Sunday, July 6, 2014

Crank Length? Whatever...(within reason)

 Discussions about bicycle crank length have been a fairly hot subject recently. There's a trend in TT/Tri circles to use shorter than "normal" crank lengths. This is done to allow folks to either open up the distance between their torso and leg at the top of the pedal stroke (to be better able to produce power) and/or allow them to lower the angle of their torso relative to the ground in an attempt to get more aerodynamic. Sometimes it's done for both. One thing that is commonly brought up is questions about how using shorter (or longer) than normal cranks affects sustainable power output. Does it...or not?

There's a lot of data out there showing that wide variations in crank length do not have significant effects on  power output, such as the work done by Jim Martin at the University of Utah, summarized here:

From that presentation, "170mm cranks would compromise the power output of the shortest and tallest riders by AT MOST 0.5%. For example 6 watts out of 1200"

Another takeaway: "Crank length and pedaling rate influence metabolic cost and efficiency only by influencing pedal speed."

How can that be? Doesn't it logically make sense that for an optimum setup, a person needs a crank length proportional to their leg length?  For example, there are websites out there that claim to identify the "ideal" crank length for a given person ( However, one thing that most folks miss on that site is that the lengths of cranks recommended with that method uses an underlying assumption with no apparent basis, i.e. "The standard crank length of 170mm is optimum for a cyclist with a 31-inch inseam." The entire method is anchored in that assumption...and yet, the Martin study summary I linked to above shows that power production is relatively unaffected over a wide range of crank lengths for a wide range of leg lengths. Is there an optimum?  Perhaps...but, even if there is, the Martin data shows that varying from that optimum (even by relatively large amounts) doesn't appreciably affect power output.

But, doesn't a shorter or longer crank affect your "leverage"? Sure...but that difference in leverage can be compensated for elsewhere in the drivetrain. The crank isn't the ONLY lever between your foot and the ground. The gears themselves, along with the rim and tire diameter are also "levers", which can be varied. One thing to keep in mind is that physiologically, our muscles have a preferred shortening speed, which results in a preferred foot speed at the pedal. I'm not talking about cadence...I'm talking about the tangential velocity that the foot travels at during the power production (i.e. downstroke) phase of the pedal cycle.

The Challenge:

A little over a month ago, in a discussion on the Wattage Google Group about the effects of differing crank lengths, the following was posted in response to the idea that changing crank lengths could be "compensated" by varying gearing selections:

"The gearing argument has been raised many times. It's usually raised and argued a lot by people not on the big side of the spectrum,  ie people that are already on somewhat ideal cranks.

My counter argument is this:

If you are riding say 170mm cranks and you firmly believe that gearing can wholly make up for cranks of the wrong size. I encourage you to take the "Pepsi Challenge" that challenge is this:

1: Go TT up your favorite 15 minute or longer hill at LT or greater power (by HILL I mean something with average grade 6% or greater) with your current setup and time it a few times.
2: Then put on crank arms that are only 87.5% in length of your current setup.  That would be 148.75mm if you are currently riding 170s. You now have my proportional experience setup of me riding 175mm cranks.
   Take your own medicine and re-gear appropriately. Now ride that setup for a few weeks and go TT up your favorite hill a few more times and time it.

If you can show me you can do the same time with your medicinal gearing going full tilt, then I'll happily eat some humble pie.

I've offered that challenge for a few years now, and suprisingly enough NO ONE HAS TRIED IT. Go figure.  News flash, I and many other long legged people that have actually taken that "pepsi challenge", have come to the very real conclusion the longer arms are faster. Now we can quibble over why they are, but that's kinda irrelevant, fact is they are.

So I really challenge you or anyone else who is of normal size to take that challenge and see what it's REALLY like to be riding something so far out of what for what your body needs. Been over 5 years since I first offered that challenge, and I'm still waiting for someone of "normal" size to try it out...patiently.


Not being one to shy away from a challenge like that...especially since the one doing the challenging is someone I know...I decided to take it up. Luckily, I have a friend (Greg Steele, of Beehive Bicycles in Salt Lake City who let me borrow a set SRM cranks with the adjustable length option.

