In 2023 the National Hill Climb championships are being held on “The Struggle” – an iconic British climb.
The climb is 4.31km at 8.4% and gains 364m of elevation but it’s not an even 8.4% (to be honest, that wouldn’t be much of a struggle…) – it gets very steep at the top. A climb with uneven gradients becomes an interesting pacing challenge for a rider time trialing to the top.
This blog isn’t a preview of the national champs, but more of a lesson in how you might want to pace the effort. For our purposes today, we will assume a 70kg rider with the ability to average around 400W for the duration of the climb, also as this is a hill climb let’s assume our rider’s bike plus kit weighs 6.5kg giving us a system mass of 76.5kg.
A brief note on aerodynamics
The average speed for a rider doing 400W as a perfectly smooth effort up this climb is high enough for aerodynamics to play a small but not insignificant role – accounting for roughly 10% of resistive forces. A reduction in cda from 0.4 to 0.3 is worth 16 seconds under these conditions and a further reduction to 0.2 (essentially a time trial position would be required for this) is worth another 20s. For today, we will set the cda of our rider to be 0.3, which is a rough assumption as hands on the hoods with slightly bent elbows.
Test run 1: A perfectly smooth effort
I know we do this on almost every blog at the moment but optimising the power distribution over a course can lead to significant time savings.
With a VI of 0 (NP/averageWatts) we can see that this effort is perfectly smoothly paced. Notice also the Wind Penalty is minus ten seconds, telling us that we have a slight tailwind. A brief nod to the hill climb nationals would not be amiss here, if the wind picks up during the day, it’s big enough to make a measure-able difference on a climb of this length.
Using the gradient distribution to set some pacing rules
One neat feature of myWindsock is the ability to view the gradient breakdown of a course. During a climb this is particularly interesting as we can see how much the gradient is lying to us – ie how true is the 8.4% average to reality? Let’s take a look…
All of a sudden, the story of this climb looks very different. In that order, let’s make a new plan based on gradients with a general rule of the steeper it goes, the harder we go but let’s not deviate more than 100W from our planned average of 400W for the climb.
As with any pacing plan, it has to be realistically implementable thus our rules should be simple. Let’s say that anything less than 6%, we ride at 300W, anything from 6-12% we ride at 400W and anything above this we ride at 450W… What does this do to our average power and the average time for the ride?
Ok, so we can see this pacing strategy results in a power that’s slightly too hard so we need to find a way of reducing our power by 24W. That said, the combination of the pacing strategy and the increased power is worth 46s in total time. Notice how the average speed increase has reduced the total impact of the tailwind as the relative wind speed is lower.
Now we need to find a means of reducing our average power, one way might be to get aero as the speed increases. Let’s tuck down above to 20kph and reduce our cda to 0.25, this saves another 7 seconds giving us the ability to reduce the power in other sections. Reducing our flat power to 200W (essentially a brief recovery ¾ of the way up the climb) we can get that average power down to 414W. Finally, we further reduce the power on the flatter sections…
This simulation comes from a more even pacing distribution, a fast start to get up to speed then 200W when the gradient is below 6%, between 6 and 12 we ride at 350W and then full punch at 460W when above 12%. The other change is the reduction in cda to 0.25 when the speed is 25kph or above. With an extra watt, we’ve saved 17s of time!
It’s hill climb season and tiny margins matter. Yes, differences in strategy based on pacing may only be worth 17s on a thirteen minute effort but this is enough to make the difference. If you’re already doing all the other marginal gains, myWindsock can make the difference between first and tenth! Sign up here today.
Cycling and triathlon are both demand driven sports. This means that an athlete’s performance is down to being able to meet the demands of the day. Sometimes, figuring out exactly what these demands are can be difficult. If you want to execute a ride, be that a ten mile TT or an Ironman bike leg, in a specific time then you’ll need to have a good idea of what the demands for that are on a particular course. Obviously, we recommend using myWindsock for this.
A demand driven approach is quite a simple one. You identify a goal, figure out what’s needed to achieve that goal and then prepare specifically to meet those demands. It’s a means of identifying gaps in your performance profile that are limiting your performance and broadly is how world class coaches and performance engineers approach problems. For example, it’s easy to say you want to run a race, ride a TT course or something in a given time but actually breaking down that goal into trainable components and addressing these specifically is how you’re likely to achieve it.
With anything, analytics tools are useful for this. We would put myWindsock in the same category as something like Training Peaks, WKO5 or a lactate test with regards to its usefulness from a performance perspective. We don’t focus on physiology, but a good performance in cycling (especially a time trial situation) relies on maximising the ratio of energy in to energy out. Your physiology is your physiology and come race day you’re unlikely to suddenly crank out fifty more watts than you’ve managed in training but you can certainly maximise the power that you have available with aerodynamic testing and optimised pacing.
