The cheerily-named French website cyclisme-dopage.com has been hosting research by engineer Frédéric Portoleau who estimates the power needed to ride with the leaders over the main mountain passes.
The estimates suggest the Vuelta saw some of the highest power-to-weight numbers of the year. Some will use this to draw conclusions but for me it’s more a tool to analyse the racing.
The Methodology
Many use power meters. These are strain gauges incorporated in the cranks and transmit data to a computer mounted on the bars which displays real-time output and saves the data for post-ride analysis.
Portoleau instead uses estimations. He’s not a bio-mechanic, instead he’s a software engineer. But it’s not as wild as it sounds because the engineer only attempts this on a climb with a significant gradient. Ideally a slope above 6% where the speed is below 25km/h and on a day without much wind for this removes much of the aerodynamics out of the equation and therefore the model is easier to work since you don’t have to factor in the effect of one rider sitting on the wheel of another. It’s similar to the use of VAM only more comparable from climb to climb because the VAM number tends to vary substantially with gradient, more so than power.
Portoleau then introduces the concept of puissance étalon or “reference power”. Rather than know the exact weight of the particular riders and their bikes, he instead works out what power would be needed for a 78kg rider to go alongside the rider in question. So if you weighed 68kg (150lbs) and were on a 7kg bike with clothing and water making another 3kg that’s 78kg meaning you’d have to average around 445W to stay with Rodriguez on the club to the Lagos de Covadonga.
That’s the data from the climb to the Lagos de Covadonga from Stage 15. Most headings should be self-explanatory but remember the puissance étalon explainer above, it’s not the power produced by the rider named, but the output needed for a 78kg rider and bike to match them, therefore it allows for comparisons. Note the measurement is taken during a section of the climb and excludes the final part with its irregular gradient. The full climb took 33.40. It is still a short climb and came after a relatively flat day of racing meaning more explosive riding was possible.
This climb has been a regular in the race and so we can use it to compare performances over time. Here’s an estimation of the winning power needed each time over the years:
Adjustments were made for the 1980s to allow for rougher roads and heavier, inefficient bikes. Roughly speaking the reference power for our 78kg rider and bike is 420W in the days of Delgado and Millar, it rises to 475W for Pavel Tonkov, Laurent Jalabert and Alex Zülle before falling back a bit. Clearly there’s been a big jump in power for the mid-1990s when EPO use was widespread.
Here’s the data from the Bola del Mundo:
The Bola del Mundo is a good one to analyse as it was one of the longer ones in the Vuelta and therefore comparable to other long climbs from the Tour and Giro, and it also came after some longer climbs earlier in the stage. All three riders overall waited earlier on the climb and it was only later than the riders reached their max, in fact Rodriguez set a new record for the climb but it had only been used once before in 2010 and this time there was a light tailwind too.
Overall Portoleau’s verdict is that the Vuelta saw the highest wattages of the the three grand tours. Our “reference rider” would need to pump out 425W to hang with Rodriguez, Valverde and Contador when the estimated number for Ryder Hesjedal in the Giro was 400W and Bradley Wiggins pumped out 415W in the Tour.
My take
This is a useful but like all analysis, it’s how you handle it and the context that matters. For example the numbers in the Vuelta could be comparatively high because many stages had little climbing until the final ascension. Unlike the typical mountain stage, there was no series of high mountain passes to cross before the final ascension of the day meaning fresher riders for the final climb.
Is there any proof of doping? No. There is only proof that Portoleau has been estimating data from the race, the rest is speculation. The data are like a Rohrshach Test. If you want to see supporting evidence of doping you will spot it. If you want to see evidence of clean riding, you will spot it.
Portoleau has worked with coach Antoine Vayer, the former Festina team coach and Vayer often uses the numbers to point fingers. Right now Rodriguez is on the frontier. As someone who became familiar with watts in the late 1990s, today’s numbers look rather low but we can see they still remain above the 1980s. But any more is speculation, get out the crayons and join the dots to suit your view.