These cranks allow for an adjustment range of 150-190mm. To do the testing, I decided to use the first 2 miles of a local ~8% average grade hill. From the Kirby Palm "method", I used as the first test a setting of 175mm since that's what his formula claims is the "optimum" for my 32" inseam (as measured per his instructions). In order to ensure that my lowest gearing was as equivalent as possible, I chose a 12-25 cassette for use the 175mm cranks.  Since the crank length was pretty close to my normally used 170mm, all I did for adjustments was lower the saddle slightly. After setting everything up, off I went to the hill and give it "full stick" for my baseline. Here are the results:

Time = 12:50, Power = 289W, Cadence = 73 rpm, HR = 170bpm

For the short crank length, I wanted to use something that was equivalent, or below, the "87.5% below optimum" in the challenge above. For that, I chose 150mm, which is actually only 85.7% of the "optimum" 175mm length. In order to assure nearly equivalent low gearing, I chose a 12-30 cassette to pair with my 39T small ring.

Now then, when making such a large change in crank length, it's important to make sure to maintain the relationships between the saddle, the bars, and the foot during the "power", or downstroke part of the pedal cycle. As such, when I adjusted the crank length from 175mm down to 150mm, I did 2 things with the saddle.  First, I raised the seatpost such that the distance between the portion of the saddle I rest my sitbones on and the pedal at max extension was the same distance, but I ALSO moved the saddle rearward on the rails so that in the end it was a full 25mm further rearward relative to the center of the BB than when the cranks were set at 175mm.  This was done to keep the relationship between my lower leg and the pedal equivalent at the "3 O'clock" portion of the pedal stroke.  Because of the movement in the saddle (both up and rearward), that also meant I need to move the bars the same directions to keep the bar to saddle relationship as close as possible. Luckily, I had a stem handy that was short enough, and with a high enough rise so that I could do this with a fairly simple stem swap.  It wasn't the exact dimensions required, but it was close enough. Shown below is an overlay of the 2 setups.

So...after making the setup change a few days after the baseline run, I then ventured out to the test hill and with the only "accomodation" being the ~15 minute ride over there, I did a run with the 150s.  Here are the results for that:

Time = 12:50, Power = 286W, Cadence = 84 rpm, HR = 171bpm

Or...basically the SAME time, power, and HR as the 175mm crank length run. As expected, the cadence increased so that my tangential pedal speed and average pedal force were the same.  This can be easily seen with a Quadrant Analysis plot of each run.

175mm Crank QA

150mm Crank QA

Pretty much identical, no? what does this all mean?  It means that determining an "optimal" crank length for bicycling isn't very important. Instead of worrying about it from that standpoint, just understand that a wide range of lengths are acceptable, and use it as a "lever" (pun intended) for other things, such as fitting issues. One thing you might want to be wary of is going too large, in that that actually can start causing problems at the top of the pedal stroke and/or prevent the most aero position for a given event. It's somewhat hard to go "too small" (within reason) on crank length...just make sure you're geared adequately for the course profiles.

Oh...and it also means Kieran needs to eat some "humble pie"....nom, nom, nom ;-)


  1. Did you see average power and time while doing the tests, or was the bike computer hidden?

    1. It was displayed. But, I've done that climb enough to know "alls I can do is alls I can do". I gave it full stick for both.

  2. Thanks for a great article Tom.
    I spent a lot of time during winter 2012-2013 trying crank lenghts from 150 to 170mm using an adjustable Powercranks with Dual Mode (to be able to lock it up) and a Powertap wheel. Same thing no difference for me so I ended up using 155mm (that I've been using ever since) instead of the 165mm I was using in 2012 (170mm in 2011) because 155mm was the best for my pedaling comfort and my position.
    I'm still using 170mm on my MTBs (very complicated to find anything shorter for MTB + the higher saddle can cause other problems) and it's no trouble going from 170mm to 155mm or the other way around.

