Sketch Variation Along Consecutive Planes

[John_Doe]

Hey all,

I've got a funny shape that's basically a circle that transitions to a square, the total axial length of which is L.
I've also got 200 non-equidistant axial positions from 0 to L which I have imported as expressions in NX11, e.g., Axial_Pos_1=0, ..., Axial_Pos_200=L
I then created a plane pattern using these positions, so I got 200 planes.
In the expressions tab, I have also defined the parameters that result in the shape change, for each axial position. For simplicity, I'll name two, say Corner Radius and Perpendicular Distance, i.e., at x=0, where the shape is an ideal circle 2*Corner Radius = Perpendicular Distance and at X=L, where it's a square Corner Radius = 0. Therefore, for every axial position I have a unique value for Corner Radius and Perpendicular Distance.

Is there a simple way to automate this as much as possible? e.g., create 200 sketches, one for each plane, and then depending on the plane, the sketch would change accordingly by reading the correct parametric values for that plane/axial position.

Many thanks

 

[cowski]

If you can write down the process and hand it off to an intern and they return some time later with exactly what you were expecting, then there is a very good chance that it can be automated. However, the chances are somewhat lessened if some of the steps are something like "tweak this spline until it looks good".

Now for a little unsolicited advice. It sounds like you are setting up cross sections for one of the swept type commands in NX. If so, my advice is to use the minimum number of sections required to get the overall shape you need. Start with 3 sections: 1 at either end and another somewhere in the middle and add more sections only if you have to. Even without knowing anything about your design, I'm 95% confident that 200 sections is overkill and will result in a heavy model with a lot of ripples and other undesirable surface characteristics. I've seen swepts with 20+ sections that resulted in strange model behavior (wouldn't shell or blends could not be applied to edges, for example) or outright bad geometry (self-intersecting faces or consistency errors). After reducing the number of sections to 4 or 6, the result was both better looking and well-behaved.

Also note that the swept command has an "area law" option that will let you specify how the cross sectional area of the swept changes along the length of the swept feature. This can be very helpful if you are dealing with fluid flow.

Edit: one thing I forgot to mention, the NX sketcher does NOT like zero length lines or zero radius arcs.

www.nxjournaling.com

 

[Toost]

Corner radius = 0 is an old Autocad convention that has caused so many hours of detective work in modern cad systems that the guy who "invented the idea" should be put on hard labor.
In modern cad systems 0 is " division by 0" but in Acad it is or was how one trimmed two lines into a sharp corner.
Trying to select an object which has the physical size of 0 can be tricky, there is nothing to pick.

Anyhow, to add to what Cowski has noted.
The problem will many sections is that NX is a very high precision system, it calculates with 14 or 16 decimals or something internally , and if the section does not exactly fit that position, the surface will do a little turn to pass that section , and we have a ripple.
If we instead let NX interpolate between a few sections , the probablilty is that we get a smooth surface.
the other option is that we open up the tolerance between surface and section to get a smoother surface, but then the point in having many sections is gone.

Regards,
Tomas

The more you know about a subject, the more you know how little you know about that subject.

 

[mmauldin]

Could it be done using a variable blend radius?