A demand driven approach to a race
I’ll use a race that I’m planning to race as an example of this. It’s a middle distance triathlon in Ibiza and the bike leg is around 90 km with roughly 900 m of elevation. In order to remain competitive I’ll need to be off the bike in under 2 hours and 10 minutes having used as little energy as possible. I’ve got around 300W at my disposal (while still being able to cobble together a decent run) for this so we will have to take that into account.
The race route runs the length of Ibiza and is not flat. It is 90km with around 900m of elevation – a course that is described as ‘rolling’ by the organisers.
Let’s start with a power budget of 310W which we will cap as my normalised power upper limit, while I know that I’m capable of more than this – one demand of a triathlon is that you have to go running after the bike leg and in order to execute this run well, keeping the ride as close to the first lactate threshold as possible is optimal.
As we can see, the goal discussed below is reasonable with a completely smooth application of 310W. We have discussed in the past how this is not a particularly efficient means of applying power during a bike course.
Given the goal is to execute a bike leg under 2 hours and 10 minutes, this ‘smooth power’ input serves a sanity check as to whether or not this goal is actually achievable. From this, it seems likely that this is the case so let’s optimise the pacing strategy within the performance constraint that we’ve identified.
The first step to this is check where on the course over-pacing climbs makes a difference. We do this with the “where power matters most” feature on myWindsock.
This plot, under experiments, shows us where on the course that an increased application of power will provide a greater return on investment for speed. This will be familiar with many experienced time trialists as it shows us steeper sections, brows of hills and headwind sections.
Now, we simplify the breakdown of where power matters most and implement this as rules.
As the road tilts up, I will ride at threshold. If we really slow down, I’ll ride at 400W. It’s important that any pacing plan is simple and easy to implement. On top of this, I will also ride in zone 2 above 50kph and stop pedalling over 60. This is not ideal for a straight TT but in triathlon, conserving energy for the run is very important.
We can see here that for a lower average power, we can go a couple of minutes quicker than the previous 2:09 with the smooth power. Playing around with different pacing strategies we can see changes in outcome time.
It’s better to pace climbs slightly harder than flat sections (which you can read about here on our blog) and given we know what power we can manage for given durations we can come up with a training session to help prepare the specific demands of the course. Using myWindsock’s rules feature we can figure out how we will pace these climbs, build an interval session and plug it into Training Peaks for all or part of the course. This means we can train for the specific demands of race day on the turbo trainer at home.
World class athletes and coaches all use a demand driven approach to competition. Setting a goal and achieving it requires an approach that identifies the things that are currently preventing that. myWindsock is a tool for this, one of many and one used by Olympic Gold Medalists, professional cyclists and teams as well as the UK’s best time trialists.
Sepp Kuss is currently in red going into the TT on stage 10 of La Vuelta. It’s a bit strange, no one expected him to be in the race lead but – if he can get through tomorrow, he has the class in the mountains to win GC. Remco is 2:22 down on the American with Kuss’ Jumbo team mates, Roglic and Vingegaard on 2:29 and 2:33 respectively.
The TT is only 25.8km in length so, while we all expect these gaps to close, the chances are Kuss will still be in the race lead unless Marc Soler or Lenny Martinez pull a really great TT out the bag. Sepp Kuss’ time trial ability has rarely been tested in the past but we all know that Jumbo can execute a good TT when they want to.
The time trial is only 25km long but we saw in this year’s Tour de France that short time trials can create big gaps but this time trial is, for all purposes, completely flat. This both benefits Sepp Kuss and hinders him. The benefit comes from the fact that not much time can be lost during a medium length flat TT as the speeds are high, however relative to the rest of the GC contenders Kuss is one of the better climbers – and his TT ability remains unknown…
Two things of note come up when thinking of this course, firstly it is basically pan flat and secondly there are a few technical sections. This time trial is essentially a test of watts and cda though – you’ll see riders who have spent time in the wind tunnel will have a big advantage – but more on that later. The main takeaway here is that the course is relatively simple and suits the pure TT specialists.
The race conditions are also relatively trouble free. Here we’ve put some numbers in to represent your ‘average’ grand tour rider and you’ll notice that the ground speed is extremely high for this power. These high speed conditions will suit Remco Evenepoel over his GC rivals and it will also suit Ganna for the stage win. The conditions remain favourable throughout the day and are quite stable seeing no shift greater than a couple of seconds due to weather. That said, this could be the time separating stage winners – look out for that in our Vuelta TT Review.
This plot shows us the effective wind that riders will feel during the course of their TT. These extremely shallow yaw angles show us that a rider will feel the wind mostly on their front and back. This is a relatively simple wind direction distribution and will suit the GC riders that haven’t done a great deal of testing or are experienced time trialists. Simply put, there’s less to be gained by knowing which wheel is fastest in crosswinds vs headwinds which is knowledge some teams will have that others don’t. These simple conditions equalise the TT somewhat.
