NX2312 Swept twisting issue

[Carbonated]

Hello everyone, I am looking for ways to optimize a design process that is becoming quite frequent as of late.

I have to bridge a connector to a pipe, but NX refuses to create a surface or body that doesn't twist.
We're currently doing this task with Bridge Curves and Through Curve Mesh, but I've noticed the curves moving along the circular edge when doing re-computations. Making 3x bridge curves for each layer of wire and for each wire end (there are quite a few!) is also an unnecessarily long task in my eyes too. Manually aligning the Swept function by using Points alignment is also susceptible to moving upon re-computation.

Is there any way to use one of the simpler functions like Swept and make sure they are lined up AND robust to re-computations ?
The attached image is the result of an attempt to line up the alignment edges by using patterned planes along the defining spline and splitting the face in two, letting me select two half-circles for the Swept function. Unfortunately, the alignment still ends up wrong (see image).

Thanks in advance,

 

[Toost]

Have you tried to re-select the Origin curve ?
In this case , it seems that maybe a Through Curves feature with the option Normal to end sections might do the job.

 

[cowski]

Is the transition section a constant diameter? If so, it could be done with a bridge curve and a tube.

 

[Carbonated]

Hey there, I'd attempted to delete the thread immediately, having found a solution right after posting. Thanks for the replies.

Regarding Cowski's question, the section is not a constant diameter, but it is a pretty small variation (< 0.3 mm).
As for Toost's, selecting the origin curve did not help (the same origin was being selected).
Through Curves with G1 could also work, but the spline definition would be useful for retaining the original shape and avoiding differences between the layer thicknesses (the solids are subtracted afterwards). I will switch between the two depending on whichever is more appropriate/require the least time.

Regarding the initial problem, enabling "Preserve Shape" did the trick, at least for the sides of the shape. The top part of the tube was still twisting a bit to the side, but was limited by the sides being aligned.

I've solved this issue by dividing the section in 4 curves with the Trim Curve : Divide function, using two planes (tangent and binormal to the center spline). Having four definition points helped significantly reduce twisting.

I've also noticed an issue where if you picked two halves of a circular edge using Stop at Intersection, the selection would still count as 1 selected curve, and the origin would be wrong. This is probably intended behavior, but made me scratch my head a bit. The new curves being fully independent from eachother, selecting all four did not repeat this problem.

 

[mmauldin]

You can break the section curves down to 2 curves instead of one, i.e. 2 arcs instead of a circle.