There are sadly few 1914 Dennis fire engines still on the road, so when the one owned by Imperial College in London needs a spare part, it can not be ordered from the motor factors and must be made from scratch. Happily, [Andy Pugh] is an alumnus with the required metalworking skills, so in the video below we see him tackling the manufacture of flattened brass tubes for its radiator core.
Forming a round tube to a particular shape is done by pulling it through a die whose profile gradually changes from round to the desired shape. We see him make a couple of tries at this, finally succeeding with one carefully designed to have a constant circumference. The use of CNC machining is something that wouldn’t have been available in the Dennis works in the early 20th century, so we can marvel at the skills of the machinists back then who made the original. Here in 2024 he makes a drawing rig with a geared chain drive suitable for larger scale production.
The video is both a fascinating look at tube drawing and a mind-cleansing piece of workshop observation, and we have to say we enjoyed watching it. If [Andy]’s name sounds familiar to you, this might be because this isn’t the first go he’s had at manufacturing vehicle parts.
Wonderful piece of tooling; cool project, fun video to watch. If Andy Pugh reads Hackaday, a couple of thoughts/suggestions:
The draw at the end of the video made some noises that concerned me. Brass can be pretty brittle.
A single, direct-to-final-shape draw (especially without annealing) may result in tiny internal cracks in the wall of the tube.
I confess I skipped around the video (so maybe this was addressed), but I think this process would benefit from annealing the round tube before pulling it through the die. In fact, it might be best to run the tube through two dies (an intermediate shaping/reduction) and then a final shape… with an additional annealing step in between.
In any event, a more generous application of lubricant during the draw would be beneficial.
I suspect the original draw die only had real machining accuracy at the output end as that is the only the point it is actually required to force the profile you want, and it may not even have been precise at all – easier to make any old die that has roughly the right shape and make everything else those tubes interact with sized to fit the result.
So as long as the the forming shape’s transition is somewhat gradual it can be quite sloppy as long as it is not significantly too small to force that figure 8 of or similar folding/kinking. And if it does that cut a bit more out of the die.
Not to knock the skills that would required, but I suspect in this case its more artistic shaping than precision machining. The precision shown here is required to replicate, but not at all when you control every aspect of the whole assemblies construction so can adjust everything that comes in the later stages to match the results from the die you made.
I’d like to try rolling brass tubing through a rolling mill repeatedly and see where it goes, but I’ve had a lot of trouble with similar forming techniques and having unsupported metal over-bend just because of the stiffness of the adjacent metal that is getting bent, leading to similar failures as his figure-8 problems.
Based on a lot of jewelry work I used to do, I think it would be interesting to draw a strip of sheet brass through a couple of dies to bend in the two tightly curved sections and draw the edges of the brass together, then silver solder the edges together. This probably isn’t a good tactic for a radiator, because of the leak probability, but when we were trying to form weird cross section tubing, drawing from sheet worked pretty well. (Easier with silver, though, because you could draw through nylon or oak dies, which made forming the dies a lot easier.)
I suspect with how copper alloys tend to work harden that progressive approach with force being reapplied again and again to the same spots would be problematic without lots of annealing (which might then have the problem of making the whole thing too soft for comfort).
But interesting thing to experiment with, can also do things like fill with sand/salt/wax before putting it through the roller – when I built my copper tubed watercooled PC long ago I had to do that a fair few times to get pretty results as the required bends where pretty sharp for the wall thickness – it can’t collapse and kink inwards if if that space is already full is a remarkably effective method of getting good final results, and when the filler is easy to get back out afterwards not a hardship to do, at least on small artinsal process. Perhaps not suitable for the production of a radiator on the production line.
Fill the tube with lead before forming. Brass instrument makers do it that way.
Just an idea, (I’m not a metalwork, so I could be way off here), but instead of lead, 3D print some cores before forming.
The idea behind using lead is that it can then be easily melted out and reused. Probably what’s actually used is a lead tin alloy, or nowadays just lead free solder.
I enjoyed watching this, except for his nails. OMG
Don’t watch guitar videos
For some of us, fingernails are tools, not fashion statements.
I suspect the original draw die only had real machining accuracy at the output end as that is the only the point it is actually required to force the profile you want, and it may not even have been precise at all – easier to make any old die that has roughly the right shape and make everything else those tubes interact with sized to fit the result.
So as long as the the forming shape’s transition is somewhat gradual it can be quite sloppy as long as it is not significantly too small to force that figure 8 of or similar folding/kinking. And if it does that cut a bit more out of the die.
Not to knock the skills that would required, but I suspect in this case its more artistic shaping than precision machining. The precision shown here is required to replicate, but not at all when you control every aspect of the whole assemblies construction so can adjust everything that comes in the later stages to match the results from the die you made.
Yeah that can the most fun part about stuff like this… There is no spare. Or if there is, it’s on a farm in the middle of Wisconsin and you’ll never find it. You gotta make it yourself.
For a brief moment I thought that Radiator Core was a new trend.
why not just squish a piece of copper pipe with a piece of flat bar inside? i’ve seen flattened plumbing scraps on the road that look like his radiator core. better formability and thermal conductivity than brass as well.
In London, it’s pronounced “aluminiumnus”.