FAILURE Shimano Cranksets

[Slack]

I run the warranty and customer service department for a retailer with a number of shops and a turn over in excess of $8.5m a year. Our biggest market is the mid level 105-Ultegra road market and I have never seen a crank fail like this. I would guess we've sold over 5000 bikes with 6800 over it's life span and we've never had a single crank back for arm separation like this. I've seen axle failures and failures from over tightening but never this. I would suggest it must be something to do with enviroment of the market that these are being sold in. I couldn't say what is it, but if it was a common 6800 fault globally we would have seen at least one chainset back with this fault.

[XCProMD]

I don’t know globally... I know of cases in Sweden, Norway, Denmark, Finland, Russia, Germany, Netherlands, Belgium, UK, France, Italy, Switzerland and US.

[pdlpsher1]

In looking at some of the pics, it appears that the two halves of the crank have come undone. And I assume the crank looses a substantial amount of strength when the two halves have seperated, leading the crank to crack and break. So my guess is that the culprit is the gluing process or the glue itself, not so much a metal fatigue failure. Or it's possible that there's some external factor (salt from sweat) that caused a chemical reaction that caused the glue to fail.

A better manufacturing process would be welding the two pieces but of course that would be cost prohibitive and not good looking cosmetically.

[2lo8]

Gluing is done because you can make thinner and more precise shapes. SiSL cranks are also bonded, but I haven't seen many failures like this. (Magic Motorcycle cranks, the precursor to SiSLs howevcer, were known to fail). Welding has issues like inconsistancies, HAZ, inclusions, etc. Welded cranks, like BMX cranks or sweet wings try to minimize the size of the welds. If you've ever picked up a plain (non-DOM) welded tube before, you can actually feel the weld on the inside. Welding tends to not be the most precise joing method and has its own drawbacks and benfits from being overbuilt where the weld is.

The prior method used for hollow crank arms was to cast/forge a tubular section that was open on one end, then pinch it shut through forging, avoiding the complications of welding.

[pdlpsher1]

The prior method used for hollow crank arms was to cast/forge a tubular section that was open on one end, then pinch it shut through forging, avoiding the complications of welding.

I do recall some early Hollowtech cranks don't use glue. Would this crank fall in that category? It seems Shimano has resorted to gluing possibly due to cost and to achieve better performance at a slight cost to reliability. You can clearly see the glue in the latest generation Dura Ace.

[2lo8]

Yes, those older ones start off with the pedal eye end open which is them forged shut. That's why the failures you see with them come from being worn through because of shoe rub, no real structural welds or bonds to fail. The bonded cranks (9000, 6800 onward) tend to be broader, more rectangular, more sylized, and slightly lighter. I'd guess due to the larger outside dimentions they are also stiffer, but I don't have issues with crank stiffness so I couldn't say. Based on these characteristics I'm guessing the inbound R7000 cranks are also bonded.

[pdlpsher1]

Thanks. I'm also aware that their higher-end chainrings are also glued together. I have one of those but it's on a Quarq carbon crank. I wonder if any of those glued chainrings have come apart.

I have read several articles on Shimano and I recall they are very keen on unit production cost. It makes sense since they are the volume leader. And this is why most of their parts use aluminum because aluminum production is highly automated compared to carbon which tends to be extremely labor intensive. And those parts that do use carbon it's the thermoplastic kind not prepreg carbon (with their carbon wheels being the exception). We probably won't see a carbon crank from Shimano for a long time. Shimano did try a carbon crank once but it was killed before it made it to production.

[5DII]

I run the warranty and customer service department for a retailer with a number of shops and a turn over in excess of $8.5m a year. Our biggest market is the mid level 105-Ultegra road market and I have never seen a crank fail like this. I would guess we've sold over 5000 bikes with 6800 over it's life span and we've never had a single crank back for arm separation like this. I've seen axle failures and failures from over tightening but never this. I would suggest it must be something to do with enviroment of the market that these are being sold in. I couldn't say what is it, but if it was a common 6800 fault globally we would have seen at least one chainset back with this fault.

what was the cause of the axle failures?

