Quantification of a training session using power

Discussion in 'Power Training' started by Oly87, Nov 9, 2009.

  1. Oly87

    Oly87 New Member

    Joined:
    Jun 29, 2006
    Messages:
    4
    Likes Received:
    0
    Hi,

    I am starting a PhD in relating the effect of training on performance, looking at it from a Statistics and Operational Research point of view.


    I was wondering if any of you had a method, or knew of methods, to quantify a single session in terms of power.

    As you can appreciate, power outputs in a training session fluctuate, so I am interested in knowing how people define the work done in a training session when viewing Power outputs, or Heart Rate for that matter.

    Any help would be gratefully received.

    Regards,



    Oli[FONT=&quot][/FONT]
     
    Tags:


  2. Alex Simmons

    Alex Simmons Member

    Joined:
    Mar 12, 2006
    Messages:
    2,471
    Likes Received:
    20
    Average Power is pretty clear and tells us how much mechanical work was done but it masks many things.

    Normalised Power and Training Stress Score tell us more and account for both the highly variable nature of power, the time course for physiological responses, as well as the non-linear relationship between physiological responses and intensity (power):
    Normalized Power (NP), Intensity Factor (IF), and Training Stress Score (TSS)

    along with how that feeds into an Impulse-Response model of training, a la the Performance Manager:
    The scientific inspiration for the Performance Manager
     
  3. daveryanwyoming

    daveryanwyoming Well-Known Member

    Joined:
    Oct 3, 2006
    Messages:
    3,857
    Likes Received:
    97
    Follow Alex's links for more information, but in terms of quantifying a training session folk's look at:

    - Sustained Average Power for various durations of interest where the durations relate to specific metabolic systems/processes targeted during that training session. So it might be looking at AP for 5 minute intervals if the workout was targeting VO2 Max work or AP for 15 - 30 minute sustained efforts if the workout was targeting Threshold work. Approaches like that are generally more useful and informative than looking at overall AP for the workout.

    - Total energy in kilojoules for entire workout is a literal measure of 'work' in the physics sense and useful in terms of planning refueling or guiding weight loss programs. It's the simple integral of AP over time and doesn't account for varying stresses very well but it maps pretty closely to Calories burned which is a useful metric.

    - Normalized Power (NP), described in detail in Alex's link but it estimates effective steady state metabolic load from a variable workout on the premise that short duration hard efforts take more out of us than lower intensity efforts. So a workout with a bunch of short sprints followed by rests is tougher than a steady ride with similar AP. NP only starts to make sense for rides/efforts longer than 20 minutes or so as it attempts to estimate effective 'metabolic' stress so it's most applicable for durations that are long enough to be primarily metabolic. Not a lot of value in calculating NP for 3 to 5 minute efforts as a training metric but NP is pretty useful in terms of understanding why a one hour race with many speed changes (like a criterium) is much harder than riding an hour steady to yield the same AP.

    - Intensity Factor or IF derived from NP and the rider's sustainable one hour power (FTP) is a good indication of how hard the workout was relative to the rider's individual and current fitness level. So it helps to describe why sustaining 250 watts for an hour or more is tough or impossible for many newer riders or amateur racers but hardly a workout for top professionals relative to their capabilities.

    - Training Stress Score (TSS) is a good single ride metric for overall training stress taking into account both the relative rider specific IF described above and the way that sustained relatively hard workouts are more stressful than sustained relatively easy workouts. So the score is weighted both to the individual rider's current fitness level and to show that intense workouts are more stressful than similar workouts even if the total kj or Calories burned are similar. TSS is an aggregate metric similar to total work in kj so it doesn't in and of itself give insight into the composition of the training that led to that score. It's still a very useful metric but has to be handled with caution when viewed in isolation or it can lead us down the high mileage/low intensity path as that's an easy way to rack up a lot of TSS but not necessarily the best way to develop bike racing fitness.

    As Alex points out there are additional metrics that accumulate over time and are very useful for tracking training over days, weeks, months and years(CTL, ATL, TSB) and there are other ways to process power data to gain insight into things like torque/leg speed requirements for different activities(Quadrant Analysis),. individual rider physiological characteristics (Monod-Scherrer CP Analysis), rider/bike aerodynamics (CdA regression or Chung method testing) and so on but in terms of single ride metrics it usually comes down the bullets above in addition to the traditional metrics of ride duration, mileage, possibly elevation gained, etc.

    Good luck,
    -Dave
     
  4. frenchyge

    frenchyge New Member

    Joined:
    Apr 3, 2005
    Messages:
    4,687
    Likes Received:
    4
    It's typically viewed post-ride. The software suites currently packaged with all cycling power meters will show work done, along with a host of other metrics related to the training session or selected portion thereof. The better ones will also put the workout into a training calendar/journal where mileage, kJ, hours, etc. can be viewed over a period of time (week, month, season, etc.).

    The links provided above go a long way in relating training stimulus to physiological response, and that software is an add-on that many of us here use and recommend for those who want to get "serious."
     
  5. RChung

    RChung New Member

    Joined:
    Sep 12, 2006
    Messages:
    306
    Likes Received:
    1
    Hmmm. As others have mentioned, there's already some stuff on training and performance. On the other hand, I think it could be interesting to look at power production on a single ride by modeling it as something like a renewal process with stochastic decrements and nonlinear renewal. I don't think that's been done yet. I'll bet there's some interesting extension of Blackwell's ergodic renewal theorem that could be cute.
     
