The Tire Pressure Revolution

The Tire Pressure Revolution

Of all our research on tires, the most revolutionary finding is this: Tire pressure has almost no effect on a tire’s speed. We did not believe it at first, either, so we’ve tested it numerous times. It’s been confirmed numerous times, with different methodologies.
The real revolution is not how you use your pump… What has totally changed our riding are the wide, supple tires, which only work because of this new insight.
First, let’s look at the data. Here is one experiment: We ran three different 25 mm tires (a supple clincher, a supple tubular and a harsher-riding clincher) at pressures from 4.5 and 9 bar (65 and 130 psi). These tests were done on very smooth asphalt (above), a surface where high pressures should offer the greatest advantages. The graphs below show the power required to ride the bike (above) with the tires at a constant speed of 27.8 km/h, but with different tire pressures.
There is no relationship between tire pressure and performance in the tested range. (Lower and higher pressures are unsafe to ride.) The graph above shows some variation in power output (lower is better), but there is no trend. The CX tubular rolls fastest at 5.5 bar, the CX clincher is a little faster at 6 bar, while the Rubino is fastest at 9 bar, but almost as fast at 6.5 bar.
Take-home message: Don’t stress about tire pressure!
This finding has revolutionized our understanding of tires. In the past, we all thought that higher tire pressures made tires roll faster. And that presented a problem for wide tires: A wider tire puts greater loads on the casing than a narrow one. To compensate, you have two choices:

  1. Beef up the casing, which makes the tire less supple and slower.
  2. Lower the pressure, which we thought made the tire slower.

No matter which route you took, then-available science predicted that your wider tire would be slower. It was a catch-22, and for the best performance, you stuck with narrow tires, where you could have a supple casing and high pressure at the same time.
But after realizing that tire pressure doesn’t matter for performance, we were able to explore new possibilities. If lowering the pressure does not make tires slower, you can make supple, wide tires. You run them at lower pressures, and you don’t give up any performance on smooth roads. On rough roads, you gain speed, because the tire (and you) bounce less. And on all roads, you are more comfortable. Instead of a catch-22, you have a win-win-win situation.
It’s this research that has led professional racers to adopt wider tires. They are up to 25 mm now. (Wider ones won’t fit on their bikes!) For the rest of us, there is no reason not to go wider. I now ride 42 mm tires at 3 bar (43 psi), knowing that they roll as fast as a 25 mm tire at 6 bar (85 psi) – or 9 bar (130 psi), for that matter.
To get the most benefit out of these lower pressures, you need supple tires. A stiff sidewall takes more energy to flex, so the tire will be slower. And since the sidewall is stiffer, it also will be less comfortable. You could call it a “lose-lose” situation.
Professional racers have known this all along: As much as their equipment has changed over time, they’ve always ridden supple tires. They usually ride hand-made tubulars (above), but for the rest of us, supple, wide clincher tires now make it possible to enjoy the ride and speed of supple tires on any road.

Share this post

Comments (124)

  • Peter

    Great research Jan (as always). So that leaves us with two reasons (myths?) why narrow tires are faster:
    1) they climb faster because they are (a little) lighter
    2) they require less power (esp. at higher speeds) because of less aerodynamic drag
    I would be quite interested in learning about the effect of tire width on air resistance. Negligible at 20kph, major component at 40kph? Purely academic interest, on my own power I only manage 30kph at most 🙂

    January 26, 2015 at 5:15 am
    • Jan Heine, Editor, Bicycle Quarterly

      The faster climbing is true, but too small to measure. If it was a major factor, the climbers in the pro peloton all would ride smaller wheels, which are lighter.
      Aerodynamics: We tested that, and found that the difference between 25 mm and 32 mm tires was too small to measure reliably in the wind tunnel. In fact, with a rider on board, I am not even sure that wider tires are less aerodynamic. It’s conceivable that they act as wind deflectors that smooth the airflow around the rider’s spinning legs – sort of like the deflectors you see on the cabs of trucks to smooth the airflow over the trailer.
      The studies and advertising claims that show narrow tires being more aero all measure the front wheel by itself. In that case, a narrow tire will have a smaller frontal area and thus be more aerodynamic, but the result is meaningless. It’s like testing the semi-truck without the trailer.