On the methodology, Portoleau has compared his work to real data taken from the SRM system used by many riders and it has matched up very well. But I’ve seen cases where his numbers don’t quite match too. A well-calibrated SRM power measurement system taking power direct from a rider has a +/- 2% margin of error meaning even capturing the data live is hard. Given this it is possible, it means that a figure of 415W for Wiggins could be 406W at the lower end and 423W at the higher end… and therefore comparisons across the season are very hard.
Conclusion
It’s tempting to draw conclusions from the numbers but if Rodriguez’s numbers are high, they are not outrageous and there’s a margin of error in the numbers so we need to be careful not to take every number right down to the last watt.
Instead we can use the numbers to help understand events in the race. In the past we might have looked at the average speeds of stages to get an idea of how hard the day was but that’s not a useful statistic as the peloton tactics vary, as does the wind. Timed ascensions, like Alpe d’Huez are better comparisons but they still aren’t perfect as the wind varies and other factors come in to play. VAM is useful but its variability with gradient means comparison are harder. In short power is the best measure of effort and since only a few riders share their data, estimations are the next best option.
Maybe one day we will all see the live race telemetry? SRM did offer this for some riders and you could view several riders’ power, heart rate and speed data online during the Tour de France but this was banned by the UCI at the start of the year.
Good article, as always!
Any idea why the live SRM telemetry was banned by the UCI at the start of the year? Seems odd, but coming from the UCI, nothing surprises me anymore…
Thanks. I think the ban came in because the UCI doesn’t want third parties to broadcast things from the race: only the TV company has the right. On-bike cameras are banned and this was extended to the data.
Personally I think we should be going the other way and making rider data available for everyone. If the UCI hosted a website where this was available for all riders in a race it would be very profitable. And we could have on screen graphics for digital TV viewers who want the stats whilst most of the audience – who want the scenery – can leave it out.
+1
this might be only a technicality, but wouldn’t you think live telemetry to have been banned due to race tactics? it’s easy to see how strong each rider is on a specific climb, then plan your break accordingly. (i might not be thinking straight, it’s getting late)
I think that’s a big reason. Does the sport need to have ever more docs and sports scientists analyzing data so the DS can scream into the rider’s earpiece -“keep going, your rival’s at XXX watts, he can only hold out for XXX more meters!”? It was bad enough when BigTex was supposedly getting the lowdown from The Belgian who was on the phone with Ferrari who was watching them climb Ventoux on TV in Italy. When I watch MOTOGP on TV I don’t give a rat’s a__ about how many rpm’s Stoner’s bike is cranking up or how far he’s got the throttle cranked open. Nor do I care much about the on-board shots. I want to see MAN vs MAN racing. Telemetry should be banned, not only from cycling but MOTOGP and F1 as well. Drives costs up and sporting qualities down. One of the few things I think the UCI does well is their efforts to assert the “primacy of man over machine”. Well, maybe they don’t really do it well, but at least they try.
I agree with your view. Mine is that if those statistics are available to the rider or DS, they might as well be available to the viewer. Best if they were simply unavailable.
Formula 1 – famous for keeping data secret still manages to transmit, on screen, live telemetry data including rpm, speed, g force, KERS status etc. I am not sure that SRM data would be any more revealing. Except it might reveal what the UCI wants to keep secret.
In F1 we also get live team-driver race radio as well at times. And we get it in V8 super car racing as well, indeed commentators in V8 super cars can talk with drivers live during the race. It doesn’t affect race outcomes and adds to the enjoyment for TV audience.
No need to prevent data being made available post-hoc. That’s not a UCI/race organiser issue anyway. It’s my power meter and my data, I’ll do with it whatever I please.
Live data feed/transmission however is simply a matter of money, nothing more. Pay the race organiser enough euro for the right to transmit data from their race, and hey presto, ye shall have live telemetry.