While this course is flat, there are parts of it where power will matter more than others. There’s a few steep ramps at 3km and just before 6km. The orange bars on this plot show us where inputting a decent amount of power will have the most return on investment. The most interesting instance is the orange bar just before 8km. This is a portion of descent that flattens out and pushing hard on this section before the road drops down again is an energy investment that leads to more speed being held on the lower part of the descent.
CdA and Time
Other than Jumbo and Sepp Kuss, I don’t think anyone knows if he’s spent any time in the wind tunnel. I’d imagine some rudimentary testing has been done but the question of how aero Sepp Kuss actually is seems to be a little bit of a mystery. Apparently, he’s 61kg with an FTP of 383W (this might be nonsense though, I just found it reported online somewhere) but let’s put these numbers into myWindsock and see how varying the cda impacts the time.
It’s clear that aerodynamics makes a huge difference on this course with time spent in the wind tunnel being worth almost two minutes on this course. Sepp Kuss will hope that he’s aero enough to maintain his position on GC. It’s clear he has the power to do a great time trial though. Aerodynamics won’t be a priority for Kuss who primarily rides as a mountain domestique so his time will depend on two factors – how aero he naturally is and how fast Jumbo’s equipment is.
Remco vs Ganna
As well as the GC drama there’s the prospect of a stage win up for grabs. Let’s estimate some times for Remco Evenepoel and Pippo Ganna to see who will come out on top using what we think we know about their various numbers.
Remco’s predicted time is on the bottom, Ganna’s on the top. These numbers are estimates and the time prediction can of course change with the weather but we notice two things – Ganna’s weather is marginally more advantageous and the course suits him slightly better. The official myWindsock prediction is that Ganna will win the stage… just.
Working out whether or not crosswinds will produce echelons is a key part of any cycling enthusiast’s arsenal of skills. It allows you to answer the vital question – should I bother watching this or just catch up on GCN? myWindsock, as per, has the answer! That said, it’s the sort of thing that is a little easier with a bit of knowledge – though one day myWindsock might have a feature that delves into crosswind likelihood in bunch races… perhaps.
Under what conditions do crosswinds cause echelons? Luckily, in 2021 someone did a study…
In crosswind scenarios, road cyclists encounter additional air resistance and a destabilising sideways force. Under such conditions, a collective of cyclists adeptly adjusts their positioning to mitigate these forces by forming an echelon. This echelon formation involves a diagonal single-file line of riders positioned diagonally across the road. This contrasts distinctly with the arrangements chosen in wind-free environments.
To scrutinise the impact of crosswinds on the drag and lateral forces experienced by cyclists, we conducted wind-tunnel trials using a scaled-down model of a cyclist. By employing a load cell, we accurately measured the exerted forces. Various configurations were explored, involving one, two, and four cyclists, with alterations in yaw angles.
The findings reveal that in a fundamental arrangement featuring four riders at a yaw angle of 50 degrees, a sheltered cyclist within the echelon encounters less than 30% of the air resistance experienced by the cyclist positioned outside the echelon, who struggles against the crosswind along the roadside. Moreover, we demonstrate that the adoption of an echelon formation in crosswind conditions becomes advantageous only when the yaw angle surpasses 30 degrees. At this critical angle, the air resistance on the exposed cyclist increases twofold when the gap to the leading group expands from 10 cm to 1 m (on a real scale).
These outcomes hold substantial implications for devising strategies in road cycling races. They also highlight specific configurations worthy of deeper investigation through more intricate experiments.
The main key takeaway here is this:
The study identifies that the echelon formation is most beneficial when the crosswind’s yaw angle exceeds 30 degrees. Beyond this threshold, the benefits of reduced air resistance are prominent.
How can myWindsock help us then?
We have recently added a host of stage races to myWindsock including all the grand tours, just head to planner and hit “La Vuelta” (the grand tour that’s on at the time of writing) and you’ll see all the stages ready to load into the planner. We originally did this to help journalists and teams out with pre stage analysis but it’s also a handy feature which tells the viewer whether or not they should bother watching.
In fact, as I write this now I have the Vuelta on and a rider from some random continental team is on a hopeless solo break while the peloton enjoys a coffee ride behind him. The key lies in the Yaw Distribution plot…
Load the race of your choice into myWindsock’s planner, if it’s not available directly on our site or app then you’ll probably be able to find one from the organisers.
Open the route and set the speed to something roughly similar to the race.
Open up the yaw distribution plot…
This tells us the number of minutes across the route where the yaw angle exceeds various points – it’s set into bins.
A rough rule of thumb would be if the total time above 25 degrees of yaw exceeds 30 minutes, something might happen…
Let’s check today’s Vuelta stage…
The stage in question, a pan flat ‘boring’ stage but that wind flowing off the coast might cause something to happen…
This is the yaw angle distribution plot. This is what we might describe as a ‘typical’ day with yaw angles focussed around zero to ten degrees. It’s unfortunate today that it seems this particular stage will be a bit of a snooze fest however – echelons seem unlikely. For something interesting to happen we would be looking for a lump around 25 to 35 degrees of yaw or perhaps an even distribution of yaw angles which would simply tell us it was blowing strongly all day.