  3. Intriguing... when I first started on a TT bike, clearly the bike shop didn't mind putting on 175mm cranks for a S frame for a girl 5'2"... and I didn't know any better. Several years later, once I started reading about crank length and swapped to 165mm, the difference was surreal. I wasn't developing extreme fatigue in my hamstrings trying to pull the pedal back underneath me and up, and yes I could get that wee bit more aero.

  4. I would even suggest that at steeper gradients that the higher cadence with the short cranks would allow for the saving of precious W'/AWC given you wont be recruiting larger motor units.

    1. I'm pretty sure the whole point here is that cadence doesn't matter pedal velocity does and pedal velocity doesn't change. Are you suggesting that there is a point where it is cadence that matters?

  5. The picture up top is *perfect*. My favorite silly cranks.

  6. this means nothing since you knew at what length the cranks were for each trial. it would have been better if someone else would have set the crank length for you and you didn't look at them. this would be called a "blinded" study. no scientific data based on how you did your "experiment" would ever be accepted.

    1. I'm thinking you may want to get better information on the purposes of "blinding" in scientific studies ;-) What sort of bias, intentional or unintentional, would you be trying to prevent? As shown by the power, HR, and Quadrant Analysis results, the efforts and outputs were basically identical. Any "preference", or bias, would have shown up in at least one of those measures.

    2. it doesn't matter that the efforts and outputs were basically identical. when you knew what lengths the cranks were, you may or may not have been unintentionally or intentionally biasing yourself.

    3. As I asked above, in what manner would that bias manifest itself? How could my knowledge of the crank lengths (which would be obvious within just one pedal stroke even if it was completely blinded due to the different pedal circle diameters) influence the results? Besides, I never said I did this to submit to a scientific journal, I just did it to show someone they were demonstrably wrong ;-)

    4. bias is bias, it doesn't matter how it is manifested. i'm not agreeing or disagreeing with your hypothesis or results, just making a comment on your methods. i understand this is not going to a journal, but these type of "tests" happen all too often on the interwebs (most often in video/audio compression tests) and are passed off as "data," but if the tester knows what they are testing before the test even begins, they are introducing bias into the test/experiment. in scientific research, when blinding can be introduced safely and successfully into an experiment, it becomes necessary, so that the "placebo" effect can be negated. in addition if you really want to be sure of your data, the experiment should be completed at least three times (or as many as you need to get your standard deviations to a reasonable level) and run some statistics on it. these type of basic statistics can even be done quite easily on uncomplicated software like excel.

      if the crank length would be known after one pedal stroke even without previously knowing the length, another type of test would have to be designed. since i am far from a bicycle expert, i have no idea how it could be tested.

    5. I don't think you're following. Sure, if bias or placebo is possible, it should be controlled for in the test setup. However, in this case, what could be affected by bias or placebo without it showing up in the measures of effort and output? I've asked you twice now with no answer as to what type of bias the blinding would be intended to guard against.

      If you can't identify how it could be affected, then saying that not being blinded "means nothing"...well...means nothing.

      Also...don't worry...I know all about calculating and determining statistical significance. For the purposes of the intent of this blog post, it's not required ;-)

    6. As Gregory states, bias is bias, and is best removed from the testing. You can argue this point all you want, but it's best to just accept it.
      The argument "what could be affected by bias or placebo without it showing up in the measures of effort and output?" doesn't hold. For example: take a fairly standard double-blind medication trial. If the testers were to inform the subjects whether they were taking the "real" med or the "sugar" pill, then the (reported) results would vary. To ask why this is so leads to a variety of explanations, some valid, some maybe not, but no-one could ever definitively prove how a "real" med might suddenly be reportedly less effective were it to be announced as a "sugar pill".

    7. Sigh...again, how would placebo effect have changed the outcome of any of the measures in this test? There were no perceptual observations. I think you might want to look up the utility of blinding to control for bias and placebo.

