How to make oval tubing from round

G'day,
trying to make oval tubing from round tubing...
we're trying to make oval tubing (two half circles seperateed by tangential parallel sides) if we try squeezing the tubing we end up with the sidewalls folding in on them selves, and the ends "pointing" themselve - the oval looks like a squeezed figure eight
The tubing (with the same perimeter measurement) is 1.5" OD and less than 1/16" wall thickness and we need pieces in the order of 24" long
On the presumption that a internal mandrel is used - how is it done? Do we start with a round slip fit which gradually changes to the shape we're interested in? Do we push it through, pull it, roll it?
unfortunately we're not looking at needing many of these, so specialist manufacturing isn't an option.
Thanks, Des remove the German antispam device is replying directly

des bromilow wrote:
  Quoted Text Here  
Have you tried looking for an off the shelf flat oval tube of about the size you need. In the UK there are a number of suppliers that do such sections in various materials.

des bromilow wrote:
  Quoted Text Here  
I would try packing the tube with sand before you try to squeeze it.
John

  Quoted Text Here  
Probably you need to start with larger tubing and draw it through a die (as the metal stretches in the drawing, it's easier to form against the surfaces of the die). Alternately, you could start with smaller circumference and apply internal pressure, but that requires a large form to expand the tubing into. Pressure can come from a hydraulic pump, or explosives.
A combination of fill-with-sand and hot rolling might get you something similar to the right profile, with lower applied forces.

des bromilow wrote:
  Quoted Text Here  
Can you define the material? copper, aluminium, brass, steel? and post a drawing of the section? Whichever it is the tube will need to be annealed before forming. i went around a copper tube draw shop a no of yrs ago, and they used a mandrel as well as a die. The draw bench was some 30 ft long, an endless chain with the draw head fixed one end, the drawtongs were attached to the chain by a drop through hook that disengaged by itself when the chain went aroundthe end sprocket. so you could make up your own bench, tho it wold be cheaper to get a draw shopto do it all for you, as the draw tension runs into tons of load. I have a drawbench some 12 ft long that would do the job, as its dsigned to make wire from 1/2 in thick some 2 ft long ection rod. I cast first, then roll down in a power mill. mainly silver Then anneal , pickle and hammer a taper to get the metal through the 1st die hole. repeat. This bench has a 3in geared wheel meshing with 3 ft dia geared wheel on which shaft is another 3in geard wheel engaging with a toothed rack some 6 ft long. Needs 2 men on the handles to use on that sized material. Draw plates are carbon steel some 12 in long 1in thick by 4in wide. One possibility would be to turn up a pair of rollers to the right section and mount them like a clothes wringer.so they could be adjusted up or down with a screwthread in a cross head. Put that in a draw bench,depends how many lengths you ultimately need what are the structural loads etc. Depends also how cost effective it has to be.
Ted dorset UK.

I got fairly good results with copper tubing, by placing a flat bar inside it, and flattening it in a wide vise.
If you still get the center collapsing with a bar, you might try a thinner bar that has slight double convex sides ().
Also, for crushing/flattening/bending in general, the hardness of the materials (inside bar and vise jaws) may have a favorable affect (urethane vs. wood faced/covered vise jaws, for example).
I agree with the other comments, that drawing a mandrel-of-sorts thru the tubing may provide the best results, as in forcing a ball bearing thru a tube to take out a dent.
A set of rollers may also work well, as suggested, with a filler bar possibly, and/or variations of bar shape and hardness of roller coverings.
From what I've read about the forming of close 180 degree fittings (U-shaped tubes), pulling a string of greased ball bearings thru the tube during the bending process prevents the tubing from collapsing (as an example of a procedure).

  Quoted Text Here  
When they bend trumpet tubes, they fill the larger ones with pitch. The smaller ones are filled with a soap and water mixture and frozen. The soap keeps the ice from expanding and splitting the tube.
Look about two fifths down this page: http://www.zacharymusic.com/Zachary_Music/ZTR900Factory.htm
I am not sure these techniques would work to compress a pipe to an oval cross-section. In that case, the volume is being reduced. The point of filling the pipe with incompressible stuff is to keep the volume constant while bending.
Your idea of the collapsing bar in the center is a good one. It might help (although I have never tried it) to also use 2 shallow angle V-blocks to compress the pipe instead of just the parallel vice jaws.

  Quoted Text Here  
In industry, a set of progressively smaller roller pairs compresses the pipe, while containing the longer dimension. About three compressions will go from round to oval without bad distortions.
K&S metals is probably the go-to company for technology on forming brass. Copper might not be too far a feat from brass.
LLoyd