Shownotes

Introduction and why we are nowhere near peak aero 

01:20 -

• I am not good at guessing what surprises people, but there is one claim I hear frequently and disagree with completely, which is that we have reached an era of peak aero where very little is left to be found.
• I think we are still very early in the aero game, there is a great deal still to be gained, and records will continue to drop as fast as they have been dropping over the last few years.
• Of course I work in the field and would be expected to say that, but it is the only reason I continue to do this work.

Where the remaining gains are across disciplines 

02:39 -

• This applies right across the board rather than to triathlon alone. Triathlon has a much larger age group population that is interested in aerodynamics, and probably a more obsessive one, than road cycling, even though there are far more road cyclists overall.
• In gravel, which I started working in last year with a professional rider, the gains are enormous, though a lot of that sits with rolling resistance rather than aerodynamics in the strict sense, so it is better described as performance optimisation of the equipment.
• On the road side there is still plenty to be found in tyres and wheels, and in tactics.
• By tactics I mean questions such as how you optimise a lead-out train at the Tour de France, how long each rider pulls, what position they hold, when the lead-out breaks, and what the optimal position and approach is for the sprinter himself in the final metres.
• There is still a lot to be learned there. The proof that we are not at the limit is that World Tour teams continue to invest heavily in this, and if anything, that investment is accelerating.

Working with teams and with athletes at every level 

05:00 -

• I started Notio in 2016 as a joint venture with Argon 18, who were sponsoring Astana, so I worked with that team until around 2019.
• I did some work with Team Sky during that period, and more recently with NSN, formerly the IPT team. Along the way I worked with Bahrain, helping them optimise their team time trialling.
• That is four teams, plus a lot of contacts and friendships built over the years, which gives me a reasonably good picture of what is going on at world tour level.
• My main focus is supposed to be development of the measurement technology itself. Aerodynamics is the big one, but we also measure road vibration and how riders move around on the bike.
• In practice, I get to test that technology on athletes of all levels. In one recent week I was in Girona testing World Tour riders, and two days after coming back I was testing a 72-year-old masters cyclist preparing to win at Worlds. There is something to be learned from working at every level.

Why road validation matters 

07:27 -

• I believe in using every technology available when it makes financial sense, so I use the wind tunnel, I use velodromes and I use a bit of CFD.
• But we do not race in any of those environments; we race on the road, and at a minimum you have to validate on the road. The differences between these environments are large enough that skipping that step gives you questionable results.
• What we see very often is someone going to the tunnel, getting into a perfect position while watching a small camera feed that motivates them to hold it, producing an incredible CDA, and then never translating any of it into a race on the road.
• That is extremely common. If you are not validating on the road at some point, you are leaving a lot on the table, and finding the right balance between the different tools is a large part of the optimisation work that is still to be done.

Choosing a venue when your budget is not infinite 

09:25 -

• Age groupers have a specific constraint, which is that most do not have unlimited budgets, and given that the tunnel is probably not the first venue to choose.
• Recently I have been involved in what we call aero camps, where a tester we work with books a velodrome and brings in 20 to 25 age groupers over several days.
• That is extremely cost-effective and results-effective, partly because he arrives with 12 helmets, which is not something you will have at your disposal when testing alone at the side of the road.
• It also depends on the athlete. I came into this as an age group athlete who had retired relatively early from a career in tech, so I had time. I am a geek, a data person; I like tinkering, I am patient and I do not mind losing results and trying again over and over.
• Not everybody has that mindset or the willingness to spend that time, and age groupers in particular are busy.
• If you want to get in and out efficiently, an aero camp is a very good route.
• If you have the time and patience and want to learn along the way, learning to self-test is genuinely interesting, and it is becoming less frustrating than it was five or six years ago, though it still has its challenges.

