Hidden speed: crankset fatigue and benefits of swapping DA-9000 for 9100

FINDINGS: Replacing a 53/39 DA FC-9000 crankset with a DA FC-9100 resulted in  1.5k/hr gain flats (approximately 135 seconds for 40km above 39-40 km/hr); 2 km/hr uphill at 3-4% gradient and 3 Km/hr downhill gradient at -2%. Replacement resolves several mysterious physiological ailments such as plantar fatigue and calf strain assumed as saddle or shoe issues. The manufacturer’s pursuit of light alloys may introduce drawbacks that can affect item lifecycle and performance. Crank fatigue can also induce the rider into believing that frame may be deficient, BB at fault or the bicycles itself becoming outdated. Crank fatigue and lifecycles are important subjects for future research.

BACKGROUND. May 2020, 22,000 kms noticed one too many chain drops, replaced chainrings. No more issues, quiet system with Campy Record 11 speed. August 2020, notice plantar fatigue particularly right side, and foot numbness, after very high intensity rides with sustained efforts above 40m/hrs. Began assuming BB or frame fatigue, shoe carbon fatigue as suspects. Ordered the 9100 mainly to move outwards the chainline, as speed and power were going up, PRs times dropping, and lower gears angled the chain. September 2020, DA-9000 accrues 26679 kms by Sep 2020. The systems runs a SRAM HG-1190 cassette, with Campy record 11 speed chains, predictable 3700 kms to reach 132.4 mm/6 link spacing (Campy max is 132.6). The result is superior to any native Shimano system; quiet, powerful, predictable, 120-150 ms per rear gear change, aided by Yokozuna state of the art cables. No cable oxidisation or elastic deformation on these cables at 15,000 kms. Pedals: Speedplay Zero Stainless, 27.000 kms. Rider weight, 83kg with accessories/clothing/water shoes etc.

On the Columbus, observers may notice my Ultegra RD-7900 derailleur. July 2020, the RD-9100 stripped the cable bolt plate internal thread, rendering the RD disposable after 11,007 kms. It took 2,000 kms of odd shifting to diagnose, and the bolt was no longer threading to 4-6 N.M. torque. Got the 9100 from Jensen USA yesterday, but a week ago installed an Ulterga that arrived faster. As with other DA components, the internal thread suggests a softer alloy. A similar fatigue was observed on all DA 9000-9100 'spyder' spindles casettes; Soft plastic core, soft metal, soft chains. The use of SRAM XG-1190 resolves these issues, as did the Campy chains. Superior metallurgy, 0.75km/hr gain, with Shimano front and rear mechs with a Shimano crankset.

METHODOLOGY: Speed, cadence, gear and perceptual/sensorial. No advanced deflection instruments used. Accurate GPS based speed capture, flats up or downhill, are an accurate metric of crank to crank speed change. I log the mileage for each component.

OBSERVATIONS: Sep 24 2020 install DA FC-9100 and go for a test ride at 17C. Immediately feel the effects. Noticeably faster, less standing in the saddle to push, noticeable more torque. Clear speed gains, reaching on flats 53/11 50-54 km/hr, whereas past two years I barely dropped below 12t. No more physiological feedback following the first high intensity rides; neither plantar nor calf fatigue post ride. More spinning,  less ‘pushing’. No pedal deflection or deformation noticed. Repeat results a second time, at 18C. Third attempt, a longer distance ride, done at 23C; lower air density and better grip. In each case, crush PRs from previous weeks. Several more hundred kms confirm the delta. If blindfolded, i could have sworn having a different bicycle. No structural defects are naked-eye visible on the outgoing FC-9000.

DISCUSSION: Shimano claims the 9100 as stiffer, and it physically features a 4th retaining bolt. It is indeed stiffer, but the delta between a new 9100 and a used 9000 appears to be substantial. 135 seconds gain is just … a lot? Chainring bolts are original; despite requesting the LBS to replace them at 22,000 kms,  the LBS opted to reuse them. They are a suspect intervening variables but do not explain fully left and right loss of power, and more mm of flex on the drivetrain side. Furthermore, comparisons with an Ultegra 7800 crankset (secondary bike) at 15,000 kms reveals some stiffness variations as well. The 7800 is stiffer than the used 9000, but, perceptually, less stiff than the 9100.

Crank fatigue and catastrophic failure is discussed throughout literature, but causes remain elusive to riders, be it most likely known to manufacturers. These keep that information to themselves to avoid the predicable aftermath of recalls, PR issues, accountability or liability. In the literature, testing methodology reveals clear deflection on new equipment as well as engineering weak spots (Fairwheelbikes 2019, Loveridge 2020, Peak Torque 2020). However, available research does not offer a long-term deflection test. If done, it remains in house with Shimano, SRAM or Campy.