[2lo8]

Thanks. I'm also aware that their higher-end chainrings are also glued together. I have one of those but it's on a Quarq carbon crank. I wonder if any of those glued chainrings have come apart.

I think is also has to do with the forces a crank arm sees, all sorts of torsional and bending loads probably imparting shear stresses on the bonding. I don't think it is just chemical failure by itself.

[11.4]

Everyone's talking about glue like it's a cheap and inferior alternative. Done properly, it's the strongest way to join two pieces, and this kind of hollow two-piece glued crank design is in fact stronger and more reliable than a solid alloy crankarm.

My experience has been the same as Slack's, above. I've seen a lot of crankarms and not seen a single split except in a very violent crash, accident with car, etc. I'm not sure how some people are having multiple crank failures.

[1415chris]

Well, it still doesn't change the fact that there is a quite number of reported Shimano cranks' failures.

[mattr]

Done properly, it's the strongest way to join two pieces,

that's the problem with bonding. It's very very easy to get wrong.
Even someone with sweaty fingers in the factory touching the wrong spot is going to potentially cause issues. Then you have the hundreds of different environmental considerations. Sweat, salt, road grime, energy drinks, piss, heat, oils, waxes, solvents, degreaser.
Any of them in certain combinations could cause the bond to begin breaking down
That's why the forged ones were so good. Very robust manufacturing process. Very tolerant to damage.

[2lo8]

Everyone's talking about glue like it's a cheap and inferior alternative. Done properly, it's the strongest way to join two pieces

Old hollowtech was one piece though, and clearly something isn't being done properly.

[11.4]

Done properly, it's the strongest way to join two pieces,

that's the problem with bonding. It's very very easy to get wrong.
Even someone with sweaty fingers in the factory touching the wrong spot is going to potentially cause issues. Then you have the hundreds of different environmental considerations. Sweat, salt, road grime, energy drinks, piss, heat, oils, waxes, solvents, degreaser.
Any of them in certain combinations could cause the bond to begin breaking down
That's why the forged ones were so good. Very robust manufacturing process. Very tolerant to damage.

One can use that same argument against any construction technique. Thermal instabilities can kill the reliability of forged ones, as can minor contamination of the alloy, smaller finite design errors, and so on. It's actually not that hard to get adhesive bonding right and it's less technical to implement than alternative techniques. It works on much more demanding and complex manufacturing processes, and cycling equipment is frankly a pretty low challenge. We have carbon rims, brake lever housings, bars, frames, and so on that are equally subject to such contamination issues. If we can get all of them right, we can bond some cranks.

I still don't see enough examples of broken cranks for it to look like a trend. When I was collecting rim flange failures a few years ago, and collected several hundred to study what was going wrong, most of the failures ended up being the result of car crashes. There will always be a certain statistical incidence of product failure and we shouldn't interpret that to mean that a product is defective. Hell, airplanes fall out of the sky for the same reason, and yet we have pretty damned good reliability in our components. I remember watching old Nuovo Record crankarms break like crazy. And 7700/7710 hollow forged crankarms are notorious for breaking on the track at the pedal threads.

[11.4]

Everyone's talking about glue like it's a cheap and inferior alternative. Done properly, it's the strongest way to join two pieces

Old hollowtech was one piece though, and clearly something isn't being done properly.

As I just mentioned in another response above, 7700/7710 cranks had a difficult propensity through most of their lifetime of snapping at the pedal threads. Early 7710 track cranks, manufactured as part of that product series, were downright unreliable. It took at least a couple years to work the bugs out, mostly by increasing the weight of the crankarms and by changing how the spider absorbed stress from the crankarm. They also increased the dimensions of the crankarm around the pedal threads so there was significantly more metal to deal with. I wouldn't hold Hollowtech out as an example of something done better. Statistically I saw results a couple years ago from a test lab that said the bonded alloy cranks were the most reliable of anything in Shimano's history.