Velodrome testing and what it removes from the problem 

12:12 -

• Aero testing has a lot of inputs.
• The power meter sits at the core of it, because it measures the power applied to the pedals, and everything else is a matter of working out how much went to rolling resistance, how much to altitude change, how much to overcoming acceleration and how much to aerodynamic drag. 
• On top of that come wind, road quality and traffic, all of which add complexity. If you can eliminate some of those, things become simpler.
• In a velodrome, and especially an indoor velodrome, wind disappears, the surface is flat so elevation change disappears, and barometric pressure, temperature and humidity are relatively constant and easy to measure.
• It is a far simpler environment. The main drawback for age groupers is that not everyone knows how to ride a track, and it takes a little time to adapt, though most people who have never ridden one are fine within an hour.
• If you get the chance to go to an aero camp at a velodrome, take it, and make sure a slot is allocated for learning to ride the track.
• Road testing brings all of that complexity back, and worse, because the variables are constantly changing.
• The idea that a steady wind on the way out becomes an equal and opposite wind on the way back does not exist where I live, because wind changes direction and strength second by second.
• That is why venue choice, time of day and protocol discipline matter so much. You test something three times, test something else three times, and once the deviation between runs is small, you return to your first configuration to confirm conditions have not changed.

Power meters and what actually matters 

15:47 -

• Power meters have become far more accurate and consistent than they used to be, so there are very few bad ones out there.
• There is one older single-sided unit that still turns up for tests and makes me cringe, and there are meters that are very consistent but read high.
• What disappoints people about a meter that reads high is that they are delighted by a 300-watt FTP and then dismayed by a 0.27 CDA.
• Reading high is not the end of the world. Your CDA numbers will come out higher than reality, but the relative change between tests is what matters, and the absolute figure is not that important.
• What you are looking for is the improvement from one helmet to another or one position to another, and a consistent meter will do that job.
• I use Favero pedal-based power meters, and they are excellent.
• At World Tour level, I have been in situations where a team was sponsored by a poor power meter brand, and we fitted Faveros discreetly so that we were working with accurate numbers.
• I cannot point to cases where people got badly wrong numbers from a single-sided meter, though it depends on the rider, because a strongly left- or right-dominant rider will get an artificially high or low reading and therefore an artificially high or low CDA.
• Ideally you want something measuring both sides, either crank-based or dual pedals, but I would not refuse to test because I only had a single-sided meter.
• It is not the first thing I would worry about.

Wind, and what virtual elevation lets you see 

20:15 -

• If you want to minimise the effect of wind, and you live in the Netherlands, move.
• Otherwise, test early in the morning. I usually test around half past nine or ten, when the wind is typically very low. Beyond timing, one of the strengths of virtual elevation is that a wind discrepancy shows up in the plots. 
• Robert Chung puts it better than I do, but virtual elevation is as much a diagnostic tool as a measurement tool.
• Virtual elevation tools also let you type in a wind, so if you enter five kilometres an hour of headwind on the way out and five of tailwind on the way back and the curves fall into line with the measured altitude profile, you have a good idea of what happened.
• I use virtual elevation on almost every test I do, even now, with devices that produce standard deviations and confidence figures.
• I take the wind measured by our device, put it into the virtual elevation plot, and check whether the picture makes sense.
• It allows you to handle wind, or at least see it and validate the test, even without a wind measurement system on the bike.

What virtual elevation actually is 

23:05 -

• Virtual elevation takes your power and your speed together with the environmental parameters, principally air density, which is calculated from barometric pressure, temperature and humidity, and those can be pulled from a weather site.
• From that, it calculates what the terrain profile would have looked like for a given CDA. You assume a CDA of, say, 0.26; it produces an elevation profile, and you adjust the number until that profile matches the real terrain. When the two curves line up, you have your CDA.
• Wind shows itself immediately in that process. If a single CDA value lets the outbound leg match the real elevation but not the return leg, something is going on, and it is probably wind.
• It is available in tools such as Golden Cheetah, among others.
• I use it for diagnosis as much as anything. You see a blip where the curves separate, and you can work out what caused it, whether that was a pothole or a car, and in my case very often a logging truck going past mid-run. It lets you visualise that very simply.

What on-bike aero sensors measure 

25:39 -

• The devices on the market differ, but there are things all of them do. They all measure air speed, and wind speed is simply air speed minus ground speed.
• They all measure environmental information such as humidity, temperature and barometric pressure, from which air density is calculated, and they gather everything in one place so it can be dropped into a spreadsheet or an analysis tool.
• Where they differ most is altitude and inclination measurement.
• On a perfectly flat venue that matters less, but on a course with rolling hills you want a device that handles it well, and the ones I have tested are not equal in that respect.
• Our device also measures vibration, which we use for tyre pressure optimisation on a given road and for run comparison.
• One of the challenges in road testing is knowing whether the rider followed the same line on consecutive runs, and vibration patterns tell you that, because riding on the shoulder produces a different signature from riding out in the road.
• The team behind Aerosensor in the UK take a similar approach to rider movement with a small camera that checks whether the position is consistent between runs. We do the same thing differently, using IMUs on the athlete's body, and each approach has its advantages.