      January 26, 2015 at 5:47 am
      • Brandon

        “The faster climbing is true, but too small to measure. If it was a major factor, the climbers in the pro peloton all would ride smaller wheels, which are lighter.”
        Except they can’t because UCI rules dictate wheel sizes.
        “Aerodynamics: We tested that, and found that the difference between 25 mm and 32 mm tires was too small to measure reliably in the wind tunnel. In fact, with a rider on board, I am not even sure that wider tires are less aerodynamic. It’s conceivable that they act as wind deflectors that smooth the airflow around the rider’s spinning legs – sort of like the deflectors you see on the cabs of trucks to smooth the airflow over the trailer.”
        Do you have any data on this to share? With all of the hullabaloo down to being able to measure the difference even in spoke count, this is tough to swallow.

        January 27, 2015 at 12:58 pm
        • Jan Heine, Editor, Bicycle Quarterly

          Except they can’t because UCI rules dictate wheel sizes.

          The UCI requires wheels larger than 55 cm in diameter. Current wheels are 65 cm. So there is plenty of room for smaller wheels.

          Do you have any data on this to share?

          Yes. We tested this in the wind tunnel of the University of Washington. More details are here, and the results were published in Bicycle Quarterly Vol. 6, No. 1.
          If you test a wheel alone, you can determine even tiny changes. Once you put the rider on the bike, it all gets lost because it’s so small. Even a full set of aero wheels supposedly decreases the wind resistance of the bike-and-rider by only 2-3%.

          January 27, 2015 at 3:02 pm
  • Gert

    Thank You
    Very interesting results I would say. Is it weight dependant? Is it front and rear same pressure?
    Looking at the graph, to me there seems a bad area around 7-8bars, where the tires are better below or above.
    In these “deflagate” times it is nice to know not to make a great deal out it

    January 26, 2015 at 5:21 am
    • Jan Heine, Editor, Bicycle Quarterly

      We adjusted the pressure of the front wheel, because it carries less weight. Of course, optimum tire pressure does depend on rider/bike weight, but since pressure doesn’t matter much, it’s not that important.
      You are right, the “moderately high” pressure of about 7-8 bars (100-110 psi) that most riders use has the highest resistance. This actually was statistically significant, so it’s not just noise in the data. But even that effect is small, and it probably depends on the road surface. In our roll-down tests on rougher roads, we didn’t see this.

      January 26, 2015 at 5:51 am
      • Skyler Des Roches

        You only varied tire pressure in the front wheel for this test? Seems like it would be worth testing this when changing both front and back tires, to try to amplify any trend that might not be significant in the pressure range you’re testing at. The fact that the environment for the test can’t be perfectly controlled (breeze, rider corrections to maintain balance, etc.) and that the rider’s weight may only be 30% on the front wheel (depending on frame geometry), makes me wonder if there is indeed a trend that could be picked up with 100% of the rider’s weight suspended by any given pressure, and with much larger sample sizes to weed out small variations in other variables.

        January 29, 2015 at 5:23 pm
        • Jan Heine, Editor, Bicycle Quarterly

          No, we varied the pressure on both wheels. The front tire had a lower pressure, since it carries less weight. We kept the tire drop the same on both wheels for each run.
          The statistical analysis shows that we did pick up differences in tires (for the runs where we tested different tires) and not noise, showing that our methodology is sensitive enough to pick up small differences between tires.

