Nick Baines
Pictures by the author
YOU MAY HAVE NOTICED that locomotives are not built out of flat plates, they have curvy bits as well. Steam locos in particular have boilers that are mostly circular. Some Victorian engineers had other ideas and produced boilers that were oval or had various humps and dents in them, and of course Messrs. Gresley and Bulleid hid their boilers behind casings, which means that modellers don't have to worry about them. But then the A4s and their ilk had casings with reverse curves and curvatures in two directions at the same time. No, I'm definitely not going to model one of those.
Anyway, I think we can agree that curves are an important feature of just about every loco you are likely to see or model. If you don't agree you can turn the page now because this article is about how to form these curves. If you scratch build, you will have to address the problem. As for kits, a lot of them today come with pre-rolled boilers or even boilers cast in resin (a material I dislike working with), but there are still those that come in a box or a bag of flat sheets, leaving it up to you to form them where it is required.
That is where people sometimes come unstuck. When I demonstrate, there is often someone who comes along with a sad face and a boiler that has been rolled into something other than a circle, or has a seam that does not meet, or (oh dear!) has kinks in it. Usually I can sort it out, unless the kinks are really bad, but there are better ways to do it.
This is not a high-tech, “cutting edge”, “pushing the envelope” method, like my other recent ramblings in these pages. It has been around since the dawn of metalworking, and boilers for the real thing are made using similar methods, but with rather larger and heavier equipment of course.
The first few boilers that I made were done by bending the material by rolling it with an assortment of rods and tubes, while pressing it on to a surface with a little bit of give in it. Telephone directories (remember them?) were good for that. I can hear a chorus of people saying “I still do it that way”, and good luck if you do. It can work and I have seen some very decent models made that way. But it seemed to me a bit hit and miss, and in any case, I always have a hankering for the right tool for the job – that is just me, you understand.
The rolling bars ready for work. Crank the handle to turn the drive roller which is hidden in the photo. The pinch roller above it is held down by the screws on top of each end and clamps the plate to be rolled. The deflecting roller is at the front and can be moved upwards by the two screws beneath it.
The opposite end from the handle shows the gear drive between the drive and pinch rollers. The gears are not essential, but having them reduces the clamping force required on the pinch roller. If you are alert you will notice that a thick workpiece between the rollers will tend to move the gears out of mesh. But I don't roll thick materials, and even if the gears are slightly out of mesh, the forces involved, the speed of rolling, and the frequency with which I use the tool make that acceptable (I say that for the benefit of any transmission engineers reading – everyone else can just accept that it works).
Sketch showing the operation: (a) deflecting roller set low to make a large radius bend, (b) deflecting roller set higher to make a smaller radius bend
So how do you roll a flat plate into a circular boiler, or indeed a conical one if you are of the Swindon persuasion? Let's deal with the circular one to begin with. The principle of bending rolls is shown in the sketch. The plate is clamped between a drive roller and a pinch roller. The drive roller is, er, driven; by hand in our case, or by motor if you happen to be making a full size boiler. The pinch roller holds the plate down and makes sure it moves when the drive roller rotates. In some cases, the two rollers are connected by gears to make sure they rotate together, but if not they rely on the clamping force.
The third roller is the deflecting roller, which can be moved upwards. The higher it is raised, the smaller is the radius of curvature of the rolled plate. Generally, I find that it is best not to attempt a bend all in one go unless it is something like a cab roof that has a very generous radius. Set the deflecting roller low to begin with, run the plate through, and then progressively raise it and repeat until the required curvature is achieved. I also find that to get a constant curvature for a truly round boiler, it helps if you always run the material in the same direction, i.e. don't reverse it, and turn the material round between successive rolls in order to swap the leading and trailing edges.
This shows more clearly the deflecting roller and one adjusting screw in the endplate.
For a taper boiler, raise one end of the deflecting roller higher than the other. The plate will take up a smaller radius at that end and you will get your taper. Getting the right taper requires some careful adjustment with the height settings at each end of the roller, and that is another good reason for doing it a little at a time.
That is the general idea, and the sketch and the photos should fill in the details. And if it all goes wrong, put the material in upside down, set the deflecting roller very low so that when the plate is fed in it unrolls it as far as necessary, and have another go. As with all things, it is a good idea to experiment with a few pieces of scrap to begin with in order to get the feel of it. Once you are up and running you can in short order produce more boilers than you know what to do with. I should also say that this is not the only possible arrangement of rollers, and other arrangements may be found in industrial equipment, but for our purposes this is generally considered the best.
Now to the practicalities. Can you buy the tool? Maybe – a few manufacturers have made them at various times. Keep an eye on the traders’ ads in the Gazette and on the web. The one in the photos I made myself (no surprises there) based on an article by George Thomas in Model Engineer back in the 1970s and republished in The Model Engineers Workshop Manual. However, model engineers tend to work in scales larger than 7mm and the original design was too big for my needs.
The key dimensions are the diameter and length of the rollers. The diameter sets the minimum radius to which you can bend. The materials that we commonly use, such as nickel silver and brass, have some spring in them. So, the smallest boiler you can roll is probably half as big again as the drive roller, give or take a bit. In Thomas's design the rollers were just over one inch in diameter, so the smallest diameter boiler one could realistically expect to roll is 1½ in. or thereabouts, say a scale six feet in O gauge. That is usable if your interest is the US Big Boy locomotive, but it was not for me. I scaled all the dimensions by half (more or less, I made some adjustments for the stock materials I had available at the time) which made it suitable even for the boilers fitted to early locomotives, which tend to be in the region of a scale four-and-a-bit feet in order to fit between the driving wheels.
Once you have rolled a complete circle, how do you get it off? You might be able to spring it off, often that works, but if not the pinch roller can be removed and the workpiece slid off that way.
The length of the rollers should be no more than you need because they will bend under the clamping pressure. You should not clamp them too hard, no harder than is required to make the plate that you are rolling move without slipping, but even that can cause a noticeable bend if the rollers are very long. Mine are about 120 mm which is quite long enough for any boiler I have had to do in our scale. For the larger locos I have to roll the firebox separately, but that is not a problem because I invariably make the boiler and firebox separate anyway.
Here is a short test piece being rolled. The rolling bars are lightweight by industrial or conventional model engineering standards and are usable for brass or nickel silver up to about 0.4 mm thick, but anything thicker is probably asking too much of them.