Where to start if you want to test on your own 

28:50 -

• I would start with virtual elevation and Chung style testing in Golden Cheetah. You will get good data, and more importantly, you will build the understanding needed to interpret whatever one of these devices eventually tells you.
• Anybody who thinks these devices are black boxes that simply hand you a trustworthy number is being very optimistic. Working through it yourself shows you the challenges first-hand and then shows you how the sensors overcome them.
• That is exactly how I progressed. I had time, I wanted to learn, I read about Golden Cheetah, went out and tried it and had reasonable success.
• Then I heard people were doing it at the velodrome, so I went and tried that, and it was the Alphamantis group who built the first track system, who introduced me to Argon 18, which is where the whole thing started.
• If you have a race next month and you are going to Kona, find a competent aero tester and use them. If this is a long-term interest, learn virtual elevation.
• That matters because in many places there simply is no aero tester within a reasonable distance, since it is still not a common profession.

Building a protocol and choosing a venue 

31:16 -

• Safety comes first, which means finding a road with minimal traffic, both so you are not at risk and so the results are not skewed.
• After that you want as little elevation change as possible, because rolling hills are manageable but real gradient adds complexity, and you want something you can do as an out and back.
• The ideal is an outdoor velodrome, where you simply go round and round with no elevation change and wind from every direction, but most people do not have access to one.
• For an out and back, the tools are increasingly able to clip the turnaround, so you ride from point A to point B, coast, turn, and ride back, and the software removes the transition.
• That protocol also cancels elevation, because whatever you climbed in one direction you descended in the other, leaving a net change of zero.
• You also want somewhere you can park a car with helmets, spare equipment and tools, because you will be changing positions and wrenching things.
• Venue selection is the hardest part.
• In Girona we spent a whole day scouting test venues with the team mechanics driving us around. I like a one and a half kilometre out and back, and what we settled on was slightly rolling. 
• Girona is getting harder, and venues I used three years ago are no longer usable because of traffic.

Run length, frequency and combining testing with training 

34:57 -

• I prefer more short runs to fewer long ones. If you do a five-kilometre outbound leg and catch a short-lived gust, it skews that run, and with only one run you cannot identify the problem.
• With three out-and-backs, where the gust appears in only one, you have two consistent runs to work from and a much better result.
• That said, I also aero test while doing something else. The advantage of five kilometres is that you can run a threshold interval over that distance, which is hard to do over a kilometre and a half.
• I am usually doing three things at once: a workout, an equipment test, and a test to improve the device itself. I am a strong believer in that, because aero testing should not be a once-a-year exercise.
• A lot of the big optimisations will come from testing continuously, watching for deviation from what you believe is your baseline, and then working out the cause, which might be as simple as being sloppy with head position that session.
• A typical age grouper already has to swim, bike and run, so if aero testing can happen during training, it should.

Baseline runs and how often to return to them 

37:21 -

• With experience, and particularly with an aero device that flags issues for you, you can return to baseline less often, though you should still do it.
• I always repeat the baseline test at the very end, and what I am mostly watching for there is temperature.
• As tarmac warms up, rolling resistance changes, so the assumptions you made at seven in the morning do not hold at half past eleven.
• Our device produces a standard deviation for every run, so we know whether a run was good and how much the CDA varied across it.
• If the deviation across a kilometre and a half out and back is small, the test is solid, and two runs may be enough rather than three.
• If you are doing virtual elevation without an aero sensor, you definitely want to return to baseline frequently.

A worked example of a full session 

40:05 -

• With the age group athlete I tested on the Tuesday, I started with a recon out and back so he knew the route, the turnaround point and when to press the buttons on his head unit.
• Then he did two out-and-backs as his baseline, and from there we went straight into a helmet change, testing two helmets and two skin suits.
• I downloaded after every test and checked the standard deviations and the confidence in the data, and nothing looked strange, with a very stable power meter and speed and wind figures that made sense.
• We only returned to his baseline at the very end, and it retested well.
• In sequence, that was recon, then A, A, B, B, C, C, A.
• That would have changed if a configuration had come out dramatically better or worse than predicted, in which case I would have spent time working out why and gone back and forth between the two to find it.

