Pipe with constant cross-sectional area

[brianvonlehe]

Hi, 

I'm looking for tips on how to model a pipe/rod structure (solid rod technically, it's not hollow) that changes its cross-sectional shape but maintains a cross sectional area. The part will be compression molded in carbon fiber, so the constant area is important for even compaction. 

The part changes from a squarish oval with the long side on the y-axis to a squarish oval with the long side on the x-axis. The rod also makes bends along the way. My plan is to use a loft between multiple cross-section sketches, each manually set with the correct area, but I'm worried that the cross-section area will change somewhat as loft is computed around the bends and as the shape changes. I've tried the pipe command but it doesn't seem to have much flexibility for shape changes. 

Is there a more clever or less tedious way to pull this off? I can upload a model later as I make more progress, but wanted to ask before I get too deep. 

Thank you, 

Brian

[g-andresen · Accepted solution]

Hi,

1. pipe has a constant profile.

2. Different cross-sections can be connected with the lofttool.

However, it is not possible to control the respective cross-sectional area along the path.

günther

[davebYYPCU]

Sweep with Twist?  

Will work with the same profile with change of attitude, but not if the profile is actually different.  

Be careful with Loft, and organic shapes, it seems to want to shrink wrap a skin over a frame of profiles and rails.  As if there is slightly lower pressure inside the volume than outside,  sometimes only minor effects, but not what you are expecting.

Might help....

[brianvonlehe]

Thank you. Sounds like my concerns are not unfounded. Maybe the right answer is to adjust the design so a simple pipe or sweep command can be used.

[chrisplyler]

---

@g-andresen wrote:

However, it is not possible to control the respective cross-sectional area along the path.

---

If it was a rectangle, then you could at least manually control the area of each section profile. You could set up a User Parameter for the desired area, manually dimension each Length, and set each Width to be the Area parameter divided by the Length.

You could do this with an ellipses too. Although it's a different formula of course.

And I suppose you could even do it with a rectangle having rounded corners, as long as you kept the corner radius constant and calculated how much area they subtracted from the whole.

[brianvonlehe]

It's essentially a rectangle with filleted corners and manually adjusted to the correct cross-sectional area, which does control it at each loft section, but that doesn't prevent distortion as the loft is carried around bends, even with guide rails.

[chrisplyler]

Right. The only thing I can think of is this:

Manually controlled section profile at both ends of every straight, with the end of one straight being exactly the same as the beginning of the next. Then use separate Lofts to build all the straight sections, which should change linearly, maintaining the area. And separate Lofts (or just Sweeps) again around all the turns, which should be constant since both ends of the turn are the same.

The only downside I can think of, besides the extra work involved, is hard edged changes in direction instead of smooth, flowing geometry.

[brianvonlehe]

Ah, loft the straight sections with shape transitions and sweep the corners with static shape, that might just work. I'll try it out, thanks!

[brianvonlehe]

I made a few experiments on combining sweeps and lofts, so wanted to report the results here. Not exactly successful.

First I made a loft between a filleted rectangle profile and a circle profile, both with the same area. In the middle of the resultant loft, there was a minor 0.22% deviation in cross-sectional area. I did another test with two profiles that have the same number and orientation of faces and fillets (instead of transitioning from filleted rectangle to circle), and the shape transition is perfect with 0% deviation in area. However, this is only for simple connected transitions.

The third test was with G1 Tangent continuity from the lofted section to the swept sections of pipe, and the results are much worse, especially where the transition is more dramatic. Even with matching face and fillet orientation between two rectangles, just shape difference, this test (in the attached screenshot) resulted in a 20% deviation in cross-section area at the midpoint.

It seems like the ultimate answer is going to come down to very careful use of multiple loft profiles that control the area along the transition. I'll have to talk to the molder, but it seems like G0 continuity between sections is not going to be good for machining nor the final product.