No. The wear occurs when the link bends over the chainwheel or off the cog, and it's always in the same range. You can clearly see it when you press the pin out. No wear occurs when the chain is straight because there's no movement, and no wear occurs on the derailler pulleys or at the bottom of the chainring or cog because the load is low. Turning the chain over effectively rotates the pins and bushings, and moves the wear surface to a new area of the pin. It nearly doubles the life of the chain. IIRC that was attributed to Shimano engineers a few years ago.

em

So this pin wear is where exactly? Where the pin passes through the plates? Do I understand your assumption correctly?

If that is the case then it doesn't matter where the wear is. Any wear will cause an elongation and lateral play in the chain. The elongation and lateral play cause the distance between the center of the rollers to increase and no longer match the teeth on the cassette. That leads to poor shifting performance and eventually slipping under load. If I understand your assertion correctly, flipping the chain over isn't going to solve anything.

Wrong again. In a modern derailer chain, the wear occurs between the pin and the inner side plates, which are shaped to bear against the pin on the inside of the hole and to bear against the roller on the outside of the hole. the roller moves around and wear on it doesn't cause much trouble. Chain "stretch" is really wear between the pin and the inner side plates. That wear is concentrated in a range of about 30 degrees on the pin and the corresponding surface of the plate. That leaves about 330 degrees of unworn surface in even the worst chain. If you turn it over, you can move the load to the clean part of the surface.

So if the pin only wears while it travels around the cassette and through the pulleys, and the wear is in a narrow section of the pin, and when a link is traveling either to the rr der or away from the rr der, there is low stress and therefore low wear, then the only time the wear in the pin becomes a factor is when it passes over the cassette and through the pulleys?
So you are saying that the effective length of each link changes at this critical point and that once clear of the cassette and pulleys, it returns to normal length because the worn part of the pin is no longer bearing the load? A different section of the pin is?
If this is true, I should be able to measure a difference between the center of the pins when a link is bent under load and when it is straight and under less load?

This sounds like a job for DIGITAL MIC!

Whew! --OP

For the sake of argument, let's stipulate that wear occurs only as you describe, resulting in a wear arc offset from the centerline of the chain.

1) In one sense it doesn't matter. At least a major component of the wear is along the chain. Wear (24 pin net wear) is measured along a straight chain. Flip the chain over and the measured wear is identical because nothing has changed. Unless you come up with some other way to monitor wear, you're pedaling blind in terms of the actual condition of your chain.

2) Again accepting your wear mechanism, let's look a little closer at the numbers. On a 40T sprocket (average front ring), the link flexes 9°. If wear occurs evenly along the sweep, the wear arc is offset 4.5°. On a 16T (average rear) sprocket, flex is 22.5° with the arc centered at 11.25°. Tension is the same so the wear footprints are additive and actual arc centerline is roughly 8° offset.

The difference in wear between max wear (at 8°) and wear on the chain centerline depends on the radii of the two wear surfaces. Since the pin-to-inner plate clearance is pretty small, the difference between radii is also pretty small, resulting in a wide arc. Empirically (i.e. from memory), the arc is about 90% on a worn chain. The difference in wear over 1/11th (8°/90°) of that arc is going to be very small - on the order of 1/11 of the clearance. Or if you want to project to the new wear arc, it would be roughly twice that (16°/90°) or 2/11ths of the clearance.

In other words, your net gain in wear is at most 18%. Hardly the "double your wear" claimed above.

Nope. I just looked at the wear arc on a pin from a worn chain. (Okay, I'm weird. I keep my pins and pieces bagged together for/with each chain.) It is approximately 180°. Wear difference would thus be 16/180, or 9% max.

Meanwhile, the sprocket teeth - which you can't flip - continue to wear.

And we haven't accounted for other wear mechanisms, such as high chain angles, vibration, etc...

I don't understand what that computation means, but if you have noticeable wear 180 degrees around a pin, you have either already replaced and inverted the chain, or you have a very badly worn chain. The fact that you can see some wear around the rivet doesn't tell you anything about the depth of the wear. You would expect the greatest wear to be somewhere in the middle of whatever the total range is.
None of that matters. It all still occurs in the same limited range.

em

That's exactly right. The differences in one link are small, which is why you measure wear over a long length
of chain. By inverting the chain, you change the process from a link moving from a slightly worn spot to a more worn spot, to a link moving from a slightly worn spot to a less worn spot.

The changes are small, but small changes are what causes excess cog wear.

em

Tension is higher on smaller cogs because there are less teeth to distribute the load over. Tension is also higher with smaller chainrings because you have more leverage than with a larger ring. The actual point of highest wear is probably further than you estimate, but it would depend on a number of factors.

again, you are probably overestimating this effect. The overlap is probably less, and the contact arc is also prbably less than 90 degrees.
So you agree with me that the effect is there, you just have a different estimate of its value.

em

I don't buy it. But it doesn't make much difference to the overall analysis.


No, I'm still skeptical. I can as easily believe that the lead tooth and engaged roller take the vast majority of pressure and thus wear at a nearly zero link angle.


For the value, if the difference is significant then why doesn't flipping a worn chain on an imprinted cassette cause lousy shifting? Or does it? And if it does, what's the point?

Chain was replaced, never flipped. Stretch was about 1/8" as I recall. I took another look with a glass and the length isn't quite 180°. Somewhere between 120° and 150°. Depth of wear is pretty close to the limit of my calipers, but something on the order of 0.02mm between center of wear area and a measurement at right angles to it. (For 1/8" stretch, wear per link would be ~0.062mm, split between the pin and the inner plate. Can't tell which is harder so can't tell proportion, but 0.02 on the pin is certainly in the ballpark.)