Analysis, and why you want the data behind the number 

42:00 -

• If you are choosing between a device that gives you a number and a device that gives you a number plus all the underlying data, take the second.
• A single CDA figure is the most valuable output, but you need to be able to dig into what sits behind it.
• If a power meter stops transmitting for three or four seconds during a run, the device sees a rider moving quickly on little or no power and hands you an artificially good CDA, so you want software that will show you that. 
• Golden Cheetah lets you see a great deal, including power drops and speed sensor drops.
• What I would avoid is a tool that hides the data, whether that is meant to protect the user from complexity or not, because when things do not add up, you need the detail.
• The less experienced you are, the more you want the summary, but the option to dig has to be there.
• We are also starting to use artificial intelligence for analysis of CDA tests, helping people interpret their own results. Handed a test, it will quickly identify power drops, wheel circumference discrepancies and other issues you would otherwise have to check manually.
• For a given configuration, I look first at whether the numbers make sense, using what we call a confidence score, which is mostly built on standard deviation.
• With the athlete on Tuesday, the CDA started around 0.205 and finished at about 0.192, and what I was checking was whether it stayed reasonably constant through each test. I confirm each configuration before sending the rider out for the next one, and if there is an anomaly, I have him repeat it.
• That is one of the real difficulties of testing alone, because you finish, go home and only then discover something odd about the B runs, whereas I can analyse at the roadside while the rider is out doing the C runs.
• I also try not to tell the athlete their CDA during the session, because knowing it can skew what follows.
• At the end, I generate a report covering how stable the tests were, and I ask for race distance and target power, in his case 40 kilometres, so I can convert every configuration into a time saving. 
• Very often it is not a case of this is faster, so I take it. It is faster with compromises, and the rider has to decide whether to accept them.
• If it is ten watts faster but costs you power, or two watts faster but you cannot see cars, that is a decision only the rider can make. Most of the time there is a decision of that kind to be made.

Common mistakes 

48:16 -

• The biggest mistake is trying to do too much in too short a time, and we see it in every venue.
• It is worst in the wind tunnel, where the cost pushes people to cram tests into an hour or ninety minutes, and the result is garbage. 
• If they recognise it as garbage, that is bad but survivable. The real problem is when they do not recognise it and make decisions on it.
• Doing the same test repeatedly and getting consistent results through the day is a genuinely valuable step in the learning process.

The hierarchy of what to test 

49:37 -

• It depends heavily on testing history.
• A typical age group athlete who has never been tested will often find 20 aero watts, simply because there is so much opportunity. In order of size, position is first. 
• Helmets and skin suits are second and tend to be close to each other. Tyres and rolling resistance sit nearer to the helmets than to the third group, which is hydration and related items.

Body position 

50:55 -

• The first question is what you are racing, because a 15 kilometre time trial and a 180 kilometre Ironman are different problems.
• For an Ironman, comfort matters, not so that the ride is enjoyable but so that you can hold the position for the full distance rather than coming out of aero after 40 kilometres because your body cannot sustain it.
• We will still test aggressive positions with an Ironman athlete, so they understand how much they are actually gaining, and can then decide whether two or three watts is worth it.
• Head position has the largest single effect.
• I always film the athlete to see what the head is doing, and how it might be improved, and the picture is completely different for someone sitting upright with their head high than for someone already visually optimised. 
• From there we start trying things to see what works for that individual.
• There is a misunderstanding that runs through a lot of this.
• When the fashion was for steeply inclined aero bars with the hands next to the nose, people assumed the change itself was fast, blocking or redirecting air, and built theories around it. 
• What actually happens is that the change affects other parts of the body.
• When the arms come up it often becomes easier to drop the head, whereas very low arms tend to lock the neck in place so the head cannot come down. 
• The same applied to the very long Superman positions. If you went longer but deliberately kept the head in the same place, there was no improvement, but for some people going long makes it easier to drop the head.
• So I am usually looking for changes that unlock another change, and head position is very often the one underneath.
• Getting a rider steeper, further forward on the saddle rather than back, opens the hip and tends to feel more comfortable.
• I am not a fitter and do not pretend to know fit, and my best aero tests are the ones where I am working with a competent fitter, but opening hip angles, whether through position or shorter cranks, helps power production. 
• We frequently see a position that feels more powerful but tests two watts slower, and plenty of riders will take the more comfortable option.
• If you are seeing a fitter and you know you will be aero testing afterwards, ask the fitter for the ranges he would be willing to try you in, how much more reach and how much steeper, and I will test that range and go back to him so we can work out the optimum together.
• This is iterative rather than one and done. Riders often think they will hold a new position, then go and train and discover it is not for them and revert to where they were, so build that into the process.