>In F1 we also get live team-driver race radio as well at times.
This is incorrect. It’s not live, but on a couple of minute delay.
Fair enough, it might be on delay in some sports. The point is they are transmitting actual conversations to the viewers.
Can’t find the link atm, but there was a post with Jens Voigt’s SRM data after his USA Pro Challenge solo win. Would be nice if Garmin (who sometimes share this) would show the SRM data of Talansky for the Vuelta, to compare it with the above method.
It was a Velo article where Jens’ data was posted I believe (try here if the link works: http://velonews.competitor.com/2012/08/analysis/power-analysis-jens-voigts-winning-power-file-stage-4-of-the-usa-pro-challenge_235832)
The above analysis reminds me of the sorts of work done in the past by Dr. Ross over at SportsScientists.com, which is linked to at the right. If I recall from their work, w/kg values in the high 5s for extended climbs of around 40 minutes seem to be at the reasonable upper-limit for a clean rider. If a climb of that duration saw numbers in the mid-6s it would likely raise questions.
Could you imagine a broadcast with live rider telemetry on the screen (HR, power, cadence, estimated calories…) and how that would change the viewing experience, especially for those that ride. I realize that teams and riders may have an interest in protecting this info, and the UCI in monetizing it, but maybe each team could designate one rider (likely a domestique) to participate each race.
+10 to the last idea…balances it out and so interesting to the viewer.
The Sky website had a link to one of their co-sponsors which put out Eisels data on a daily basis during this years Tour. Included not just power, but also altitude, speed and HR.
Data overload perhaps, but seemed to be a good tool for the sponsor to market their goods.
“Vayer often uses the numbers to point fingers”
That is very true.
Here’s what he said about the Tour de France:
“The last comparison is more thrilling than bluffing. In 2011, after 16 years of scrambling for heavy doping products, we were at last cheering, in these columns, for the absence of riders performing above an average of 410 Watts on the last ascents of mountain stages : the detection threshold of poison. Alas ! There is again four of them, this year who crossed that bar : Wiggins, Froome, Nibali and Van den Broecke, with 415 Watts for the first three of the classification, and 410 Watts for the fourth one. We are now longing for 2013 and the return of Contador and his tainted meat ! Until then, it’s doubtful a cure will have been found.”
That’s my point. He sets out 410W as an important threshold and accuses those at 415W to be doping. Only we don’t know if 410W is the upper limit and the reports of 415W are still an estimate. A 5W difference is easily within the margin of error but when he looks at the Rohrshach test he sees a blood bag. I’m not trying to make claims either way here. But Vayer’s certainty should be more measured.
I agree
There is something quite outlandish about claiming the rider is doping using this particular method. If he had access to SRM readings or Powertap for individual riders, then he might be a bit more surefooted in these claims. It seems quite strange to use 410 as a watts level, as someone with a build like Pantani may not be able to put out this power, yet still might be doping.
If he thinks 5 Watts (regardless of weight or other details of the ‘mountain stage’) over some fairly arbitrary threshold means anything this guy should go back to school.
Why was it banned by the UCI? Really annoying…
The UCI wants TV viewers. It does not want positive tests results, or data that are indicative of doping.
Another factor that needs to be corrected for, if you want to be able to compare power outputs, is air density, which impacts oxygen availability. Varies with at least altitude, temperature and weather. Barometric air pressure is a useful proxy, I think.
Yes, it’s good for comparisons on the day. But cold vs hot and humid vs dry makes a significant difference. A hot and humid day is significantly faster compared to cold and dry air.
Humidity has only a very minor (almost negligible) impact on air density.
The biggest influences on air density are temperature, air pressure (barometric) and altitude.
Increasing temperature, lowering air pressure and increasing altitude all reduce air density (and vice versa).
Increasing humidity (whilst very minor) also reduces air density.