Yaw angle tells us the ‘effective’ wind angle, it’s the wind we feel when riding rather than the wind blowing on a stationary object. myWindsock has all sorts of features like this which will have you lining up at the start of your next race, chaingang or group ride more informed than any other rider – sign up here.
The national 25 is almost upon us. The best time trialists in Britain will gather in Scotland to unleash their watts onto the WE25/03. With CTT providing such a helpful course description, we thought this preview might be an opportunity to check out the advantages that can be gained from using myWindsock to check out the course relative to your rivals who only check CTT.
The good people at Cycling Time Trials put together a course description, elevation map and GPX file for the good majority of open events hosted there. It looks something like this…
CTT Course Description
Start on the A926 west of Forfar Academy, proceed west to join A90 southbound to leave the A90 at the Forfar south exit signed A94, take the 1st exit at the roundabout to go under the A90, take the 1st exit at the next roundabout to rejoin the A90 northbound to leave at the Brechin south exit signed A935 to take the 2nd exit at the roundabout to rejoin the A90 southbound to proceed to the second Forfar exit signed Forfar and Kirriemuir A926 to take the 2nd exit at the roundabout to finish on the A926 heading towards Forfar shortly after.
It’s comprehensive, for sure. I can even imagine the readers and riders thinking, who on earth would I need myWindsock? I have all of this information for free on CTT?
Let’s check the course out with myWindsock and see what insights we can gain…
For the purpose of today, we will avoid weather related chat. For up to day information on conditions from the best provider of cycling weather information you can either follow us on Instagram, where our story will be updated regularly in the 48 hours leading into the event or check the forecast out here yourself. For all insight and analysis available, sign up to myWindsock here.
The most interesting feature on this course is the 219m of elevation gain, not hilly by any stretch but considerably rolling for what is essentially an out and back along the A90. This will require riders to pace the effort more efficiently if they’re looking to win. This is not a course that can simply be settled as a watts per cda contest. It’s plausible that a rider that is less aero and less powerful could beat a rider more so simply due to pacing. The climbing is one factor, the gradient distribution is another, the vast majority of the climbing is done between 0 and 3 percent. Check out myWindsock’s gradient distribution plot below…
The gradients are primarily shallow with almost all of them being less than 6%. This is a course that should be ridden in the aero bars whenever possible. It’s still the case that on shallow climbs the majority of the work is being done against air resistance.
Roughly 15% (though this number will change rider to rider) of the work done on this course is done against gravity. This means, for a rider averaging 300W across the course, 45 of those watts will be used to overcome the elevation gain. That’s not insignificant by any stretch – this said, most of the climbing done is at relatively high speeds so this course might suit lighter riders that climb well in the aero bars. There’s definitely a more significant watts per kg component than we usually see in a time trial along an A-Road.
Where power matters most
myWindsock premium users can see highlighted sections of the course where power output has an outsized effect on speed. These sections are usually the steepest parts of climbs but can also be accelerating out of sharp corners.
Rolling time trials like this require a slightly different approach to pacing than a flatter TT – the most efficient effort physically is not the most efficient effort physiologically and you can go faster for less watts by focussing the power in the right place. For more generic advice on pacing a hilly time trial, check out myWindsock’s blog on this and also using rules as a pacing tool. In order to be the most informed and best prepared rider on the startlist, check out myWindsock here.
New to myWindsock are Virtual Athlete Rules. They are, to a programmer, what we call conditional operators. Conditional operators are fundamental constructs in programming that allow you to make decisions and control the flow of your code based on certain conditions. They enable your program to execute different sets of instructions depending on whether specific conditions are met or not. The most common conditional operators are if, elif (short for “else if”), and else. These operators are primarily used in languages like Python, C, C++, Java, and many others.
In myWindsock – they work broadly the same as in high level programming languages. For example, we might want to see what would happen to our overall time if we drop our power to a given number over certain speeds, sit up during certain portions of the course or raise our power over certain gradients.
Setting Rules on myWindsock
Firstly, add a course or segment to your planner. I’ve added a triathlon that I’m racing in October in Ibiza – pacing is vital in triathlon so playing with a pacing change on a triathlon course seemed smart.
We start with a perfectly evenly paced 300W for the bike course. This gives us a time of bang on 2:15 for the bike.
How to set rules…
Open the forecast on myWindsock.com
Scroll to settings and click the orange “Change Settings” tab
Here is where you will find performance rules…
Let’s add some rules and see what happens…
I’ve added speed related rules – the aim is to simply whack climbs and coast fast descents…
This has taken almost two minutes off our bike split for less power and essentially the same normalised power (I certainly can’t really tell a 1% change in NP). We can see here where the rule change has allowed us to use less energy.