Balancing aerodynamics against power production 

56:20 -

• Making that trade-off quantitatively is very hard. I was part of a small group using Moxy sensors to measure the effort required to hold a position while I measured the CDA effect, and it was not great.
• It is much better to leave the aero test knowing that position A costs you three watts, then go and do two three hour rides in it and decide whether it is still worth three watts.
• Trying to resolve both questions inside a short test is very difficult.
• The same logic applies on the road bike. In Girona we tested riders in the drops, on the hoods, and on the hoods with the elbows tucked in tight, and the differences were large.
• So the guidance to the rider becomes tactical. If you are in a three or four-man breakaway and you are on the front, do this, and when you are not pulling, do that; and since this position is much more aero, work on it in the gym and on training rides.
• Dan Bigham used to say the body can adapt to almost any position within reason. If a change is worth 20 watts, put the time in.
• If it is worth two watts, question how much time it deserves.
• The point of an aero test is not to come back with absolute numbers, but with data you can make decisions on.

Helmets 

58:33 -

• At the 2025 aero camp we had around 21 people and something like 12 to 14 helmets, with everyone testing three or four, which produces useful statistics.
• Across that group, the MET Drone 2 and the newest Rudy Project helmet consistently came out first and second. 
• Roughly a third of the time the MET was fastest and roughly a third of the time the Rudy was, and when one won the other was second, with everyone else picking up the occasional win.
• Triathletes seem to have noticed, because the Rudy is selling exceptionally well, and it is a very good helmet to consider if you are not going to test.
• We fit a temperature sensor inside the helmet, added for four dollars to a sensor that was already measuring head position, so that we can measure how warm it actually gets in there.
• Helmet choice has an aerodynamic component, but visibility and heat both have to be weighed as well, and neither can be judged without wearing the helmet.
• Every time I run a helmet test, I ask for immediate feedback on heat and visibility, because otherwise people fixate on the CDA number and simply have not noticed, unless they were dripping with sweat.
• Helmets are genuinely difficult, but the trends are well known and following them will not put you far wrong.
• I used to wear an aero helmet for all my day-to-day testing, a Giro Aerohead 2, and I have since switched to a road helmet, a Specialized Evade 3.
• The two are close in speed, and the road helmet is obviously much cooler. The MET Drone is quite a bit faster than either.
• What I would test is a genuine aero helmet worn with sunglasses against a road helmet, because you gain a lot of cooling that way.
• People are complaining that the Rudy is very warm behind the visor, and I suspect a Rudy with sunglasses would be faster than the MET, though that is a guess.

Clothing 

1:03:47 -

• The aero camp is run by someone who has a shop just outside Amsterdam and his own line of suits, so participants would test one of his suits as well, partly to gather data on them. The typical person arriving at an aero camp brings a very high-end suit.
• The trend we saw was that the high-end suits usually tested fastest, but not by much- nothing like 12 watts.
• A lower-cost suit that fits you well costs perhaps three or four watts, with three being about average. 
• Fit was everything, and both too tight and too baggy are problems. Clothing makes a difference, but I am not convinced a very high-end suit is the best watts per dollar.
• The athlete I tested on Tuesday had flown from Canada to the UK, tested at Silverstone and had a custom Vorteq suit made, and it tested only marginally faster than his off-the-shelf Nopinz suit.