  6. Oly87

    Oly87 New Member

    Joined:
    Jun 29, 2006
    Messages:
    4
    Likes Received:
    0
    Thank you for all your responses, very useful and helpful in giving me some idea on what to look at when quantifying a training session.

    I feel with the normalized power method it doesn’t take into account the duration of the session and is somehow not correct. However, that is for me to look into.

    Just one other question, do you guys have a way summing up a session in terms of Heart Rate? This is a good indicator as to what the effort levels were during the session and relating it to power may help.

    Thanks again

    Oli
     
  7. Alex Simmons

    Alex Simmons Member

    Joined:
    Mar 12, 2006
    Messages:
    2,471
    Likes Received:
    20
    Did you read the links provided?
    Not really, because it's not.

    Of course you can look into TRIMPS and EPOC. But these are response based methods and of course HR is subject to variation due to factors unrealted to training.
     
  8. frenchyge

    frenchyge New Member

    Joined:
    Apr 3, 2005
    Messages:
    4,687
    Likes Received:
    4
    Well, you could count the beats.... and then do all that cute stuff that Dr. Chung suggested. ;)
     
  9. RChung

    RChung New Member

    Joined:
    Sep 12, 2006
    Messages:
    306
    Likes Received:
    1

    Hmmm. Do you also feel that the mean power method doesn't take into account duration and is somehow not correct?
    Well, good luck with that.
     
  10. daveryanwyoming

    daveryanwyoming Well-Known Member

    Joined:
    Oct 3, 2006
    Messages:
    3,857
    Likes Received:
    97
    As rchung suggests NP is an alternative form of averaging so it takes duration into account just as much or just as little as a direct arithmetic mean. Yes, it yields a single number just as an average would so by itself it is more closely related to intensity than it is to overall load which is where TSS comes in.
    Take a look at this document, especially Figure 1. in terms of your assertion that HR is 'a good indicator as to what effort levels...': http://www.midweekclub.ca/articles/coggan.pdf

    But realistically you should be surfing Pubmed, not the cycling forums if you're actually gathering information or performing a preliminary literature review towards an advanced degree. Has academic research really sunk to posting questions on internet forums?

    If you want to see what some of unpublished thoughts from folks (not necessarily academics and typically not structured research) committed to power based training you should join the Google wattage list: http://groups.google.com/group/wattage

    But there are hundreds of published peer reviewed papers related to sports performance metrics, but you won't in general find them here or on other enthusiasts forums.

    Good luck,
    -Dave
     
  11. dkrenik

    dkrenik Member

    Joined:
    Dec 5, 2003
    Messages:
    440
    Likes Received:
    8
    Not just academic research. It's amazing the type of advice folks are seeking in these forums (medical? C'mon...). There's even a thread about "which saddle?". I'm notsure how many subjects could be much more individual than that one.

    Finished with rant,
    Dave
     
  12. Oly87

    Oly87 New Member

    Joined:
    Jun 29, 2006
    Messages:
    4
    Likes Received:
    0
    Within the NP method, the moving average is used to smooth the data and remove the noise from it. So duration of the session isn’t accounted for, as the transformed data is still taken at 5secs intervals.

    For example, a session lasting a 30mins may NP 280, a session lasting an hour may have the same NP of 280. I feel these sessions cannot be equal in terms of what work was done.

    I would agree the TSS takes into account duration.

    Hope that explains my point of view.

    Thanks for your responses.
     
  13. daveryanwyoming

    daveryanwyoming Well-Known Member

    Joined:
    Oct 3, 2006
    Messages:
    3,857
    Likes Received:
    97
    You seem to be missing the point. NP is a form of averaging, it is not intended to represent overall workload but rather it is intended as a measure of effective metabolic intensity. Your point about a half hour versus hour long ride applies equally well to average power or any other weighted or unweighted averaging approach.

    So yes, NP does not represent session workload, it was never intended for that, nor does AP nor other averaging approaches. That doesn't devalue NP as a metric, just that it doesn't provide a proxy for total session work. But then again the aggregate metrics including total energy burned in kj or TSS don't in and of themselves provide insight into intensity your same argument applies.

    IOW, 2 rides could both burn 1000 kj or rack up 100 TSS but in isolation that tells you nothing about the intensity or energy delivery systems targeted with those workouts. A rider could burn 1000 kj in a 5 hour ride at ~56 watts with no appreciable training benefit or in a one hour ride at ~ 278 watts.

    My point is that you won't find a single metric that fully describes training any more than you can characterize the general performance of an automobile on a single metric (mpg, hp, peak speed, peak engine torque...).

    -Dave
    P.S. FWIW, the 30 second averaging isn't as much a noise reduction mechanism (I work in signal processing and can see how it's possible to view it that way) as a mechanism to sort out metabolic vs. neuromuscular and purely anaerobic efforts. The 30 second window was chosen on typical exponential half lives of adaptation for physiological processes, IOW high power events of much shorter duration do not really cause the body to respond the same way as longer sustained efforts.
     
  14. acoggan

    acoggan Member

    Joined:
    Jul 4, 2003
    Messages:
    3,047
    Likes Received:
    9
    You need to study the algorithm a bit more.
     
Loading...
Loading...