Let’s take it a little further and shrug extra hard on the flat sections and let’s say, for argument’s sake that this reduces our frontal area by 5% and this has a 1:1 mapping with cda (which may or may not be true – seems like the sort of thing you might want to test). The other thing we will do is really whack the steeper sections and hit 400W when the road tilts over 9.9%. The assumption of the rules is when none of the conditions are met we ride at a steady 400W.
This has shaved another 23s off our total moving time and reduced our average power by a single watt.
Let’s further adjust the pacing strategy by simply setting the uphill power to 330W and the downhill power to 200W – which will be the default settings until our pacing rules that we’ve set kick in as conditions are met. We will also alter the cda values to a slight increase when climbing and decrease when descending…
We’ve managed to further drop our time by another two minutes with a lower average and normalised power by descending slightly slower and climbing slightly harder.
We can see here that, as well as pacing a long effort evenly, riding the road is also important. Riding harder on the climbs and gentler on descents will mean you get from A to B both faster and having used less energy. At the end of our little experiment our moving time is over 4 minutes faster than the perfectly evenly split effort for less watts.
If you want to take your pacing and race planning to the next level, join athletes who have won Olympic medals and World Championships by signing up to myWindsock here.
Recently I was out for a ride with a couple of people and we were discussing how best to pace a rolling bike course. The questions that came up were the following…
Is it faster to ride a smooth power or ride climbs harder?
If it’s faster to punch up climbs, how hard should I do this?
If it’s faster to ride climbs harder, why is this the case?
The plan here is to take a look at some physics, then use myWindsock to answer the questions that we’ve asked…
Aerodynamics and speed
As you ride a bicycle, you experience resistance from the air that opposes your forward motion. This resistance is called aerodynamic drag. When you move slowly, the drag is relatively low, and it doesn’t significantly impede your progress. However, as you go faster, the resistance from the air becomes more pronounced, and the aerodynamic drag increases.
Imagine sticking your hand out of the window while driving a car. At lower speeds, you’ll barely feel any resistance, but as you increase the speed, you’ll notice that your hand encounters more and more force from the air.
This relationship between aerodynamic drag and speed is not linear but rather exponential or quadratic. It means that when you double your speed, the aerodynamic drag doesn’t just double; it increases significantly more. This non-linear relationship is why it becomes much harder to maintain high speeds, especially on a bicycle, as you need to work against the increasing resistance from the air.
The non-linearity is what’s important here. It tells us that not all input power is created equally. Remembering that power is the ‘time derivative’ of energy – ie given a total energy budget, the power is the speed of energy expenditure. High powers deplete your energy reserves faster and low powers deplete them slower. Riding at 300W for 5 minutes is the same amount of mechanical work as riding at 600W for 2.5 minutes (of course, it probably uses less energy as humans tend to be less efficient at 600W).
So we have two concepts here – an energy budget for a given course and the idea that when you go twice as fast, the resistive forces increase 4 times as much. This should lead us to conclude that by going harder when the bike is moving slower, you spend less energy on your surroundings. The increase in speed for a given increase in power is greatest when speed is lowest.
We have our first answer then, it’s faster to ride the climbs harder as opposed to riding a smooth power.
How hard should I go?
This is the age-old pacing question. If you cook yourself on every single climb – you’ll likely run out of gas at some point during your race so you probably don’t want to do that, you don’t need a physics engine to tell you that riding every 5 minute climb in a 25 mile TT as hard as you can is not smart – the second climb will feel very hard. That said, go too gently on the climbs and you’re wasting watts in the wind on the descent. I think it’s time to run a myWindsock simulation and have a chat about power-duration curves…
A power duration curve from 2023 for an average time trialist (me).
This is my 2023 power duration curve and it tells me my best efforts for this year, this is how I would judge the power that is sustainable on climbs. For example, imagine a 25 mile TT with 4 climbs that total 5 minutes in length along the course. The way I would do it, is check my duration curve and add up the lengths of the climbs. So in this race, the duration of climbs is 20 minutes (4 x 5 minutes) – thus I would pace the climbs at this power. There’s no science behind this method, but personally I find that this allows me to recover quickly enough on the descent to begin pushing on the flats.
Let’s go into how myWindsock can tell us which is faster…
Simulation time…
The windsock map of our chosen segment, a climb then straight into a descent.
For this example I found a segment called “climb and descent” in France. Which, on the day of writing, happens to have a stonking tailwind.
Run 1: Even Split Pacing
We can see a very low wImpact but notice the time, 14 minutes and 15 seconds and the weighted power and the average power are the same. Now let’s try an extreme power split and see if we change the time.
An insane pacing strategy – whack the climb and freewheel the descents
This pacing strategy is much the same but with a lower average power and a higher weighted power. This lower average power means the same time (roughly) is achieved with a lower overall energy expenditure. That said, you need to train to achieve robustness to these higher average powers.
A more sensible pacing plan…
This effort is almost a minute faster than the smoothly paced effort despite being physiologically almost the same. This suggests that there’s an optimal measure of whacking the climb and peddling on the descent in terms of a pacing strategy.