    8. Blinding doesn't necessarily apply here. Unlike medication studies, where the effects are subjective, here the results are completely beyond the control of any bias. His heart rate remained the same. Power remained the same. Therefore, same efficiency.

  7. Sigh...@Lightweight Buffmeister, good comments, but I gave up on trying to explain good scientific method here, realizing my efforts are fruitless. As Dr. Dean Edell used to say, you're welcome to your own opinions, but you're not welcome to your own facts.

    1. And the facts are the power, time, HR, and QA plots shown above. It's your opinion that it all "means nothing."

      It DOES mean something in the context of the "challenge" it was meant to address. If you're looking for scientifically publishable results, take a look at the information in that Jim Martin .pdf I link to at the beginning of the blog post. There should be plenty in there to meet your satisfaction.

  8. Ahhh hard scientists once again trolling good 'ol practical engineers.

    While I still personally believe there are advantages to longer cranks for my 6'4 frame - I agree the test is valid. There is always visibility in engineering benchmarking and testing in general because we have to understand the kind of results we're designing the tests for for them to be meaningful (in the engineering sense).

    I'd still argue I could generate more power longer with a larger crank in the same regards that I swear I can grind faster (and easier on my chain) using the same (nearly) gearing on my 52 big ring than I can on my 47. The ultimate gear reduction might drive the same power to the ground, but the stresses on the intermediate components (the ball of my foot) are lessend with the longer crank.

    1. Just poking in here to say I have an undergraduate degree in biomedical engineering, a graduate degree in bioengineering, and published articles in peer-reviewed science and engineering research journals. I guess that makes me a "troll" that is a "hard scientist" and a "good 'ol [sic] practical engineer."

    2. and yet with all that studying you have failed to understand the difference between blind study and a test done with measuring instruments

      you have been told several times that power, time, HR, and QA was taken and compared and that those instruments are not effected by emotions, how can it be so hard to understand

      rider follows power meter to give same effort on same distance than you compare if the time changes

      or rider rider same distance in same time and than power meter and other results are compared

      point out the part where bias had influence

      please explain how rider bias effects a measuring tool, you have been asked several times and completely avoid it

      completely avoiding the key points and measuring tools makes you troll or maybe your mindset is not very scientific after all'

      this was not drug test so there was no need to do the test with same procedures... he used instruments to do it

  9. I also turn 105-110rpm on my 175 crank though (flatland and short climbs) and don't think i could reasonably turn faster which is required with the shorter crank. Again another variable in cadence style to consider - slow grinders versus fast turners.

  10. Hello Tom and All,

    Regarding your neat experiment about crank length:

    #1 Time = 12:50, Power = 289W, Cadence = 73 rpm, HR = 170bpm

    # 2 Time = 12:50, Power = 286W, Cadence = 84 rpm, HR = 171bpm

    For me I am not convinced that my Heart Rate would be the same at the same Power with an 11 pedal stroke difference in cadence, even though the pedal force was less ..... as I see a heart rate change (for me) at same power and different cadences ..... higher cadence giving higher heart rate.

    It would be interesting to see what the experiment values would be with the same cadence for #1 and #2 ..... perhaps test cadence at about 70, 80, 90, 100 and then compare the other values.


    “Key Points When competitive cyclists perform prolonged exercise that simulates racing conditions (i.e., variable, low-moderate submaximal cycling), a higher cadence results in excess energy expenditure and lower gross efficiency compared to a lower cadence at the same power output. Consequently, maximal power output is reduced during a subsequent exercise bout to exhaustion after using a higher cadence. Selection of a lower, more energetically optimal cadence during prolonged cycling exercise may allow competitive cyclists to enhance maximal performance later in a race.



    +1 mph Faster

  11. Great stuff. There remains however, the issue of comfort: which crank length does the best job of distributing the load over a range of muscles and through a range of position which feels best. This wouldn't translate necessarily into a time trial result. For example, if you wore uncomfortable shorts during a ride you might still be as fast (or faster if you're motivated to get it done with). Did you find a preference for one versus the other?

  12. Any word from Kieran on how that humble pie tasted?