Monday, May 17, 2010

Fatigability and BMX performance at the Olympic level

by Hunter Allen and Andrew R. Coggan, Ph.D.

HA: I have been involved in BMX since I was 11 years old, when I competed in my first race, and BMX has been dear to my heart ever since. It gave me the skills to become an elite MTB racer and later a pro on the road. Luckily for me, though, I realized that I don’t have enough fast twitch muscles to really succeed in BMX, and endurance was more my forté. However, I have continued to follow BMX over the years and when BMX became an Olympic sport in 2006 (for the 2008 Games), I knew that I wanted to contribute to the cause (besides, I had no data on elite BMXers!). As Technical Coach to the 2008 BMX Olympic Team, it was my job to outfit the team with SRM power meters, conduct a series of on-track and off -track tests, and begin to define the demands of BMX racing along with the abilities of the best riders in the world. Some of these data are described in the section on BMX in the new 2nd edition of our book, but we thought that it might be interesting to go into a bit more detail here.

For starters, it is important to realize that the Olympic BMX Super Cross track is not your regular backyard local track. This thing is practically a motocross track! It has a 30 foot tall starting ramp, the jumps are over 40 feet apart, the first turn berm is 25 feet tall, and the race lasted about 36 seconds, which is quite long for a BMX race. It is EXTREME. The demands of the track are different than a regular track and therefore some of the best BMXers on the national BMX circuit did not excel on this track. Check out the picture below (taken at the replica track at the Olympic Training Center in Chula Vista, CA)or view this quick video (http://feelbmx.com/videos/olympic-bmx-video-men-final) of the men’s Olympic final to see just how big and gnarly this track truly is!


One of the first things we tested in Chula Vista was the effort of the riders down the first straight away. I wanted to see how much time was spent pedaling vs. not pedaling (in the air mainly) and also how many watts they were able to put out coming down the start ramp and then coming out of the first turn. These were critical areas in the track and probably held the keys to success in BMX. The next thing we tested were full race laps with each athlete by themselves. This way we could see their fatigue resistance throughout the entire course without interference from other riders. Lastly, we did a few mock races to compare an actual race to the previous tests.

When I observed the riders during the tests themselves, it was really clear to me that they have quite a few “micro-rest” periods in each straightaway. So much time was spent in the air between the jumps that I knew this could be a critical component to an athlete’s ability to prevent fatigue near the end of the race. “Floating” over the jumps and relaxing in the air is definitely not a skill that all the Olympic hopefuls had and the riders that made the Olympic team clearly had this wired. Each race in fact was a series of “micro-bursts” and micro-rests. Another thing that I noticed was that the riders that were able to “corner-start” (that is, had the ability to replicate their effort leaving the starting gate while exiting the first turn) really had an advantage down the second straightway. If the rider could produce nearly the same watts they did off the starting gate, but now do it while rolling and exiting the first berm, it made a big difference.

When I looked at the data, I was blown away by some of the wattages that these athletes were putting out at the start and then down the first straight and exiting each turn. Another thing that blew me away was that not only were the best athletes putting out over 1800 W exiting the first turn, but they were pedaling at over 160 rpm and the best were over 180 rpm! Right away, it was obvious from the technical demands of the track and their physical performance that these BMXers were highly skilled and elite athletes. There was not a slacker in the bunch, and this was clearly not a sport for sissies.

I asked Andy to take a look at this data to see if he could see the same things that I was observing in analyzing the BMX power data and my observations at the track. I knew that if he put some thought and math into the data, that we might even learn more…..

ARC: When Hunter asked me to see what I could make of the power meter data mentioned above, the first question that came to mind was just how much of a decrease in power actually occurred during the race. In particular, I was curious as to how the fatigability of these athletes compared to other data that are available, e.g., to published standards for the fatigue index measured as part of the original Wingate test, and/or to the large amount of data we have collected on road and track cyclists since the power profiling was developed in 2003. This question, however, could not completely answered by simply looking at the “raw” power data. This is because single-speed BMX bicycles are typically fitted with low gears, e.g., 50-55 gear-inches, to help the rider get the “hole shot”, i.e., to maximize the rider’s ability to rapidly accelerate away from the starting gate and thus beat their competitors to the first jump or turn. As a consequence, for most of the race a BMX cyclist’s cadence is much higher than is optimal in terms of power output – for the men at the training camp, for example, peak cadence during the time trials was typically over 170 rpm. When combined with the long crank arms such riders often use, this means that their circumferential pedal velocity, and hence muscle shortening velocity, was extremely high, which in and of itself would tend to limit their power production later in the race. In other words, power might be lower at the end than at the start of the race not due to muscle fatigue per se, but simply due to the difference in cadence.