Hydration and bottle placement 

1:05:55 -

• Hydration has been the most hyped area of the last two or three years.
• Much of that came from the period when riders were pushing bottles down their jerseys and claiming 30 watt savings, which was nonsense, and when that was banned, people decided they had to achieve the equivalent between the arms. 
• They started positioning bottles to block air from reaching the chest and in many cases, went slower.
• At the velodrome, we would ask why a bottle was in a particular position, get told it was because a particular professional does it that way, move it, and find it faster, then move it again and find it faster still.
• On the front, unless you are going to test, I would avoid mounting it too high.
• A lot of people lean on their bottles, and we found more riders were faster not leaning on them than were fastest leaning.
• On the rear, Cervelo published data on behind the saddle and between the arm bottles back around 2012, and I find that more accurate than what circulates on social media now.
• A bottle on the back can be slightly beneficial, will hopefully be neutral, and if it costs a couple of watts, that is not significant.
• Claims of 12 watts from a rear bottle are absolute nonsense. They do give me something to do, because I read them, 3D print something, go and test it, and then confirm what I suspected.
• My advice is to sort out your fluids, electrolytes and calories first. If you cannot test, put the bottle on the front, lower rather than higher, and more towards the middle of the arms.
• If you do take the time to test, moving it around and trying one or two bottles will find you minor savings, but you are not going to find 20 watts there.
• I find it interesting that the new Canyon almost prevents you from setting the bottle up in that very high leaning position that aftermarket setups allowed.
• Aftermarket parts will presumably appear eventually, but Canyon appear to have made a deliberate effort to sell a bike that is optimised and secure straight from the factory. 
• My question is not whether they are right, but whether they considered that position and what they found.
• There is still a great deal of misinformation around bottle placement, front and rear.

Tyre choice and durability 

1:10:55 -

• Tyre pressure is extremely important, and choosing the right tyre and then running it at the right pressure go together.
• Choosing the tyre is hard because you want it as fast as possible and as durable as possible, and while speed is easy for me to measure, durability is not.
• With a team, I can put a promising tyre on several riders' training bikes and gather statistics on whether it survives and how often it punctures.
• With a triathlete, I will tell them to fit it and train on it all summer, so that if they have no flats, they arrive at Kona with confidence in it.
• The Continental Archetype is an example. It is a very, very fast tyre, but my impression is that it punctures. If a decision were made purely on my road speed numbers, it would be selected, and the flat rate would go up.
• On performance, there is a rolling resistance component and an aero component, and the rolling resistance component is much larger.
• If you cannot test, bicyclerollingresistance.com is a reasonable starting point.
• You should not assume their absolute number applies to the road, but if you take the difference between tyre A and tyre B on their rollers, that delta will usually carry across to the road. There are exceptions, but most of the time it holds.
• The watt savings they publish are a different matter. Those figures are per tyre, and they are generated at a relatively low mass and a relatively low speed, I believe below 30 kilometres an hour.
• What you should do is take the rolling resistance coefficient, or the difference between two coefficients, multiply by mass, by 9.81, and by your speed.
• That gives you the watts you will save. Reverse engineer the wattage rather than using theirs, which is probably conservative.

Tyre width 

1:14:37 -

• There are effectively two worlds here, good roads and bad roads. On reasonable to good roads, wider is very rarely faster.
• The idea that a wider tyre at lower pressure with a larger contact patch is faster is a myth we simply do not see borne out on the road. 
• On rough roads the wider tyre does have its advantages, and there is a general principle that being a few psi or a few tenths of a bar too low is much better than being too high, because the penalty for too high is greater.
• So err on the low side and drop the pressure a little on very rough roads.
• The pendulum has swung back somewhat.
• A year and a half ago the expectation was that we would all be racing triathlons on 40 millimetre tyres within two years, much of which was tied up with the hookless debate. 
• Pogačar was riding 25s in the time trials at the Tour de France this year, which surprised me. On reasonable roads you are not going wrong with a 28, and a 30 is not a big mistake, but I would question a 32 or a 25.
• Tyre pressures were trending a little too low as well. There were articles suggesting three bar rather than five, which was nonsense for triathlon and time trialling.
• What we do not understand is the effect of vibration on fatigue. I do not take it into account and I probably should.
• We measure vibration, and I rode a course recently that I had not ridden in a year, went back to the data from the previous year, and the deterioration people had been complaining about was clearly there. 
• That happens to be part of the Ironman Canada course, and I put a friend on 30s for that race last year.
• It varies enormously. Ironman Xiamen in China had perfect roads. Mont-Tremblant is very good. Most Ironman courses sit on the better side rather than the worse.
• Haines City in Florida was terrible when I did it, and Ottawa this year was pretty bad. Cartagena in Colombia had a section that felt like riding a road that had been bombed.

