On my first test run, I carefully rode for 30 minutes on the standard chain and then 30 minutes with the optimized, following the same protocol for both runs, the optimized came out ahead, but just. The second run was even clearer, the optimized was faster.
Then a friend suggested using the optimized first. And then the run with the standard chain was the faster. Yikes.
I reached out to Kreitler to see if the bearings could get faster after a warm-up period. Never heard back. The temperature during these early runs, according to the Garmin we were using, registered at .8˚ Fahrenheit difference. I didn’t have the means to see if the drum was heating up. I resolved to find a temperature correction formula as well as only test one chain per day.
As for temperature, I asked around and was pointed to the book Bicycling Science. The book posits that rolling resistance (Crr) changes by about .6% per degree Fahrenheit. Some recently tried to see if they could find those numbers, but in their tests they got .8% per degree Fahrenheit difference, and are wondering which to trust. So looking at the temperature differentials on the Garmin, I could have been getting anywhere from a 1.7-2.3% difference in Crr on the first day of testing, which, assuming the power was spot-on in both runs, is hard to know if it would be enough to make a difference I could find, but a difference all the same. I base this on looking at Tom Anhalt’s Crr spreadsheet and the accompanying article. He tests an Armadillo’s Crr being .0077, and 2.3% of that is .0001771, so the new Crr would be .0078771. As I didn’t measure either the tires or the loads going in, and I know my weight fluctuates over the course of the day, I’m reluctant to calculate the Crr for the tires I was riding. And finally, it’s hard to know how consistent the Garmin’s temperature measurement is.
|chain||tires||powermeter||Kreitler||interval||power||distance||speed mph||speed m/s||cadence||Temp|
|standard||Rubino Pro||Cinquo||gate open||4:00||153||1.015||15.225||6.806184||102||69|
|standard||Rubino Pro||Cinquo||gate open||4:00||200||1.164||17.46||7.8053184||99||69.8|
|standard||Rubino Pro||Cinquo||gate open||15:00||250||4.872||19.488||8.71191552||96||69.8|
|optimized||Rubino Pro||Cinquo||gate open||4:00||151||1.026||15.39||6.8799456||102||71.6|
|optimized||Rubino Pro||Cinquo||gate open||4:00||200||1.175||17.625||7.87908||101||71.6|
|optimized||Rubino Pro||Cinquo||gate open||15:00||250||4.865||19.46||8.6993984||96||71.6|
|standard||Rubino Pro||Cinquo||no resistance||15:00||251||6.33||25.32||11.3190528||92||68|
|optimized||Rubino Pro||Cinquo||no resistance||15:00||249||6.5||26||11.62304||94||68|
|standard||Armadillo||Elsa Calibrated||no resistance||15:00||154||4.868||19.472||8.70476288||106||73.5|
|standard||Armadillo||Elsa Calibrated||gate closed||15:00||247||5.365||21.46||9.5934784||93||67.7|
|optimized||Armadillo||Elsa Calibrated||no resistance||15:00||154||4.863||19.452||8.69582208||99||71.6|
|optimized||Armadillo||Elsa Calibrated||gate closed||15:00||247||5.482||21.928||9.80269312||94||69.8|
|optimized||Armadillo||Elsa Calibrated||no resistance||15:00||154||6.059||24.236||10.83446144||98||69.8|
|optimized||Armadillo||Elsa Calibrated||no resistance||15:00||154||6.159||24.636||11.01327744||100||69.8|
|standard||Armadillo||Elsa Calibrated||no resistance||15:00||154||5.785||23.14||10.3445056||99||64.4|
|standard||Armadillo||Elsa Calibrated||no resistance||15:00||154||5.786||23.144||10.34629376||99||64.4|
|optimized||Armadillo||Elsa Calibrated||no resistance||15:00||154||5.877||23.508||10.50901632||100||66.2|
|optimized||Armadillo||Elsa Calibrated||gate closed||15:00||247||5.842||23.368||10.44643072||96||64.4|
|optim&lubed||Armadillo||Elsa Calibrated||no resistance||15:00||154||5.915||23.66||10.5769664||96||68|
|optim&lubed||Armadillo||Elsa Calibrated||gate closed||15:00||205||5.228||20.912||9.34850048||91||64.5|
|optim&lubed||Zipp 303||Elsa Calibrated||no resistance||15:00||154||8.883||35.532||15.88422528||106||68|
That said, the optimized chain did beat the standard chain at identical power in most runs. And given days with two identical runs with the same chain, the second run was typically faster.