Air pressure intrinsically varies with altitude, of course. A weather barometer can also be used to measure altitude. Temperature also is intrinsically related to pressure in theory, but it’s such a huge system that local air pressure need not reflect local temperature.
Common cycle computers already include barometers and thermometers, of course. Garmins log temperature data, but, unfortunately, don’t seem to log the barometric data directly.
Interesting. I’ve noticed myself that weather can make a difference. Though, on local climbs it seems like dense air (i.e. cold) works better for me, but its hard to measure this effect (so many variables). Also, note that there are 2 opposing effects:
1. As air density increases, oxygen concentration per unit volume increases, and so your body should be able to sustain more aerobic work.
2. As air density increases, air resistance should increase, for any given speed, and so you require more power to sustain a speed.
So this should mean that different densities benefit different types of ride. Fast cycling on the flats the aerobic power gain from dense air may be greatly outweighed by the increased resistance, and less dense air would be better (probably to some limit – you’re likely to cycle slower again when pressures get too low). While on climbs the increased air resistance may be minimal compared to the power gains, and denser air should be better.
Have racers noticed such effects?
For time trials, there is a trade off between reduced partial pressure of O2 and reduced air density as altitude changes. While the physical changes are the same for everyone at a given altitude change, the physiological impact of reduced partial pressure of O2 varies between individuals (and within an individual depending on acclimation etc).
There is a “sweet spot” range of altitudes where the benefit of one outweighs the cost of the other, and so overall speeds are higher.
Where that sweetspot range is depends on the relative O2 demand for the event in question.
For a longer TT, then it’s an exclusively aerobic demand, so that altitude range is lower than for, say, a 1km TT where the proportional O2 demand is lower. Hence a 1km TT record is more likely to be set at higher altitude than say for an hour record. Hour records have mostly been set at sea level, but that’s as much about where suitable tracks are located than altitude.
Ah, the length of event / the level of aerobic effort is another variable into determining the optimum? Interesting. Several non-linear functions involved here, so finding the optimum is non-trivial.
Hmmmmm… I hate to say it but, yes I am suspicious-especially of Valverde. This guy has only a handful of results all year, then on stage 17 drops Rodriguez like he’s standing still, and almost catches Contador for the win. All this on the day after the rest day. If they say you have to be stupid to fail a dope test, then maybe the testers should get smarter. How about testing a rider, then retesting him again a few hours later! If I was a doper, I’d dope right after being tested. Let’s just get serious about catching these cheats and be done with it!
I hate that I feel suspicious! But at least I feel confident the next generation is clean. Unfortunately, they have to settle for the minor placings until cycling finally gets it’s house in order. We’re getting there… I do believe Hesjedal’s Giro victory was clean!
Hesjedal’s blood values for the Giro have been released, and, a bit like the power levels above, you could see doping in there (or at least microdosing), or a clean GT win.
It’s understandable to be suspicious and with a rider returning from a doping ban it is essential to be vigilant. But there’s a line between this and going into print with accusations. Vayer seems happy to do this but I’m not confident the data is strong enough.
I would actually look at this the other way around. Maybe Rodriquez’s transfusion didn’t go so well and he was feeling crappy that day. Didn’t Valverde have teamates/other riders that helped catch Contador once he had gotten away from Rodriquez?
Yes. Valverde sat behind Rodriguez until the bottom of the climb and then jumped across first to Quintana then to Intxausti. In fact after Intxausti was used up Valverde pulled Henao and Verdugo to the line. Since two riders from the break managed to stay on Valverde’s wheel dose this say more about Valverde’s speed or Contador’s fading at the end? Handful of results all year for Valverde? Then how does he rank #5 on the current CQ Rankings? And if the Bola de Mundo offers the best comparison to the Giro and Tour then all three Spaniards are about equal to Hesjedal’s watts and far below Wiggins’ watts. My conclusion, let’s not let worrying about who might be doping ruin our enjoyment of the sport. If doping is proven later we’ll have plenty of time to be pissed off then.