Using myWindsock to set some rules
We have set a rule such that if the speed of the virtual athlete is greater than 60kph, the power that athlete rides at goes to zero. We notice a loss of 13s here but the power reduction is huge. This kind of selective pacing is something that may interest triathletes in particular as the aim is to get off the bike with as low an average power (thus less energy expended) as possible to be in the position you want to be on the run.
These personal rules are quite easy, can be programmed into a pacing file or just remembered. If I am going at some given speed, ride at some given power. Using myWindsock rules is an excellent, simple, means of setting a pacing strategy for a rolling course. By calculating your cda you can work out roughly how many watts you lose to air resistance at a given speed and once this is over a given number – which will depend on duration of the race – you should stop and freewheel.
myWindsock has a host of features to help you nail your next race – sign up here.
We get a lot of professional triathletes in our Instagram DMs asking various questions and getting us to do various simulations. One thing that’s of great concern to this group is the impact of drafting in a race. In Ironman branded racing, athletes must ride 12m apart (though it’s observed to sometimes be slightly closer than this on occasion). Sometimes there’s a 20m draft rule but this is often enforced inconsistently so for the purpose of today we will focus on the difference that can be made by drafting at 12m.
Without spending days doing a bunch of computations – we can have a look at some off the cuff experiments done by various internet nerds and input the numbers they received into a route in myWindsock to have a look at time savings based on what a pro might do.
Our virtual pro
In order to run the numbers, we will simulate a virtual pro. Let’s call him Andy, as I know an Andy that likes to spend a long time off the front of professional Ironman races alone so the name seems to fit. Let’s imagine his system weight (75kg athlete plus 10kg tri bike) is around 85kg, his cda is around 0.25 (he’s not been in the wind tunnel much but is fairly aero) and he rides an Ironman at 300W. If we head to Kona and simulate this course we can ask ourselves two questions…
How fast does Andy go?
How much energy does he use to go this speed?
Ironman bike splits have two competing priorities, get off the bike as fast as possible but having used as little energy as possible (remember, you still have to run a marathon after this). Let’s head into myWindsock…
Here’s our virtual pro’s bike split when they’re riding alone. A decent ride…
Ok, so Andy’s doing a decent Ironman bike split here but, let’s say for the sake of argument he swims a bit slower and has a mate to ride with. How does the calculation change?
What is the impact of drafting on cda at 12m?
This is a question for Aerosensor or Bodyrocket and hopefully as these devices make their way onto pro bikes for races we will see these numbers solidify slightly (as well as accurate distances thanks to RaceRanger) but some experiments were done and we will head to the forums to find out what the nerds of the internet have found…
Here we found a cda reduction of 0.033 and other members of the forum appear to trust this bloke so we will take his numbers as ‘true’ because it’s all we’ve got. This is clearly not the best experiment in the world but it’s a starting point to work on the impact of drafting and as these numbers become clearer we can change our models.
Changing the cda…
So, Andy’s cda is now 0.25 when he’s on the front and 0.217 when he’s in the wheel of his mate according to some bloke on the internet – we’ll call his mate Ben. Let’s imagine Andy and Ben have agreed to do half the work each – this means that Andy’s average cda will be the average between these two values. We will assume it to be 0.233 for this reason. There’s a whole bunch of caveats at play here – for example the relative importance of cda vs other factors depend on where you are on the course but that’s partially addressed by us using the Kona course which is an out and back along a relatively flat, straight road so if our approximations are valid anywhere it’s here.
Andy is able to ride at the same average power of course as he’s turned up to the race with some given level of fitness and this is the maximum pace he’s decided to ride at while maintaining his ability to run well. Let’s check out the numbers…
Andy riding with his mate saves over 5 minutes!
Ok, so Andy saves over 5 minutes here…
What about the chase pack?
Let’s imagine there’s 4 guys all working together in the chase group trying to get up to the front of the race, to keep things simple we will assume they’re all working together equally (this doesn’t happen) and also we will assume there’s no compound effect of being 3rd in the group – ie the cda reduction the guy at the back receives is the same as the guy in 2nd (also not really true, it’s likely being 3rd wheel is easier than 2nd). This would mean an average cda for each athlete of 0.225. Again, this is an approximation that comes with a host of caveats of course…
The chase pack are even faster, saving 8 minutes on the solo rider out the front and over 2.5 minutes on the pair riding up front.
What does this all mean then?
There’s a couple of implications of these simulations and a couple of other things which need looking into. Primarily though we all need to stop pretending that Ironman bike splits can be judged without context and that non-drafting racing is actually semi-drafting. The impacts of legal drafting in triathlon are real (and we haven’t even begun to touch on the impact of motorbikes). If I had unlimited money, I’d organise the following experiment at some professional race with BodyRocket or Aerosensor (other options are available)…
Measure the cda of each athlete in isolation on the course
Plot the cda(t) for each athlete during the race and normalise against their isolated cda
This would mean that we can (assuming all the athletes are following the laws, which they generally try to do in my opinion) see where a triathlon is ‘easiest’. We will be able to answer the questions of…
How much draft do lead motos give in a race?
