• Chris Andrews

    More than a minute per km on the return leg?! That must’ve been some headwind!
    You’ve gotta treat a headwind like an uphill, pretty standard pacing decision…

  • John Murphy

    Just want to say this was a great article, very interesting and well thought-out and written.

  • Pete

    So in the initial flat simulation, what was held constant? Total energy expenditure? If so, that’s a non-trivial assumption to make. If you start with a lower power, then increase it gradually, you can conceivably increase your total energy expended, compared to maintaining a constant power output. The missing factor is… that people get tired.

    • Nick Squillari

      There was no ‘negative split’ (read: increasing power to attain the same average power). It was ‘fast start’ for all simulation, with power then decreasing to the predetermined wattage being tested.

      People without doubt get tired – the studies (and the review) make mention to them (especially with respect to W’). So your point is totally valid and recognised. Only it was a more macro view of what strategy would win – one a consistent power output or one that was more varied. Which may sound like given but there are still many who race without this taken into account.

  • bacon

    I’m keen to hear everyone’s thoughts on negative splits for a climb. For example, say you aim to hold constant power through a TT and then approach a climb – surely it is better to hit this at your current power then aim to increase as the climb goes, resuming your base power once the climb has finished.
    This would avoid blowing up trying for a power you cannot maintain. Or would it be better to just know what power you need to raise to for the climb, and then aim to hold that constantly throughout?

  • David McAdam

    So Zwift results where in line with your prediction. Maybe Zwift is based on exactly the same studies you were reading.

    • Pete

      The Tour of Bright TT should be held on Zwift to save driving up Thursday or super early on Friday morning.

    • Hamish Moffatt

      Or Zwift and the studies are just physics..

  • Alex Simmons

    Those studies were, well, a bit lame. They only consider the (relatively easy to solve) physics and completely ignore the realities of physiology.

    They make no attempt to recognise that one cannot simply vary power in such a manner without some physiological cost. The more you vary power (for individual segments longer than about 30 seconds), the lower your average power will be if riding to minimise time for the entire distance/duration. Assuming total energy output is a constant (or average power is the same) is not a physiological plausible assumption when the power is delivered in such variable fashion.

    So while increasing power for ascents and lowering it for descents does make sense when riding a variable gradient course, there is also a significant physiological constraint in play that limits the level of variability in power that is optimal.

    Vary power too much in line with the gradient changes and you’ll end up going slower overall because you will be forced to ride with a significantly lower average power overall. It’s also more than simply to do with one’s anaerobic work capacity, since replenishing anaerobic capacity is a wholly aerobic process, IOW an energy demand required to replenish O2 deficit is added to that required to drive the bike forward. Very soon that results in one having to ride below their expected mean maximal power output for the duration.

    I (and others) have modelled TT pacing accounting for both the physics and physiology for many years and tested outcomes versus real TT performances. I’ve done it for world tour riders down to club amateurs. Yes, variability with gradient is optimal, but the level of variability is not always as big as many might imagine. If you are unsure, then your mean maximal isopower for the expected duration is actually still a very fast strategy (it’s actually pretty hard to do), and varying power a little in line with gradients will save a bit more time.

    It’s possible with some analytical techniques I’ve developed to do a post hoc assessment of a given ride that also accounts for variations in environmental and other factors on the day, and to give an indication of how far from optimal the ride was.

    As for head/tail winds, while it’s sort of true that you can think of them as a gradient, it’s not quite as simple as that. For a start the physics for power-speed relationship for overcoming air resistance is different to that for overcoming gravity (or getting gravity assist when descending). But the bigger factor is simply the duration of the head wind sector will be the majority of the ride and so big variations in power are not really feasible since you will bump up against your mean maximal power capability for that duration.

    In general, the optimal power variation for a head/tailwind scenario is not big. It depends a little on the strength of the wind but in reality since you will be spending more than half the race duration into the headwind you are not really going to be able to increase power on that leg all that much unless you significantly underpowered the tail wind section in which case you will still loose time overall.

    Most people actually go too hard on the ascents and don’t push hard enough on the descents. And they often forget about the opportunity when cresting a rise. The biggest and by far the most common pacing mistake riders make (assuming you don’t miss your start that is) is starting too hard for too long. It only takes ~15 seconds to get up to cruising speed and if you are overpowering (over threshold) beyond that then you are costing yourself time later on.

    As to comparing results from different rides, unless one has performed a very decent assessment of the differences in environmental conditions and other factors such as aerodynamics, I would be very careful in assigning performance improvements solely to a different pacing strategy. It only takes a reasonable difference in air pressure to see a minute difference in TT times over 40km for same power/aero. The Aussie national TT titles were held on same course last 2 years but the conditions were quite different.