From what I saw on the TV coverage, of the three contenders, Rodriguez spent the most time with his nose in the wind.
Contador had team mates plus collaborators when he bridged to the break, then Tirrelongo to help him out. Valverde sat behind Rodreuez then Quintana and Intxausti just about the whole stage.
No wonder Rodrgz was pooped at the end, poor lad.
From what I understand, the majority of the teams support the ban of live telemetry as they do not wish for their riders data to be studied by rival Director Sportives during racing where they can identify impending tiredness and communicate it to their own riders.
However, I do feel more information ought to be released post event for comparison and analysis.
I can think of a few reasons why the Vuelta scores higher on this analysis than the Tour:
-The stages were shorter
-The gradients were steeper, favoring featherweight climbers relative to the ‘reference rider’
-The mountains were lower
You don’t have to be a software engineer to do this analysis, BTW. Anyone can go websites like analyticcycling or (my personal favourite) the Kreuzotter calculator and punch in the numbers.
Pretty right.
Fascinating.
Just one thing: “light tailwind” on the Bola del Mundo? I was there, it was not a gale, but it was considerable enough to be felt by roadside spectators. (And Portoleau says: “vent fort de dos”).
A lot of the data from recent years suggests that riders are riding right at the limit of what’s considered physiologically possible. That’s going to make any slightly high number seem suspicious to people who are looking for evidence of doping. I’m sure there’s still some doping going on. However, even if some riders are getting themselves to the frontier of physiological possibility with some illegal help, at least the riders who can get there naturally and don’t want to dope still have a chance at being competitive, which might be the best we can hope for.
I don’t know what to think any more when watching the top riders compete. I find that I tend to lump riders into categories such as “clean until proven otherwise” or “has a past but confessed so maybe clean now” or “has a past and is in denial about it so regard with suspicion”, but the reality is this all amounts to just applying a my own life experience to make judgements about people that could all be completely wrong. All we ultimately have is the data from the UCI and from the outsiders and their blogs. We need the outsiders to keep making these calculations because while they may not be able to precisely identify individual cheats, they can at least tell us something about the state of the sport.
+1
Like a fan can state with any accuracy whether a rider is clean…”I just know so and so is clean b/c he’d never dope….”. Tyler Hamilton doped, for goodness sake. And in his book, he implicates Jens Voigt as well. Hincapie, JV, VdV, LL, TD, they all doped.
Truth is, if they are winning, they’re probably doping.
Except for Wiggo though, obviously.
… actually, probably including wiggo.
I love Jens Voigt, but I’ve never had any doubt that he, as a product of the East German sporting system, has spent a lot of time doped to the gills.
I don’t think he does it now, though.
How does he implicate Voigt? (I have the book, but the UK version, which apparently has been edited to remove some stuff, due to the UK’s stricter libel laws).
I just noticed something. Did Talansky (Talanski??) take a detour? His time is longer than Contador’s but his speed is faster.
Given advances in sports science and riders more ascetic approach these days wouldn’t we expect current watts to exceed those of the eighties?
Would Jens Voight doping really be a surprise? Given he raced through the Armstrong years i’d be surprised if he didn’t. It will still be a sad day when he finally retires. Assuming he’s currently clean.
Great article thanks !
There are a couple of pros posting rides on strava that include power data – Emilia Fahlin and Taylor Phinney at least. Taylor has also posted race data this year.
For interest, I’ve done similar in the past using Alpe d’Huez ascensions.
http://alex-cycle.blogspot.com.au/2010/07/ascent-rates-and-power-to-body-mass.html
At these powers/speeds, on a 6% gradient with zero wind, ~ 76% of the energy demand goes towards overcoming the force of gravity, with the balance split between air resistance (~ 16%), rolling resistance and other drivetrain/bearing friction (8%).