Fortunately, the “gate start” and “first straight” efforts that Hunter had the riders perform provided me with a way of correcting the data for the effect described above. I did so by using these multiple, but very brief (i.e., 4-6 pedal stroke), efforts to reconstruct each rider’s average effective pedal force (AEPF) – circumferential pedal velocity (CPV) relationship, as shown in Figure 1 below:

Figure 1: Relationship between average effective pedal force and circumferential pedal velocity from a representative athlete.


Based on this relationship, it was possible to calculate each rider’s:

1) maximal AEPF (AEPFmax), which is the Y intercept of the fitted line shown above;

2) maximal CPV (CPVmax), which is the X intercept of the fitted line shown above;

3) maximal power (Pmax), which is equal to 0.25 x AEPFmax x CPVmax; and

4) the CPV at which Pmax is produced (CPVopt), which is equal to 0.5 x CPVmax.

More importantly, these data allowed me to express the AEPF at any time during a full-lap effort as a percentage of the maximal AEPF (and hence power) that the rider could produce at their cadence, and hence CPV, at that time. By doing so, it was possible to determine the rider’s fatigability independent of changes in their pedaling rate. (Note that although all of the SRM handlebar computers were set to record data at 0.5 s intervals, in practice this actually means that all data are based on individual pedal cycles. This is because the SRM averages data over a full pedal revolution before calculating power, cadence, etc., and none of the riders ever pedaled fast enough to complete two full pedal revolutions in 0.5 s, i.e., in no case did the measured cadence exceed 240 rpm.)

Having determined that the above approach was feasible, I chose to analyze the data from the one or two solo full-lap efforts that each rider performed, on the assumption that these data would be more reflective of their “pure” physical abilities (i.e., skeletal muscle characteristics, fitness, motor control) than the data from the mock races, where interactions with other riders might occur. An example of the results of these analyses is shown below:

Figure 2: Average effective pedal force (AEPF) expressed as a percentage of circumferential pedal velocity (CPV)-specific maxima as a function of time during two TTs by a representative athlete.


As can be seen in the figure, the rider in question spent much of their time not actually pedaling, and even when they did pedal they did not and/or could not always do so very forcefully/powerfully. However, their pattern of force (and hence power, since the data are expressed relative to CPV) application was very consistent during the two time trial efforts, something that was true of the other riders as well. This indicated to me that the consistent variation in force/power must have to do with the placement of turns, jumps, etc., on the course, but since I was not present during the data collection I had to ask Hunter to fill me in on such details.

HA: The Super Cross course starts with a mammoth, 30 foot tall starting ramp that had up to a 53 percent drop at the steepest section, and then into a short flat section leading to the first double jump which was 40 feet from peak to peak. The riders coming off the starting ramp could only pedal so much before reaching a critical velocity and having to prepare for this first double jump. This first double jump was so intimidating that even World Champions had to ride it quite a few times before getting their nerve up to jump it. Next came another double jump into a long “tabletop” jump in which the riders couldn’t pedal over, and then lead into the first turn. Coming out of the first turn at over 35 miles per hour, the riders had to “corner-start” and pedal hard for a few pedal strokes in order to hit the next “step-up” jump, which peaked at 25 feet tall. This was followed by a smaller double jump, after which they had to jump over the women’s course berm, leaping a massive chasm onto the men’s course and second berm. If the rider did not have a solid “corner-start”, then that could play out badly at the end of the straight when they had to leap the chasm. The third straight was characterized by almost continual jumps, so much so that it was only possible to get in one or two pedal strokes and the final straight had two more jumps in it with a flat sprint to the finish.

ARC: Once I understood why these cyclists were or were not pedaling at certain times, I decided to focus on their force/power during the last two or three pedal strokes, as an indicator of how much they fatigued during each TT. These data, along with data derived from the force-velocity relationship previously described, are shown for four athletes in the table below. These individuals were chosen for comparison because although they were all very similar in terms of performance in the unfatigued state (i.e., data shown in the first four columns), the first two failed to make the U.S. Olympic team, whereas the last two won the Silver and Bronze medals, respectively.

Table 1. Force-velocity relationship during cycling, maximal power, and power at end of TT for four athletes.

Upon examining these data, what became evident is that the more successful athletes (i.e., th last two) were able to maintain a higher relative (to CPV) power output during the latter portion of the race, i.e., they exhibited less fatigue.

