TiHabanero

Spent a better part of the morning attempting to spread the rear end of an 81 Raleigh Competition GS to 135 spacing. The frame is full Reynolds 531 with butted main tubes. It was set at 125 from the factory as measured. Using a threaded rod and some nuts I spread the stays out to 170mm and got an extra 2mm after relaxing the tension.
Seems Reynolds 531 has a strong memory and returns to natural form easily. It successfully resisted my repeated attempts to coax even 5mm out of them. I was going to get brutal with it, but decided to chill and accept the tubing's limitations on my efforts. One thing for sure True Temper RC2 bends much easier as does good ol' 1020. Done plenty of both without trouble.

unterhausen

That is an interesting thing for you to say because that hasn't been my experience with 531. OTOH, I always have used a pretty long lever.

bulgie 

I think what you're referring to as memory is better described as yield strength. Most structural metals (especially steel) have a range of stress that deflects the metal only elastically, so that when the stress is removed the part returns to its original position. You have to exceed the yield strength to get plastic (permanent) deformation, a.k.a. "taking a set".

The force you need to apply to exceed the yield strength depends on the strength of the steel of course, but also its geometry such as section shape (oval vs round for example), diameter, wall thickness. And how long your lever is.

The distance you have to move it to take a set depends on the stiffness as well as the yield strength. That's why Ti frames take so dang much delfection to cold-set — it's not so much high strength, more the low stiffness compared to steel. Ti builders need their alignment table to put the frame higher up above the surface because at normal height for steel, you can push the dropout all the way down to the surface without it taking a set.

IMHO, the word memory should be reserved for a different phenomenon, easily felt when cold-setting a frame, that a piece of metal that you've just bent is easier to bend back in the direction you bent it from. This effect only pertains over a short distance though; bending it back towards where it started is easier at first but then is soon back to the same strength ovbserved before. I think this is because when the "bulk" of the chainstay (or whatever) has yielded, there are still individual crystal grains with just the right orientation that they didn't yield, so they are straining in the direction wanting to pull the part back where it was. But after only a small bend back in that direction, those grains are stress-relieved and no longer fighting to bend the part back. Whatever the reason though, the effect is real.

That's why I usually cold-set a frame just slightly too far, then bring it back a bit. I do this by feel, and after aligning a thousand or so frames, your feel gets pretty accurate, I think. (I have no numbers to prove that.)

For a typical 531 rear triangle spread, that's probably about an extra millimeter of overshoot-and-pull-back. I think it leaves the frame in a more stress-relieved state that might increase the fatigue endurance. If you overdo the overshoot, then you are most likely reducing the fatigue endurance, so err on the small side.

Glad to have input on that from any "real" engineers or metallurgists. But decades of real world experience shows this method is at least adequate anyway. Frames I've done this to generally don't break.

Mark B

unterhausen

It would be interesting to see what the full yield curve of 531 is like. It doesn't behave like low carbon steel. But the yield curves I found don't show what happens after you let the stress go.

Andrew R Stewart 

I don't measure the deflection during the setting (using that polite word) and I do one side at a time. Just by going on visual memory I'd say 50ish mms per side of deflection at the drop outs to get substantial spread. I never used a threaded rod for bending the back end, always a lever with an end pad against the seat tube. The force I use on the 3is' lever is manageable but on the high side of my 150 lb cyclist's body strength. My flat surface is only 2'x3' and the drop outs hang off one end so there's no issue with "bottoming out" the lever. One of these days I should try to measure the real drop out deflection...

Mark mention of rebound bending (as I once heard it called) is real IME. I agree that it seems that the bend back takes less effort. Sometimes way less I also try to set the width a mm or 2 wider then bring it back a touch, as I do this I am constantly checking the rear end's being centered. Andy

guy153

I use the threaded rod and the way it seems to work is you go out what feels like miles (160mm, 170mm etc.) and it keeps returning to exactly where it was. But at some point it does start to move permanently. As soon as you're in that yield zone, it doesn't take much to move it some more.

I bolt one end of the rod tightly to the right dropout, with a nut either side, and crank the other end out for a bit. A ratchet ring spanner is nice for this. Then I put another nut on the outside of the left dropout, and a locknut to hold it there. Then I wind it all back and measure. If it moved say 1mm and I want a total of 5mm, I'll move the locked nut out by about the difference-- only 4 or 5mm-- lock it again, and then wind it back out until the left dropout hits the stop. This is a good way of fine-tuning how far you go as you sneak up on the right spacing. But as bulgie said if you do go a bit too far it's really easy just to squeeze them back in a fraction just pushing them together by hand, and maybe that does leave things more stress-relieved.

Of course always check the alignment first because you may want to pull one stay a bit more than the other to correct a misalignment. In this case you can't use the rod which will push both evenly (unless the CS are asymmetric, like one is more dimpled or something than the other).

guy153

I think Bulgie is right that this is a property of just two things: yield strength and stiffness. 531 ought therefore to be similar to CroMoly in theory (as yield strength and stiffness are similar). But of course geometry is important too. Round CS are probably harder to space than ROR or round/oval because the oval part will bend a lot more easily.

unterhausen

Just to clarify, I didn't find a full cycle of 4130, which is what I looked for because it's awfully close to 531. The people that do yield tests don't cycle that often, but sometimes you can find hysteresis curves that would show how these materials react to the load being removed. Most materials recover in a linear elastic fashion with the same young's modulus as the base material. There may be some variation off of that, but it's difficult to see on the plots.

TiHabanero

Guy153, I cycled the drop outs with the threaded rod three times. You are suggesting repeated cycles until the tubes conform to my desire. Sounds like a plan to me and will give it a whirl this week sometime. Any guesstimate of how many times you have to repeat the process to where the steel gives in a bit?

unterhausen

cycling doesn't do anything, you are going to have to exceed the yield stress

guy153

The purpose of the repeating is just to sneak up on the right dimension without overshooting-- you have to relax it all in order to measure the final set. I don't think it has any effects on the material. Apart from Bulgie's suggestion of going a bit too far once and bending back to maybe stress relieve I don't think you want to yield it any more than possible.

unterhausen

I was very careless when I drafted this post, sorry. Like guy153 says, it's best to sneak up on the right width. My main point is you are going to have to bend it past where it needs to go because of the material that's still in the elastic regime. Going past and then bending back like Mark said may do something useful, but that's not clear. Some people say that frames will move back after you cold set them. If you consider the microstructure of a cold set tube, there is undoubtedly a balance between the material left in tension that wants to be restored to its prior shape and material that's resisting moving back. Going too far and bending it back may restore that balance to where the residual stress is minimized.

Fredo76

Reynolds 531 is mighty springy stuff. When re-finishing my frame, getting 10mm more spread was like pulling teeth. Rubber teeth...