    As for the rider that lost all that time on the headwind return leg, I put that down to attempting to ride far harder than they ever were going to be capable of sustaining for the duration, even on a windless day.

    • Jesse Graham

      Article is equivalent to the “Beginners guide to simple mathematics”. Alex has just dropped the “Advanced Abstract Concepts in Algebra”.

      I think “pacing” would be new concept for many ?

      • Alex Simmons

        Well the nice thing is that the principles for good time trial pacing that take into account all relevant factors are pretty simple, just not as wildly optimistic as those papers suggest.
        – start on time
        – don’t start too hard
        – a little higher power on inclines than declines (up to say +/-10% in power terms is plenty*) but expressed in terms of perceived exertion it actually feels like you hold back a bit on inclines and push hard on declines.
        – if descents are steep then tuck in if you run out of gears
        – have fun

        * there are a few variations on what’s optimal depending on your ratio of W/kg vs W/m^2

        • Nick Squillari

          Hi Alex. Thanks for your input. Always excellent to hear from an expert whose advise I’ve used myself many times.

          Interested though to know more of your thoughts on what parts of the two studies in question are “wildly optimistic”?

          • Alex Simmons

            Hi Nick

            My comments are especially wrt the variable gradient scenarios which model physiological implausible scenarios.

            We only have to read what they modelled:

            “A hypothetical rider (body mass 70 kg, bicycle mass 10 kg) was studied over three separate undulating 40-km courses comprising eight hills starting with 2.5 km of uphill and followed by 2.5 km downhill, with course gradients of 1%, 3%, and 6%, resulting in no net elevation change. A range of mean power outputs 200–600 W (in 100 W increments) were considered and power output was systematically varied by ±5%, ±10%, ±15% from this mean baseline;

            For each combination of power output and gradient, there was a frequency of variation in power output of 2, 4, 8, 16, 32, 64, and 128 per time-trial; distance of variation (in km) was thus calculated as event distance divided by the frequency of variation (20, 10, 5, 2.5, 1.25, 0.625, and 0.3125 km, respectively).”
            /end quote

            e.g. in one modelled combination the hypothetical rider does 20km at 9.9W/kg followed by 20km at 7.3W/kg! That’s just silly. Most of the scenarios are unrealistic. OK silly high power we can ignore and just consider realistic power outputs modelled…

            The big problem is all they do is model the physics of applying such variable power and completely ignore the physiological reality involved.

            One cannot vary power and expect to maintain the same average power that you could when power is quasi steady state.

            Varying power +/-15% from threshold power over significant interval durations for 40km and ending up with an average power = threshold power is physiologically implausible and creates wildly optimistic outcomes wrt what times gains are possible.

            Put it this way, if a 70kg rider with average TT CdA has a threshold power of, say 300W, then attempts to ride 2.5km 6% ascents at 345W (which for a 70kg rider will take nearly 7 minutes) then 2.5km 6% descents at 255W (which will take ~2 and a 1/4 minutes) and repeat for 40km, well they will crack before they reach the half way mark. Normalized Power in such a scenario is approaching 330W.

            In other words, the pacing strategy is not physiologically plausible and hence the modelled time gains are wildly optimistic.

            • Nick Squillari

              Possibly we’ve interpreted the studies differently. I didn’t see this as being prescriptive, rather more descriptive in terms of facing off the two primary pacing strategies that could be employed in a time trial. The authors both acknowledge the physiological limitations that come in to play in addition to never suggesting their simulations should be taken literally. It’s a base of knowledge, expanding on a previous simulation (and likely to be expanded on again). With the extreme end of the power simulations included in an effort to further tease out if there are any patterns or variations that existing withing a smaller subset of power values.

              Is there – like you’ve beautifully outlined – a whole lot more to pacing in a TT? Sure. Are those deeper points however potentially over the head of a lot of causal time trialists – and part of the barrier creates the perception that you need to apply something akin to quantum physics to enjoy and post a decent time in a TT? I’d argue that can be the case. Our intent was to try and help improve the times of those who want to go a little faster and whose knowledge base starts and ends with “hold ‘x’ watts +/- 2% for entire TT duration” regardless of terrain or conditions.

        • I like this simple advice :)

    • Cyco

      Several years ago I read an article on Boardman’s pacing in one of the longer TdF TTs. Cycling Weekly, iirc, said his and Keen’s studies showed that constant HR gave the best times.

      He was able to hold his Threshold HR +-1 for the 50ish km of the event, by pedalling down the hills as well as up. It was a ‘rolling’ course, no long or steep climbs, and their belief was that the time loss from recovering on the downhill sections was greater than the time gain by going harder up hill.