How much advantage is given to ‘pack rats’?
Does being a good swimmer actually matter in triathlon?
Information on tactical decisions – for example should Andy sit up and wait for Ben if there’s a swim gap?
It’s clear there’s a lot of interesting work to be done in this aspect and I’m sure the team at myWindsock will look into it in a little more detail one day.
The 2023 Tour de France stage 16 time trial has proven somewhat of a talking point. We have tried to make sense of the numbers, as at first, they seemed interesting. Essentially, we (Tom) vastly underestimated how fast Jonas would go in this TT but, with a couple of fresh assumptions, we can come up with some good (but not alien-like) numbers for the TT performance that the small Dane managed.
Where was the time gained?
During the short time trial, Vingegaard managed to gain 1 minute and 38 seconds over the Slovenian Pogacar. The TT finished with an ascent of the Côte de Domancy (6.05 km, 6.84%) after a rolling course in the earlier part. Notably, Jonas Vingegaard delivered a truly historic performance, outshining even Pogačar on the 22.4 km route by an impressive 1 minute and 38 seconds, while leaving all other competitors behind by a margin of at least 2 minutes and 51 seconds. Our, and others who have also done this, calculations reveal that Vingegaard achieved a remarkable 7.60 W/Kg for a time of 13 minutes and 21 seconds. This would undoubtedly rank among the greatest performances of all time.
Pogacar lost 77s over the course of the climb, which does include a bike change.
This plot shows how the time gap opened up minute by minute, where each data point shows the rider’s time split at each checkpoint. You can see, from the shading, that the time gap opened up much wider between checkpoint 3 and the finish than at any other point. This shows, as 79% of the time lost occurred in this sector.
The climb – Côte de Domancy
This plot, borrowed from Lanterne Rouge, shows the climbing performances of 2023 with the TT climb shown in red, Vingegaard is the red dot that appears furthest up the y-axis where it can be shown to what extent he outperformed all the other riders that day. Vingegaard must have done roughly 7.6W/kg for 13 minutes and 21 seconds up this climb according to these (and ours roughly match) calculations. This climbing performance is the best so far this year in the World Tour by some margin.
A myWindsock model of the performance
The nice thing about doing a simulation after an event has taken place is that you know the answer and your weather data is correct. We can see that Jonas had extremely favourable conditions and managed to fiddle with the settings to achieve a fairly realistic average power and cda based on what we know about the climb and the time gaps.
It was clear that Jonas’ strategy was a simple one, rail descents, ride the flat sections at a reasonable tempo and absolutely whack the last climb as hard as possible.
Inputting the power to weight estimates done as a lap on the last climb and setting the climbing pace to roughly this, we get the above power profile for Jonas’ TT. We would love for Jumbo to publish the data so we can see how our estimate stacks up! myWindsock premium features really come into their own for races like this.
The conditions
Vingegaard clearly had favourable conditions, shown with our Instagram story analysis of wImpact during the course of the day. This plot shows how Wout Van Aert’s time would have changed had he started at different points during the course of the day. This said, it’s only worth a small fraction of the time difference.
The settings we used for Jonas
We asked Ineos Grenadiers’ Dan Bigham about some realistic numbers for CRR and drivetrain resistance and, with a few caveats used the settings below as a rough estimate. In reality, especially with a course such as this one where speeds, gearing and cadence change so drastically these values vary with time but for our purpose today average values over the course of the TT will suffice.
Assuming an extremely optimised aerodynamic position and the ability to hold this almost perfectly coupled with the fact that Vingegaard is a small man these are the cda values we used. The climbing power was set from W/kg estimates and the other values were changed to fit the recorded time. It’s possible he was less aerodynamic and produced more power in the flats or descents. It’s tough to tell of course but these values produce a 30 minute average close to the peak 30 minute power shown by Vingegaard this tour in terms of W/kg.
By reducing rolling resistance to 0.003 and drivetrain losses to 2% we were able to further shave time from Jonas’ performance. The base assumption we have used, implementing myWindsock’s advanced options to do so, is that Vingegaard’s bike was perfectly optimised.
This was an extremely good ride from Vingegaard and he must have had to produce close to his peak average power for that duration in the time trial position. Something that is possible with incredible dedication to time trialing – something we know he has worked on extremely hard. From our analysis, it seems Vingegaard had a perfect day, on a perfect bike with perfect conditions and produced an other-worldly performance on that final climb. Given how aerodynamic it’s possible for a rider of that size to be and how well he descended, it’s possible he came into the base of the final climb relatively fresh. It would be interesting if Jumbo would publish his power files, as seeing how he paced this would certainly help us when predicting future performances.
If you’d like to bring a new level of analysis to your time trial performances, check out myWindsock here.
