Originally Posted by
T-Mar
While the load directions are different between the front and rear sprockets, the load distribution is not the same. The load is distributed over many more teeth on the front. Depending on the gear combination, the individual tooth load (and consequently the individual link wear) on the rear can be 4X to 5x that on the front. This why rear cogs are typically manufactured from steel, while chainrings can be made from aluminum.
Also, while the rollers can rotate, leading to more even wear, the other wear surfaces are the bushes/flanges on the inner plates and these only pivot though a limited range as the chain wraps around a cog. This, in conjunction with the higher loads at the rear, results in bush/flange wear that is predominantly on one side. Once the flanges wear enough, the pin will start to wear, again predominantly on one side. Attached are pictures showing (extreme) bush/flange and pin wear and how it occurs predominantly on one side. Consequently, periodic reversal of chain direction can extend chain life...
I am surprised to find such detailed photos of the chain's inner wear surfaces, I didn't even consider it worthwhile to search for them!
But I don't consider it likely worthwhile to reverse the chain even once during it's service life on account of roller-to-bush wear.
This because once the wear occurs to the more-worn side of the sideplate bushes, it doesn't get put back when the chain is reversed. The lower-wear side gets moved to the higher-wear position, but the wear still contributes to roller freeplay in the loaded direction.
And, speaking of roller free-play, this doesn't contribute to any increase in the chain's pitch, so would not be expected to affect sprocket wear or even the ability to mate with more- or less-worn sprockets. It's kind of irrelevent then, no?
Then there is the severely-worn state of the pictured chain links! While it
would in theory be a good idea to reverse a chain before (as pictured) the bush surfaces were failing asymmetrically, I believe that this chain's wear status is of the order of perhaps 2% or more, which would be about
4X of the recommended replacement point! I can't recall seeing a chain this worn on any modern bike given even a minimum of economy-class servicing, and I've studied many chain link parts over the years.
It's likely that the
pin-to-bush surfaces (independent of the rollers!) have their own wear asymmetry of a different sort, namely that the wear will tend to occur slightly off of the purely length-wise axis, and which would mean that
inside-outing (not reversing) the chain loop might somewhat actually benefit the chain's
effective pitch as it "bends" around the finite radii of the sprockets. I'm not able to measure this however, since all of my chain measuring tools/techniques pull the chain in straight tension as it is measured.
So, for those who are perhaps already removing their chain for some reason during it's service life, perhaps some benefit to the chain's
effective pitch (and thus service life) might be realized by flipping the chain's loop inside-out! But this of course excludes all of the newer, so-called "directional" chains (which aren't so much directional as asymmetric, semantics noted earlier by CliffordK in one of this thread's first replies).
And further, per jackbombay's datapoint, and to the extent that inside-outing the chain loop benefits (reduces) the chain's
effective pitch, I suppose that in the case of a severely-worn chain that this
might actually cause
slipping on the "more-worn" (than the flipped chain's pitch) sprockets that the worn chain was wedded to!