developing true length on compound curve?

hi all,
Im trying to develop a shape thats a bit complex to me, for fabrication at
work. What we have is a large copper furnace, its a cylinder lying on its
side. The cylinder is about 20 feet in diameter. I have to build a mouth
to sit on top of it, you could imagine the mouth as being a second piece of
pipe, meeting the cyllinder at a 90degree angle. The mouth itself is
actually a shape with two straight sides (parallel to the ends of the
cyllinder) and two curved sides running parallel to the cyllinder, each
curved side with a different radius.

Ive drawn out the cyllinder and the mouth, in plan and in section and the
results are showing me what i want, but its not really what i want. The
purpose of this drawing is to allow me to make dimensions for making the two
curved pieces. What i really want, is to be able to 'unroll' the curved
piece so its flat, then i can take dimensions off it at will. Imagine, more
simply if you like that im just joining two pipes together, and the second
pipe is joining into the middle of the first one. what i really want is to
be able to slit that second pipe down the middle and unroll it so its now a
piece of flat sheet. any clues how to do this? Im sure there is something
simple i can do here that im missing.

the alternative, is looking at the overhead view, taking lines every foot or
so and drawing them out onto the front and side views then taking individual
measurements.... but this really only gives me true lengths while the piece
is *curved*. Which means the sheet will need to be rolled first, then
measurements projected onto it from a hypotehtical flat line, then the
ellipse drawn, then cut. In an ideal world i would be able to calculate
that ellipse first, mark it out onto a flat sheet, cut it, then put it
through a roller and it would all fit up nicely. The sheet of steel is
about 2" thick which is why it would be easier to cut ot before rolling.

Any and all advice welcome!
Shaun
Australia

 

You need something like:

http://www.plate-n-sheet.com/

http://www.tyharness.co.uk/sheetmetal/#100

http://www.litio3d.com.ar/index.html

 

A while back I used a free utility called unwrap that, although crude,
did work for this kind of thing. Maybe it's still out there somewhere.

 

Here's a page taken from a technical manual explaining how to do the basics.
It's in french, but I'm sure you can figure it out from the graphics.

http://www.box.net/files#0:t:104368

BTW, I've never tried to post a link from there, so it might not work. I'll
check back to see how it turns out.

Dr Fleau

 

Ok. Need to register to Box.net for it to work. Sorry. I'll try to figure
something out.

Dr Fleau

 

Take 2

http://www.box.net/shared/tccvlpxs98

Dr Fleau

 

Thanks for the reply,
Looking at it, this looks pretty much like what i ended up doing, which is
the same way i would have approached it if i were drawing it manually. It
ended up taking about half a day though. This way of doing things, while
accurate is hardly quick, or elegant though :-(


I drew out a top and side view, then a front and back view to the left and
right of the side view. i then took a series of xlines at 100mm intervals
horizontally from the top view and took them across and down to the front
back views, then created vertical xlines down to the sideview and across
again to the front and back views. than i could draw a manual line at each
'crossing point' on the xlines. this gave me height at 100mm intervals, but
still not a true length as the pipe was still 'curved', if you like.

Next i took the top vew and broke the curve at every xline.... by then
measuring the arc length of each of the curved sections i could set up a
spline line and bring the vertical lengths across to it. What it means in
reality though was that i had to measure each one manually, then re-create
this arc length on-the-flat with an offset command. Sound tedious enough?
yeah, it was. at least i only had to do half the curve, because i could
mirror the other half. Given that it was about 3 metres wide, this
ammounted to 15 measurements. I had to work out two developments since the
object is straight on the sides with a different radius at the front and the
back, so i did one development for the front and one for the back. One less
than pretty thing about the final development is that the reference lines
ended up being uneven once it was 'rolled out'.... sincei took them at even
100mm intervals on the curved section, when on the flat they started out at
about 110mm aprt at the outside (but all odd numbers, like 109.7584) and
ended up at 100mm in the center.

If you're interested, i can send you a copy of the drawing, i also have a
self executing solidworks 3D model of the object with an inbuilt viewer if
you'd like to take a look, but its biggish (about 2.5meg).

I'd be interested in hearing any improvements on the way i do this, even if
it is just a small improvement over the xlines, then cut the arc and
manually measure, then offsets technique. While i dont have to do
developments like this very often, it would be nice to find a more automatic
(and hopefeully less prone to human error) way of doing it.

Thanks,
Shaun