So while gravity dominates, there is still enough variability in air resistance from drafting/rider morphology and position to alter the estimates of power from speed by ~ 5% , let alone the wind variable, which is quite significant even if the wind is barely noticeable.
e.g. a non-noticeable 1m/s tailwind (calm conditions on Beaufort wind scale) would reduce the power required at these speeds by nearly 20W, or nearly 5%. Conversely with such a minor headwind.
IOW, take such power output estimations from ascension rates with a big grain of salt.
Of course, the steeper the gradient, the more the energy demand becomes dominated by overcoming gravity.
As for using power meter data as a means to spot performance anomalies, well “data doping” wouldn’t be all that difficult to do.
Thanks. I was just wondering if technical issues, such as differences in equipment, might affect drive train efficiency, tyre rolling resistance, and the like. I’m not too sharp on these matters as they relate to the consolidated tale-tell wattage number. I assume 410 Watts output might produce a varying time to race completion, due to the technical issues I’m rambling about. Thoughts anyone? Cheers.
Yes, time to complete any course (assuming a set power output) is a function of of the total of the various resistance forces. Reduce those and you go faster.
Drivetrain resistance on these bikes is of the order of ~ 3% of the total energy demand (let’s say ~10W), and so the issue then becomes, how much difference is there between one high quality drive train and another. Even if one is 10% better, that’s only 1W. Of course one wants the best, marginal gains and all that, but the scope of these differences is well inside the noise (error ranges) of such estimations.
As for rolling resistance, well there are indeed differences between different tyres (and tyre pressures and also to an extent the amount of tyre scrub from different riding styles on a climb). We can reasonably assume they are using the same road surface though (experienced TT riders do know how to pick smoother sections of road surface though).
So, yes some pros are riding on tyres that are worse than others from a Crr point of view. Sponsors do tend to like you to use their product, irrespective of whether or not it is the best choice.
Let’s say there is a Crr difference of 10% between two tyre types and we consider the 6% gradient scenario at these powers and speeds. The equivalent wattage impact is ~ 3W. At higher speeds on flatter terrain, that goes up a bit, more like 5W.
For an idea of the relative magnitude of each of the primary resistance forces when cycling, see this chart:
http://i220.photobucket.com/albums/dd226/ASimmons/EnergyCostbyResistanceForce-2.jpg
For any individual the absolute numbers might vary a little, but the overall relative contributions will follow this same basic pattern as gradients change. In this chart it considers a rider at 300W, ~125W less than the professional power outputs we are discussing, hence the balance between gravity and air resistance is a little different because of the speed differences.
Interesting how air resistance remains a significant factor even at higher gradients.
But of course air resistance is not directly a factor of the gradient, but of the speed. We can assume that for pros, air resistance matters indeed more than shown in the graphic. Drafting is more important when climbing than it used to be, which I believe is bad for the show. Race organisers should strive to reduce climbing speeds. The Vuelta is trying by increasing gradients, but the most effective way would be a considerable increase in the overall hardship and length of the stages.
How do you put hate, desire, revenge or “a voice in your head saying ATTACK at all costs” into these formulas as variables. Does hate or a bad childhoods, which has proven to be turbo boosters for quit a few riders over the years, work better in rain or dry conditions? Does the drinking Red Bull from Coca Cola cans topped with pain killers in the final hour of the stage make any difference on the figures. The power output for sure is a very handsome statistic tool but cannot serve as doping detector. With all the above mentioned variables in mind it is only a statistic tool for individual use. Cant help thinking on the TV images of Froome looking on his (SRM) numbers every 5-7 seconds when climbing. Do concentrate on the race – not on the numbers.
+1
To me it just shows that the Tour had a weak field.
This is the first time I see an article with credible numbers, but one problem I see is how they manage to do this day in and day out. If these guys on the vuelta were not using EPO or some other juice to get them more oxygen to their blood; I am convinced they are surely are using something to help then recuperate and do the same (or harder) the following day! Think of us when we just do our “meager” rides to key Largo! How would it feel to do it 3 weeks in a row?