvature, create this as multiple faces or solids.

Attached is an example on the how i would have attacked the task.

Regards,
Tomas

 

[peiter]

Hi Toost, John,

The scandata is the ideal example for me but I can't implement it directly due to different strokes (54 for the scan, 40 for my application) and different bores (54 for the scan, 47.6 for my application). Therefore I have to scale the window heights on the bore and I can only use the axial and radial angles plus the shape in which the channel is build.

Ideally I want to be able to control the cross section area which has to decrease in a small and constant rate from the underside to the port window above.

Both your tips were very usefull to me and it seems that I am getting more grip on drawing the channel in the way as I desire it. The question that arises to me now is how to define the cross section in the bend of the channel, all sketches that I've drawn now are either horizontally (X-Y) or vertically (Z-Y) alligned. Perhaps drawing sketches at an angle relative to Y-X is the best way?


The CAD designing which I do for my job is mostly related to 3-axis milling parts, so until now I haven't practiced these kind of functions enough .

Thanks for the help guys,

Peter

 

[Toost]

Reading the last post about scaling the area.
You can in the swept feature either use scaling by law- Perimeter or Area law
In the case Perimeter, you can set a law such as linear, where the start value is the same as the first section perimeter and the end value is the perimeter of the last section.
In the case Area, you need to enter the area of the first and last section., also there you can apply different law subtypes.

When sweeping this type of geometry i would like to try stay away from more than 2 sections per feature since it is very difficult not to override the otherwise natural and smooth shape which you get with 2 sections.

Attached is an example of the Perimeter law scaling. There is a Intersection curve as the last feature, this intersection is at 50% of the guide and therefore the perimeter of the intersection should be "midway" between start and end perimeter due to the linear law. - The result is off by a few %. One can probably tweak the tolerances / orientation of the section plane etc to get closer. - The intersection curve is a way to check the scaling result only.

Regards,
Tomas

 

[peiter]

Hi Toost,

Thanks for the last example.

The last weeks I have been busy implementing the tips you both gave and I think I'm getting the picture clear for myself on how to draw the channel exactly as I desire it.

I have made silicone molds from a cylinder which is a good basis for my purpose and I'm scanning these molds this Wednesday. The raw scan data is going to be used to make some along the channel sections which I will use to build my own swept channel. By tinkering with the derived sections I plan on changing the shape exactly to my desired shape.

Regards,

Peter

 

[peiter]

Hi guys, in the past weeks I've further finalized the dimensions and performed some 3D measurements on actual KTM channels.
I'm now drawing my own cylinder and modelling the channels so I think now is the right time to check with you guys whether I'm drawing it in the correct way.

Currently I'm working on "morphing" from the window as drawn on the cylinder to bottom of my cylinder plane whilst following the scandata as good as possible with my swept.
I'm quite happy with the part nearest to the cylinder with the bend in it. Now I am trying to connect the horizontal bottom sketch to the other swept but I don't really like the outcome.

If someone has a suggestion on how to do it better, please let me konw, or if I should just continue using splines and be patient .

(The scan data is inserted as a facet body in case you don't see it)