Home General Training Discussions

Avg power higher than normalized power?

I did a new bike test this morning and something strange happened. On both my 5' and 20' tests, my NP was slightly lower than my avg power (2 watts for each test). The other intervals this morning, which were basically wu's and a cd, np was higher than avg power. I'm not sure why this happened and the only major difference is that I starting using a quarq this week as opposed to a powertap. Quarq seems to be working fine, calibrated, etc. Anyone have an idea as to why results would come out this way?

Comments

  • Common explanation for this has to do with the way the #s are calculated.

    AP is calculated on a second-by-second basis (possibly with a 3 second lag if you have "3 sec power" option). NP is calculated with a 30 second lag. Meaning it uses the previous 30 seconds measurements on a rolling basis to give the current/total NP. Some algorithms deal with this by not showing the NP for the first 30 seconds, others simply include the 30 seconds prior to the interval start as part of that intervals NP. So if you were going easier just before the 5 and 20 minute tests (duh!), your' NP will be lower than the AP for that interval.

    So for shorter intervals, especially on a trainer where you are presumably riding very steady with no curves, slowdowns for traffic, hills, etc, it is possible to have an AP higher than NP.

  • X2 on what Al said....

    AP is the true average power over the interval time
    NP is how it felt to you if there was a lot of coasting or spiking, or the true cost to your body... they should be the same if steady but this is less likely the longer you go

    AP is what moves you down the road...

    In this case use the AP the higher number! :-)
  • Thanks Al and Tim - I thought I understood the concept of NP, but your explanation makes sense. But, I'd figure over a 20 min time period that the normalized number would creep higher than the avg. I can see how shorter intervals, how the numbers would flip. Also, testing (or any ride) on the trainer vs outside should yield closer numbers between the np and avg, correct? In this case, they were very close - within 2 watts. Thanks!

  • Posted By Brad Marcus on 08 Dec 2013 10:30 AM


    But, I'd figure over a 20 min time period that the normalized number would creep higher than the avg.

    Having NP be lower than AP is pretty uncommon, and Al and Tim described it's really a technicality of the way the NP algorithm works rather than an indication of any physiological phenomenon except for one, it requires that your NP and AP be extremely close to begin with.

    So really this is just a sign that you are riding extremely steady to begin with, a feat that is certainly more difficult for longer intervals and quite impressive if you are spot on your target power for a full 20' set.

    Seeing NP < AP on the road is a different story due to road undulation and other environmental factors, it should be more difficult to replicate on the road. If you have a long variable ride that results in a NP being lower than AP, the cause is almost always that your computer is set to drop 0's from the AP calculation.

  • @Trevor - thanks for the comments. I just double checked my Garmin and it is indeed set to include 0's for power. I found this phenomenon odd, since I cannot remember it happening prior for an interval as long as 20', but maybe it was just a case of me sticking to riding steady. As I mentioned it was on the trainer. What I did not mention was that I have a trainer which allows me to dial in the resistance for power ahead of time, electronically, by programming it in. In this case, since it was a test day, I gave my 5' and my 20' a 30 watt range when I dialed it in. Of course this only adjusts the resistance and doesn't generate the power, but it may help to keep me in the zone of where I planned to be. When I take it outside in the Spring, all bets are off, and my results will not be as steady. Thanks again.
  • Oddly enough, I believe this is a solvable problem if we were to change the way AP is calculated...i.e., calculate it as the average of a rolling average (but without the raising to powers that's done for NP). In principle NP must always be greater than (or equal to) AP, but because AP is just plain average instead of pre-averaged before averaging....it's not quite an apples to apples comparison.
  • @Brad, The minutiae of the mathematics is fun for us nerds. But... Since this was your test, simply use the higher number (in this case your AP). The reality is that your NP for your 5' or 20' sets would be at least that or maybe slightly higher if not for adding in the much lower numbers in the 30s before your real test periods.
  • Ok - in case anyone is still tracking this thread, I want to make sure I got this right:

    When doing my 5 X 1'(1') intervals at Z4, I should record my AP (which is higher) and not my NP (which is lower). This is because as Al Truscott stated: "AP is calculated on a second-by-second basis (possibly with a 3 second lag if you have "3 sec power" option). NP is calculated with a 30 second lag". Therefore, my AP is the true measure of my interval.

    Since we didn't do any of these in the OS, I forgot and just want to make sure I'm consistent with the way we in EN do this. It's been so long since I've done these!

    Thanks,
    Joe

  • Posted By tim cronk on 08 Dec 2013 10:18 AM


    X2 on what Al said....



    AP is the true average power over the interval time

    NP is how it felt to you if there was a lot of coasting or spiking, or the true cost to your body... they should be the same if steady but this is less likely the longer you go



    AP is what moves you down the road...



    In this case use the AP the higher number! :-)

    Because of these artifacts of the calculation that Al is talking about, you sometimes see other strangeness, such as a VI that is less than 1 for intervals while riding on a trainer or on a long flat section of road.

    A lot of times, leading into an outside test, I will be coasting (zero power) leading up to the test, and only hit the lap button to start the test after pedaling for a couple seconds. this has the potential of adding nearly 30 seconds of 0w into the NP calculation for the interval.

Sign In or Register to comment.