Knowing the relative radii of the wearing surfaces tells you a lot about the contact patch (wear area). (This is a fairly simple solid geometry problem, but equations are nearly impossible to type on a forum.) If the radii are close to identical, then any wear will show as a patch approaching 180° long. Depth of wear will gradually decrease toward the edge of the patch. As the difference in radii increases, the patch will get shorter and the slope steeper.

To make this easier to think of, unwrap the cylinder and turn it into a flat surface. The wear patch becomes a V groove in the surface with length up to 1/2 the surface length (180° of pin circumference) with the depth of the groove sloping up from the center toward each end. The groove may be offset from the center of the surface.

To estimate the depth of the groove at any point along it, we need merely to know its length, depth, and slope. While the slope isn't quite linear, for longer grooves it is close enough to treat that way for our purposes. The problem then become simple plane trigonometry - the width of a triangle as you move up one side.

How far up the side to move is the offset of the groove from the center of the surface, or as I calculated it: 1/11 (8°/90°) of the way along one side to the centerline and 2/11 to the center of the new groove (that you'll get by flipping the chain). At that point the depth (wear) will still be 88% of the current maximum depth.

Make the offset a little larger. Make the length a little shorter (though my data shows that you should make it longer). You won't have changed much. You're still looking at a maximum additional wear of 10%.

So instead of a chain lasting 5000km, it will last 5500km? Considering I ride Campy chains (and the wear is actually much longer) is the cost of the link needed to flip the chain justified by the 10% extra use I get from the chain. Also note that this 10% extra use is at the end of the useful life of this chain so the performance will have been significantly degraded.

Basically, for the price of a link I can get another 1000km of less than perfect shifting performance.

Perhaps with a Shimano pin costing less, the extra km might be cost effective.

Chains with Master Links might benefit the most.

If the theory holds up in the real world, flipping your chain basically extends the life of a chian buy 10% at the end of it's useful life. So every 10th chain is free but you have to endure crappy shifting to get it.

I think I'll have one less beer this weekend and go with the new chain. Thanks.

DMF is correct, the wear is the same either direction.

That is your answer...you probably have reduced the chainring and cog life as well. A 20 speed is really not a 20 speed but really about a 15 max for optimal gear use. I think you will find if you set up your own protocol you will be able to extend the life of the drive train parts significantly. Good luck.

I got 10000k from my last chain - an 8 speed KMC Z92. I used a wet lube. This time around I'm trying dry lube, but applying every day, about 70k intervals. Early day yet, but is alot cleaner. Maybe I'll try the flipping over thing at 1000k intervals.

You haven't been following closely. DMF is up to 18% + 10% added life. Even if that's all it is, it still is available for free.
You are right about buying new pins. I wouldn't do that. I use connex links because I'm not confident enough that I can connect a chain using the Campy system.
You don't get poor shifting and you don't wait till the chain is at the end of its life. Once shifting deteriorates, it's too late to replace the chain without replacing the cogs anyway.
The cheapest way to maintain the drive train is the way industrial chain manufacturers recommend: leave the chain in place until it stretches 3%, then replace the cogs and chain together, and never soak the chain in solvent, which just replaces the lubricant with dirty solvent. Dirt on the outside of a chain has no effect on efficiency, although it deteriorates shifting a lot. That plan works great on a single speed, but it results in poor shifting on a derailler bike.

em

DMF didn't say the wear was the same in either direction. He actually estimated a difference in wear in different directions.

em

I think the wear is spread over several teeth, although I have no way of measuring that. I couldn't find anything on that through Google. I think the reason we replace derailler chains at 0.5% stretch rather than 3% is that as pitch increases, more load is taken by a single tooth, causing excessive wear. If you change chains frequently, they run fine on old cassetttes because new chains don't cause much wear.
Flipping the chain doesn't cause poor shifting for the same reason a new chain doesn't. You need to change out chains at very low levels of wear. The point is, if I can rotate any part to find a region of less wear, why wouldn't I do that? It's really no different than when we switch the running rigging on our sailboats end for end. You get more life because the wear points have changed, and you get it for free.

em

I was using the most conservative of the estimates, so I' settled on 10%.

I find a deterioration in shifting performance before the chain is worn.

I change my chains more often because cassettes cost considerably more. I usually get at least 2 chains out of every cassette - 3 if I am doing a lot of base miles and no racing.

No I didn't. The 10% figure was final, and IMO too high.


And that's IFF (if and only if) we assume your wear mechanism as the only significant one.

I did early on, then decided to explore your theory — not endorse it. So far I have seen nothing definitive one way or the other.

Using a new chain on an imprinted cassette does produce lousy shifting. Believe me, I know this one. :D

My overall point here is that even if you are correct, the difference is closer to marginal than to dramatic as claimed above. And that for a gain of <10% chain life there are downsides:

1) You can no longer measure the true chain wear. You can't tell which "side" of the chain is more worn.

2) The sprockets continue to wear. In fact, they probably wear faster since flipping the chain would effectively shorten it, concentrating wear on a single tooth.

3) The "extra 10%" would probably wear faster due to the effect described in 2).


Considering 3) and that there may be additional wear mechanisms, I'd guess that flipping a modern* chain would realistically yield no more than 5% extra mileage on the chain.


* Older chain designs with larger wear surface clearances would theoretically gain more.