      • Alex Simmons

        Fortunately we’ve moved on, although Keen is a smart guy and was trying to make good use of the tools available. He and Boardman were early adopters of power based training.

        Pacing with HR is tricky since HR has a sizeable response time lag compared with actual energy supply / power output. It’s like driving your car by looking in the rear view mirror. By the time HR get’s above target, it’s way too late and you’ve burned a match. Keeping HR steady through a TT also means one’s power is likely to fade overall through the event as cardiac drift is normal when riding at threshold.

        • Paul Jakma

          It’s possible that an experienced rider can “feel” an increase in load in those systems that lead the HR and manage those to keep HR steady. I.e., the rider may be integrating more inputs into their “controller” than just the HR, with the aim of keeping HR steady.

  • Micky D

    Nice article. Thanks Nick. Another way to think of the variable pacing strategy on the hills is that they generally involve some descending afterwards. The descent offers some recovery time at a lower power to balance out the uphill effort. On many descents it is simply not possible to hold your target average power for the TT anyway.

  • massarob

    This is another interesting study, accounting for physiological components (CP & AWC)
    Thank you for this article.

  • Damian Eagle

    Nick, I was that rider in the lower grade it seems.

    You obviously spent a lot more time trying to work out why I took out the first half of that race than I did :-)

    A very interesting article, and I could only aspire to be able to maintain your consistently high power figures for around 30 minutes to prove it out. But unfortunately I hadn’t done enough of the right training …

    My training had mainly consisted of time-poor sub 5 minute high-intensity work on my commute, which makes for some great Strava segment KOMs, but not so great for sustaining moderate-high average power for 20+ minutes in a real-world TT.

    It meant that through the last TT season I’ve been going like a demon for the first 10 minutes of each time trial (irrespective of power output it seems), and then gradually losing power as the duration of the TT did not match my training profile.

    I do intend to do some more endurance-style training coming into next season, and would appreciate it if you could keep up this level of analysis of my races as I go :-)

    • Nick Squillari

      I didn’t want to name you Damian, but that out leg did make me chuckle. Especially when I married it up against your finishing time. You rocketed out to that turn!

      I’d also argue the training you’re doing is in fact pretty good for TTs and that what was missing was a better pacing strategy, not fitness. You would have destroyed your W’ on the out leg. No amount of training for any rider would be able to make up for that and replenish it enough to be able to tackle a head wind like we had heading home.

      So a little better pacing and you’d have been, undoubtedly, minutes faster.

  • James Baggott

    really interesting…….. on loads of my local roads where I train, where I know the rough time spent climbing, where and what the gradients are, I have been applying my power curve numbers and been getting fast times when riding solo, which directly reflects what you have been saying. And I always treat a headwind like a gradient :)

  • ilciclsta

    Best bike split is a good tool to figure out pacing strategy for time trials, it takes into account the forecast weather conditions as well as the course profile. You can set target time, target speed or target wattage and it will give you tons of info on how to pace using power. http://bestbikesplit.com

  • Luke Bartlett

    What did Boardman say about doing the hour? “if you can hold this pace for the hour you aren’t going hard enough, if you can’t hold this pace for the rest of the hour you’re already cooked”. Obviously, that is on a track, where sustained power with no fluctuations would be ideal.

    and obviously wind could be treated as a climb except for the fact that rider power:weight is important on a climb whereas absolute power is what matters into the wind, assuming you can get aero.

    I would have thought that you have to look at it from a physiological cost vs time cost. ie, if you can sustain a higher power on a climb and gain X minutes, you can therefore hold slightly less on a flat/downhill/tailwind in order to be able to sustain the higher power on the climb/headwind. obviously putting huge power up a climb will gain more time over rivals than huge power on a flat because wind resistance will increase cubic with speed, and lower speeds more power goes into your speed on a climb.

  • Andy Ross

    Terrific article. We have done some testing using the ProShift Automatic Shifting System and have found additional insight in controlling and optimizing power/cadence further improves performance. Please visit http://www.proshiftracing.com.

  • Alan Doughty

    Factoring in rider physiology and the results could be very different. A non specialist TT rider whose strengths are power at VO2 max or anaerobic would conceivably find variable power a better option. A deficiency of the article is it treats riders as if they are electric motors with the only influence on times being pure physics.

  • Alex

    In the simulated studies was the NP of each trial run the same? I think that is a potential issue in this study because a variable pacing strategy can make it harder to maintain FTP or near-FTP after having a power spike. I used to do FTP interval workouts with 1 minute of VO2 for every 3 minutes of threshold for a total of a 12 minute interval. It was extremely painful to complete 2 let alone 3 of these. I don’t think I could have done this for the length required in a TT.


Pin It on Pinterest

January 18, 2017
January 17, 2017
January 16, 2017
January 15, 2017