The question then arises as to what might account for the greater fatigue resistance (lesser fatigability) of the cyclists C and D compared to riders A and B. One possibility, of course, is a difference in fitness/conditioning, and indeed other men at the camp seemed to be somewhat lacking in this respect. This did not, though, appear to be true for cyclists A and B. Another possibility is an inherent difference in muscle fiber type, i.e., it is possible that cyclists A and B fatigued more rapidly because they had a higher percentage of type II, or fast-twitch, muscle fibers. As shown in Table 1, however, the slope of the AEPF-CPV relationship was similar in all four men (and indeed across all of those tested, including a rider who recorded what to my knowledge is the highest-ever 5 s human power output of 25.2 W/kg), suggesting that they were all also similar with respect to muscle fiber type (a higher percentage of type II fibers would be associated with a shallower slope of the AEPF-CPV line, i.e., force and hence power would fall off less rapidly with increases in CPV and hence muscle shortening velocity).

If not fitness or fiber type, what does explain the difference in fatigability between these otherwise very closely-matched athletes? As it turns out, it appears that cyclists A and B fatigued more than cyclists C and D simply because they pedaled more. Specifically, based on the SRM data cyclists A and B completed 27 and 34 pedal revolutions during their TTs, whereas cyclists C and D pedaled only 20 and 18 times, respectively. This is not because cyclists A and B used markedly lower gearing, as they did not – rather, the higher number of pedal revolutions was apparently the result of their attempting to generate power at times when it was difficult, or even impossible (i.e., when air-borne), to do so. In contrast, cyclists C and D pedaled less often, but when they did they did so with maximum effectiveness, e.g., the three pedal revolutions at ~100% of velocity-specific force and hence power performed 7-8 s into the race shown in Figure 2. The remainder of the time, they apparently “rested” their legs as much as possible by “floating” over the jumps, etc.

HA: To summarize, then, here are some important lessons to be learned from these data:

  1. Pedal less, win more. While this is well known in road racing circles, it is not so well known in BMX. As described by Andy above, however, we found that the guys who made the Olympic team pedaled less than those that did not make the team.
  2. 100% "fast twitchers" may not make the best BMXers, at least for the SX. In particular, the guy that cracked out the biggest numbers in terms of Pmax (which is measured over a single revolution) and maximal 5 s power also did not make the team.

  3. Fatigue resistance matters. Again as described above, the Olympians fatigued the least over the duration of the race. They were able to make the most of their "micro-rest" periods, along with superior fatigue resistance gained via conditioning. On the hand, riders who did not make the team fatigued more, either simply as a result of pedaling too much (see point #1 above) or (in the case of other riders whose data are not shown) due to lack of fitness. 
We sincerely thank Dr. Steve Johnson at USA Cycling for permission to share these data and Mike Day (Silver Medalist, 2008 Olympic Games) and Donny Robinson (Bronze Medalist, 2008 Olympic Games) for allowing themselves to be identified in this article.

4 comments:

  1. The power meter is dead. Long live the power meter. :D

    Great post guys. While much info from Quadrant Analysis is pretty self evident for many, I find the max CPV-AEPF stuff extra cool. I like the thought of taking this to show fatigue resistance.

    Where figure 2 show a rider exceeding their max CPV-AEPF, is that because the max CPV-AEPF is a line of best fit through sample start/sprint data?

    ReplyDelete
  2. Make that:

    "The heart rate monitor is dead - long live the powermeter!" ;-)

    Re. your question: There is some variability in the 0.5 sec by 0.5 sec (really, revolution-by-revolution) SRM data, which is undoubtly of both technical and biological origin. Individual points can therefore exceed the line-of-best fit prediction just due to random chance, as you concluded.

    Other possible contributing factors could be 1) that the riders were more motivated when doing the full-lap TTs vs. just the "gate" and "first straight" efforts, and 2) twitch-potentiation (I'll let you look that up!)

    ReplyDelete
  3. OK, we have a new PPP :-)

    Twitch potentiation: repeated motor nerve stimulation enhances the evoked mechanical response of the corresponding muscle, resulting in an increased twitch response*.

    IOW muscles can perform more forceful contractions after some initial NM stimulation, or not lose the contraction force ability over repeated efforts.

    Time lag on that? So would that also be related to track sprints/starts which are often performed at a higher power after one or two efforts have been done and separated by quite long breaks (compared to the relatively frequent NM efforts seen in BMX)?

    * Or the more formal definition:
    touching the the TV remote control a couple of times results in the ability to make more forceful/rapid changes of channel as and when ad breaks drive you mad. :D

    ReplyDelete
  4. Thank you for a well researched and written article. I have been looking at all the facets of BMX training. Obviously there are differing requirements when coming off a flat (ie. non SX hill). But the finding of those who had the least fatigue at the end were those who pedaled less is really interesting. Something I had never considered.

    ReplyDelete