Finding optimal tyre pressure 

1:19:25 -

• The method mirrors virtual elevation analysis, where you adjust CDA until the curves match.
• We start with a tyre whose rolling resistance is well characterised, calculate a CDA, then hold that CDA fixed and vary the pressure, testing four bar, four and a half, five, five and a half, while also measuring vibration so we can characterise the road.
• We then find which rolling resistance value, at that fixed CDA, produces the best virtual elevation fit.
• What we have found is that optimal pressure depends on the tyre itself.
• A Continental GP5000 TT might be optimal at 4.8 bar while a Corsa Pro Speed is optimal at 5.1 bar, which suggests the compounds do make a difference.
• If you are not going to test, the Silca calculator will give you a reasonable number.

What the numbers actually look like 

1:21:20 -

• We tested World Tour riders on relatively good roads in Girona, on 28, 30 and 32 millimetre tyres and everything between four and six bar.
• They weighed around 70 kilos, one at 68 and one at 71, and their optimal pressure came out at 5.5 bar. 
• Dropping to five bar cost them about three watts. I would guess the true optimum was above five rather than below, probably 5.6 or 5.7, and going up to six was not much of a penalty.
• That road was not perfect, with a filled crack roughly every ten metres, and I filmed it because I knew people would argue.
• What was surprising was how little the optimal pressure varied with tyre width on those good roads. It was not a case of 5.5 bar on the 28 and five bar on the 30; the numbers were remarkably close.
• On bad roads it is a completely different story, and weight changes the picture too.
• Chelsea Sodaro, who I work with, is around 53 kilos and can get away with much lower pressure, but even she was running five bar on the roads we tested in Colorado. 
• On one road that was genuinely very bad, the WorldTour riders were still optimal at around 4.4 bar, which is not that low.
• People have come to believe lower pressures are better.
• That is true on bad roads, but the trend went too far. It started with an article a couple of years ago, I believe, based on Zipp data from what they called a rolling road test environment, which had four bar as optimal. 
• I looked at that and thought it could not be right, and it was one of the things that pushed me into doing a lot of tyre testing.
• Put a World Tour rider on seven bar today, and they will not be happy with you, but four is as extreme in the other direction. Nobody is riding above six, or should be.
• Testing tyre pressure is also demanding on the rider. One World Tour rider, Riley Sheehan, did over 100 kilometres of rolling resistance tests with me, and his coach called afterwards to ask what I had done to him, because he was racing two days later and nobody had mentioned it.

The aero side of tyres 

1:28:08 -

• On the road it is extremely difficult to separate the aero component from the rolling resistance component.
• There are protocols for it, but it is hard, so what I usually see on the road is the combined benefit. That is where the wind tunnel becomes useful, and we often verify there.
• When the Continental Aero 111 came out as one of the first publicly promoted aero tyres, I tested it on the road for about a month and could see no difference against a GP5000 S.
• The reason was that I was never getting high yaw, and side wind is where those tyres shine.
• In the tunnel, we saw they were much faster at 15 degrees of yaw, but when we calculated the probability of seeing 15 degrees of yaw in the real world, it was extremely low, with Kona a possible exception. 
• Since they matched a GP5000 S for rolling resistance, which is slower than a GP5000 TT, the aero benefit did not outweigh the rolling resistance cost, so in my opinion their appeal was very limited.
• Pirelli have a new tyre I have been testing this summer, which I have not taken to the tunnel.
• Bicyclerollingresistance.com rates it slightly slower than the GP5000 TT, but on the road I am seeing it on par with the GP5000 TT, so my working assumption is that there is a small aero benefit.
• I do think the aero benefits of tyres are very small in general. What I like about that tyre is that it appears to have good puncture resistance.
• I do not trust myself to mount a new tyre the day before a race, and I do not believe a tyre with 500 kilometres on it has a higher chance of puncturing, so I would rather race something I have been riding for at least a couple of weeks.
• I also do not think rolling resistance degrades much with use.
• At the start of the year, I repeated tests from the previous year with two sets of wheels carrying different tyres, one set ridden a great deal in between and one barely ridden at all, and I reproduced the same delta.