Occasionally cranks fail catastrophically, causing serious injury. Shimano’s answers to crank failure are available online. Their intention to improve product quality is not in doubt. The pace of resolution, however, leaves to be desired. Irrespective, Hollowtech weak points include glue adherence failure or metal fatigue, either through torque cycles or through oxidisation (Peak Torque 2020). IMO, that is unnacceptable- we do not accept it in vehicles, we should less in more vulnerable critical cycling components on bycicles matching traffic at 40-50 km/hr...

In my case, insidious loss of power transfer, despite PRs improving weekly had me deep into researching BB, frames; an exchange with Hambini regarding BB swap; and planning the next frame, all but delayed by COVID-19 and a 50 week marathon towards the completion of a graduate research project. Nonetheless, learned as much as possible about BMC Teamachine, Tarmac S-Works SL-7, F-12, Scott Addict Ultimate, V3RS or reinforced, custom Sarto Lampo. Stiffness is important, as at 83-85 kg with accessories, am different than a 65-70 kg rider... Either case, when bikes are so expensive, having the right kit and drivetrain is primordial.

The primary hypothesis is that Hollowtech aluminium alloy and its glued crank construct is insufficient and finite regarding sustained high torque applications and long mileage, or sufficient for a low count lifecycle. A secondary hypothesis is that Shimano chooses this construct as the best bang per buck and highest profit, possibly cheapest forging or flow forming method. Alternative materials, e.g. composites and laminate layers can yield multifold modulus and strength but are more expensive to make and would eat profit margins.

The DV effect is crank fatigue as a factor of one independent variable (metallurgical and engineering properties) and several intervening variables, e.g. moisture, cycles, load etc.

DV (fatigue)= IV (engineering, materials, weak spots) + iiv (Torque) + iiv (cycles and mileage) + iiv (environmental)

A third hypothesis is that improvements in frame technology and stiffness results in a higher power transfer and more elasticity deflected into the crankset, therefore, current high-power output and crankset torque is something not necessarily anticipated 10+ years ago when developing current cranksets. The more a frame begs for power and speed, the more the rider delivers, which, correspondingly, all passes through the crank and BB. Rider weight and torque evidently play a role. If the framset is less elastic, the drivetrain will absorb more energy.

To its credit, the DA FC-9000 has never fractured, and it likely performed as intended. Nor has my FC-7800 Ultegra failed; caveat, it runs on a slower TCR Advanced, and never pushed it as hard. 

Opinions on alloy fatigue or failure vary. First, are these things FORGED or flow-formed? How are they forged if so? Several high mileage riders argue that item failure is what it is, and thar it is normal. We all agree that each allow will have its own lifecycle item. However, we may agree that fatigue is a statistical outcome, but not catastrophic failure. Nor should a surreptitious performance drop occur around 20,000 kms. In 2020, there is enough engineering science to A- ensure fatigue testing; B- prevent catastrophic failure, especially one related to epoxy and C- have these last well beyond 50,000 kms without any noticeable increase in deflection. At the very least, a safety sensor could be built in, something enabling an auditory cue (metal or electronic sound) that the item is done.

Mileage will undoubtedly vary, professional riders with multiple bikes and annual part replacement may never reach the lifecycle limit. Most riders, however, accumulate the effects with mileage.

RECOMMENDATION: High mileage high torque riders should replace chainrings no later than 10,000 kms, and cranksets 20,000-22,000 kms. High mileage riders should also assess FC-9100 material fatigue. I recommend that those with the technical means/lab access demonstrate empirically the effect of variables and bring their data to manufacturers. Manufacturers are encouraged to develop longer lasting forges, laminate composites or better-quality critical components such as cranksets.

FUTURE RESEARCH:
1. Effects of crank fatigue on power metres? Most likely item fatigue eschews power readings..
2. BB Press Fit versus threaded play effect on crank spindle?
3. Frame stiffness effect on crankset fatigue?
4. How to test and detect crankset fatigue?
5. Failure rate delta between sizes; 50/36, 52/36, 53/39?
6. Spindle or drivetrain fatigue?
7. Larger diameter spindle- any advantages? 22, 25, 30 mm?
8. Forged vs cast flow cranks? 
9. Is Shimano DA the stiffness standard?

References:

Fairwheelbikes. “Road Bike Crank Test.” 2019. Accessed at https://blog.fairwheelbikes.com/reviews-and-testing/road-bike-crank-test...

Loveridge, Mathew. 2020. “Understanding an unusual Shimano crankset
failure- An Ultegra 6800 Hollowtech crank breaks – what can we learn?”
https://www.bikeradar.com/features/shimano-crank-failure/.

Peak Torque. 2020. Why do Shimano cranks keep failing?” Peak torque. Accessed at https://www.youtube.com/watch?v=Rj__lexd_BI.