Today is the rest day of the Tour de France which has left us all sitting around and scratching our heads wondering what to do with ourselves. While I’m sure ITV and Eurosport are producing a glossy magazine show talking about some inspiring backstory surrounding Thibaut Pinot’s goats and Cycling Weekly are writing articles speculating around who feels good and who has shown ‘signs of weakness’ we decided to stay in our wheelhouse and talk about the impending time trial.
This year, the Tour has only one TT and it’s pretty short in terms of distance but that doesn’t make it unimportant. Big gaps in the general classification are possible as the course is a tricky one to pace correctly and we will see slow average speeds (relative to the usual insanity in Grand Tour time trials).
The TT takes place on Stage 16 of the 2023 Tour de France and the course goes from Passy to Combloux. It has 3 climbs, the Côte de la Cascade de Cœur, the Côte de Domancy and the ride up to Combloux to finish. The descent from Passy Chef-Lieu down to Sallanches is technical and the roads are narrow (though have been freshly resurfaced).
The parcours looks relatively flat, we can assure you that it isn’t.
The slower speeds in this climb will be interesting as will the high temperatures often seen in this valley. The teams that have invested in the materials that provide a balance between aerodynamic performance, cooling and also aerodynamic performance at slower speeds (as the fastest fabric at 60kph is not often also the fastest at 30kph) will have a competitive advantage over the riders wearing their standard speed suit. The usual disadvantage will also apply to athletes in the leaders jerseys as these skinsuits are typically slower than the team’s kit sponsor’s model.
The Weather Forecast
Due to Thunderstorms we are expecting very changeable conditions for the riders. Wind Speeds are expected to be predominantly South Westerly, at a Light Breeze of 6mph, with gusts of up to 27mph. The hot temperatures and intermittent showers will add to the complexity of the course.
On dry roads, a significant portion of the descent from Cote de la Cascade de Coeur towards Sallanches, could be taken in aero bars. However, the possibility of rain during the afternoon may cause riders to leave their fine tuned time trial positions more frequently. A wet vs dry descent is likely to be the greatest variable due to start time.
Changeable conditions with predominantly SW Wind.Average Wind SpeedWind Gusts
Expected finish times
The best place to stay apprised of the conditions on race day is via our forecast which you can find here.
Over a course like this it can be difficult to predict finish times as the aerodynamic properties of riders change a lot on long shallow climbs. That said, the TT route is preloaded into myWindsock so we can check out the sort of times we might see based on a few power and weight options.
For this kind of race, we will think of a TT specialist with a cda of 0.2 and 400W to play with, along with a system weight of 75kg then we will play with the settings to see what a rider like Pogacar – the favourite in my eyes – will do in this TT.
Run 1, the ‘average’ tour pro.
A solid benchmark for this first TT is under 37 minutes (weather pending of course). A GC contender will be able to do in the region of 6.1 W/kg for this duration and might have a slightly lower system weight of around 72kg. Let’s see what this does for the overall finishing time…
What might a GC contender do?
So the winner of the stage is likely to be pushing 35 minutes and it may come down to pacing strategy. This course also favours the lighter riders…
The force breakdown plot from myWindsock.
As you can see from our force breakdown, gravity plays a larger role than air resistance on this course – which is why I find myself frustrated when commentators describe it as ‘rolling’. This is a hilly TT that will suit climbers better. Yes the TT specialists will go better in the valley and yes the climbs are short, but the only bigger guy that stands any chance is Wout Van Aert.
What can Wout do? This is a rough estimate of what he’d need to do to be competitive on the stage.
Winning is possible for Wout – as we can see above, but he’ll need to do some monster numbers and it will need a bad day from one of the other main contenders.
Blowing up on the last climb
The last climb is steep and then turns out onto the main road where it flattens out to a low gradient for the final few kilometres. This low gradient section can be deadly though – it’s a big ring climb but it’s uphill all the way to the line. We thought we’d check out what happens if a rider cracks on this final segment by going too hard on the steep section – something we see often even in the pro peloton (think back to the final TT in 2021).
The final climb is tough with a 2.4km section over 10% average gradient.
The final climb is steep and will take riders around 15 minutes from the base to the top. The first 2.5 km is steep and the final km flattens out and goes onto a main road.
What happens if a rider blows up then?
This plot shows what happens to a rider’s time as they drop off their power. On a climb it’s pretty linear, lose 10W and you’re shipping between 10 and 20 seconds on this climb. We will see some changes on the leaderboard between the final intermediate check and the finish with this parcours.
This TT is going to be important for GC but not long enough that it will decide the race. The overall winner is likely to be a GC contender but it could be won by a rouleur on a very good day. If you want to keep up to date on the conditions on the day, follow myWindsock on Instagram where we will be keeping up to date with the fastest conditions and providing live analysis of the time trial.
If you want to have a go yourself at predicting times and winners, or want to add a new level of precision and preparation to your riding – why not use myWindsock and sign up here.