Rim width and the rule of 105 

1:32:22 -

• There are two people whose opinions I trust here: Josh Poertner at Silca and Jean-Paul Ballard at Swiss Side. Josh swears by the rule of 105, where the widest point of the rim should be 105 per cent of the width of the inflated tyre, so that the tyre is slightly narrower than the rim and you avoid a pronounced bulge.
• Jean-Paul does not really see that rule in his data.
• I believe it to an extent, because in the tunnel we typically see about a watt to a watt and a half going from a 28 to a 30 millimetre tyre, so there is an effect from being too wide. Would I obsess about it?
• No. Following it is not a bad thing, but if you are at 103 or 106, I would not lose sleep over it.

What working with world tour riders is actually like 

1:35:25 -

• There is a great deal we did not get to, including gravel, where the performance impact of tyre selection is enormous, and aerodynamics matters comparatively little, and the role of artificial intelligence, and team time trial optimisation.
• What I find most interesting about World Tour riders is how much they geek out on this.
• They get excited, they arrive with ideas and with tests they want to try, and they are as influenced by social media, hype and trends as any age grouper. 
• People would be surprised how much of my time is spent anti-hyping the hype.
• Belief affects results directly, because a rider who wants a suit to be faster will unconsciously hold a better position in it, which is exactly why we use cameras and body sensors. 
• Triathlon and world tour cycling are opposite worlds in how formally and scientifically they approach this. Triathletes are generally not formal about it, with occasional exceptions.

The 12 to 20 metre draft zone 

1:38:20 -

• Those charts published without a y-axis were not released by me. I think they were well intended, and the people publishing them did not anticipate that geeks would be reading them.
• What they were trying to show is a phenomenon I call stacking.
• At 12 metres, and using deliberately conservative numbers, the first rider behind gets perhaps a five-watt benefit, the second 12 watts, the third 20 watts, and by the fifth rider it is enormous.
• That is why you would see those packs in Kona, and why the tenth rider is in an entirely different situation from the first.
• The results were striking enough that I called Magnus Ditlev's aerodynamicist, a friend of mine, to ask whether it matched his experience, and Magnus confirmed it publicly himself. He described pulling on the front, then moving eight or nine positions back, sitting up, and riding at 50 watts less.
• At 20 metres we did not see that stacking effect. Where the first rider might get three watts, the second rider also got about three watts, and it did not compound.
• That is what the charts were meant to illustrate, but it came across as though data was being hidden, which was not the intention at all.
• I have said several times that it would be worth explaining this properly on a podcast, because a single chart will always be perceived a certain way, whereas a conversation with a knowledgeable host produces an accurate picture.
• The whole 12 to 20 metre project was a really interesting experience, and although I was a little sceptical at the outset, I thought it was very well done, and the result was good.

Final recommendations and remote testing support 

1:42:20 -

• Be realistic about what you will find and how quickly you will progress. If you are starting from scratch, find somebody competent to help you, because you will gain far more that way.
• And treat it as a learning journey rather than a hunt for 20 watts, which will make the whole thing much more enjoyable.
• It depends on the athlete, and if someone struggles to calibrate a power meter, the chances of success drop, but most people could do a great deal with remote guidance, particularly on the analysis side.
• We built it into the software because we believe in it.

LINKS & RESOURCES

• Related episodes:
  • Tire pressure, aerodynamics, and smart equipment upgrades with Josh Poertner | EP#235
  • Dan Bigham: aerodynamic testing, equipment, and making your bike fast | EP#229
  • The Hour record and aerodynamics with Dan Bigham | EP#368
  • Bernardo Gonçalves | EP#397
  • The latest on aero sensors and aero testing with Michael Liberzon | EP#339
  • Aerodynamic testing in the field with Michael Liberzon | EP#294
• Resources and tools mentioned:
  • GoldenCheetah - free software for virtual elevation (Chung method) testing.
  • Bicycle Rolling Resistance - the rolling resistance database Marc uses as a starting point.
  • Silca tyre pressure calculator