After lots of runs, and lots of frustration realizing how carefully I had to control for variables and wishing I had been able to stay on the same components at the same calibration and same temperature for every run, it seems that I probably found small benefits. Interestingly, the results recorded show the difference between the optimized and standard chain to be greater than 6cm per second, not that the numbers I found are reliable due to the power meter variations and the questions as to how to correct for temperature.
The testing itself was eye-opening. The first thing is that taking the time and attention to make sure all the variables are properly controlled is hard. The second thing is that even with all that effort, I found myself questioning the results, worried that the unquantified made the measurable differences. To me, the experience showed if I’m going to sweat any detail, better sweat all the details because otherwise the effort could be for naught. It’s swell to have a fast chain, but if I ride the wrong tire, or the right tire at the wrong pressure, or too thick a grease on the bearings, the power saved by one component could be more than offset by insufficient attention paid to another. This is why the top racing teams often have people who figure this stuff out (“performance directors”) and then they nag the mechanics to get it all dialed in properly—in some cases it gets undone by a competitor making a boneheaded equipment choice. It seems that the smart strategy in the marginal gains game is pick off the easy gains first, and then, as focus and budget allows, go for the harder ones. Even non-racers can benefit from marginal gains so long as they choose wisely. On the flip side, it’s a waste of resources to spend on all one’s budget looking for marginal power savings, and then lack the funds to travel comfortably to your primary racing goal of the year.
I’ll trust Friction Facts tests because Smith has demonstrated he’s careful, transparent, thorough, and thoughtful–and Smith is also giving away the details of his wax mix on the Friction Facts website, so it’s hard to accuse him of skewing the results to favor a service he sells.. I should add “for now,” as new data questioning everything could come to light some day. With the chain, looking at the tests he’s produced and talking with him, doing the full-on optimization makes sense only when I’ve got everything in life, training, and my bike set and I’m heading to some kind of championship event. Melting plain paraffin and bathing a chain in it looks like it can get people most of the way to the special treatment, and if it’s easy to do, it could be a real help to do it right before a major event if you have the time. A good master link, like Wippermann’s, can make this operation pretty fast and simple. And short of that there are thin lubes FF tested that are good for everyday use.
One of the other questions I had with the optimized chain is how it would fare in terms of durability. Could it manage several time trials? A long road race? A stage race? What about riding in rain and grit? At the end of the test, after riding the optimized chain 111mi indoors, we took the chain outside into the final throes of NYC winter. The first outside ride was the day after a snowfall, and it was pretty messy. We did 50 miles that day, and then another 44 miles before squeaks started emanating from the chain. FF’s Smith states that the chain efficiency decreases before squeaks start happening, and our last run before applying Rock ‘n Roll Gold lube was slower than the first run after applying the lube. RnR is both what I had handy, and a lube that tested very well for FF.
The wax treatment probably isn’t effective for a stage race, but for long road races, time trials, and especially track racing, it could be a relatively easy performance boost.
I didn’t expect this test to focus on testing methodology and controlling variables, but that’s where the focus shifted as I started to analyze results. At the same time, it was a valuable lesson and good practice for the future, both in terms of testing and looking for performance gains. It is frustrating not to be able to re-create at home any assurance that the gains found in labs can be found in real-world testing, but that’s the nature of marginal gains, and why they’re hard to find.
I want to thank Velimpex for suggesting said test, providing the chains, and for Friction Facts for their work and time.