          January 29, 2015 at 7:36 pm
  • azorch

    I’m going to risk preaching from the choir to echo everything you say here. (1) I grew up in the camp that said narrow/high pressure simply HAD to be better/faster/more efficient; I disbelieved literally everything I read to the contrary. (2) For me, a dedicated “non-fast” rider, the change to wider/more supple/lower pressure resulted in a subtle – but perceptible – change in ride quality and overall riding efficiency. (3) “Subtle,” indeed – until I get back on a bike with 700 x 23 tires, as I did a few days ago. I’d made the philosophical turnabout in tires a few years ago and had not done much riding on “skinnies” in that time. The difference was jarringly, teeth-chatteringly apparent and I was shocked at how much harder my arms seemed to be working. I’m no racer, or racer “wannabe,” so the benefit to me is one of greater riding efficiency which equates to the important thing for me: more quality time in the saddle.

    January 26, 2015 at 5:36 am
  • CultFit

    Very informative – Thank you!

    January 26, 2015 at 5:38 am
  • ORiordan

    So if someone is arguing the point with me, is it valid to say that for some people, higher pressures/narrower tyres may “feel” faster because they associate a harsher ride with speed? In other words, it is perception rather than reality and the perception has probably never been challenged by comparative testing.

    January 26, 2015 at 6:19 am
    • Jan Heine, Editor, Bicycle Quarterly

      Yes, here is how it works: The faster you go, the higher the frequencies at which your bike vibrates. (The surface irregularities pass under your wheels faster.) However, higher tire pressure also increases the frequency of the vibrations. So the bike feels faster, even if you go at the same speed. It’s a powerful placebo effect, and I believe it is the reason why narrow tires and higher pressures have become popular time and again in the history of cycling.
      I’ve been tricked myself, believing we were going much faster when I once rode my bike with higher tire pressure. I had forgotten to deflate the tires after a test. My riding partner Mark laughed (he has a computer on his bike) and said: “We are going the speed we usually do around here.” Yet to me, with 120 psi in the tires instead of 80 (this was on 32 mm tires), it felt way faster.
      Documenting and spreading the results of our research hopefully will make future generations immune to this error.

      January 26, 2015 at 6:27 am
      • generaledefoutre

        “higher tire pressure also increases the frequency of the vibrations” each vibration is an information as if you were litteraly reading the road with your bike, so when you’re in a narrow tires / high pressure / fast pace situation you’re receiving much more information than in fat tire / low psi one. That speed feeling is in fact some kind of stimulus overdose.

        January 26, 2015 at 7:06 am
      • RickH

        Agree. I did a similar thing but tried from extreme low pressure to maximum. Initial feeling was “woo! this is fast”. Looking at the speedo it was maybe .2 kph faster. What! Roll down distance was barely 2 metres over 350 metres distance. No Big deal except comfort.
        As for wind resistance. Pfft! As soon as you’re on the bike you can throw all the test data out the window and that is going to vary depending on the weather and the clothes you’re going to wear.
        Unless you’re racing a time trial or a personal best in a Gran Fondo focusing on small details are for those who treat their riding as a competition.
        I always quietly laugh when I hear “wassitweigh”? Less than 2 bidons a rack a bag a jacket and a couple of bagels along with some cash a credit card and a phone or gps device.

        January 27, 2015 at 7:52 pm
  • Ty

    Great stuff as always Jan, thanks!
    I believe you have also pointed out in the past that lower pressures equal less flats, in that a sharp object is more likely to penetrate a “hard” tire with no give than a softer tire which may tend to give enough to prevent penetration.
    I certainly seem to have far less flats since I went the low pressure route, even on my Extra Leger Compass 32’s, and I have had them for over a year. I only have had one flat on those tires, and that was the first ride out and probably a fluke from a very thin wire which probably would have caused a flat no matter what pressure I was runnign.
    I would be interested to know if your new research still supports the notion that low pressure tends to mean less flats.

    January 26, 2015 at 7:27 am
    • Jan Heine, Editor, Bicycle Quarterly

      Quantitative research into flat tires is difficult, since they are both rare and random. However, there is a lot of data from riders who report far fewer flats after switching to wider tires at lower pressures. So it’s safe to say that lower pressures equal fewer flats. It’s certainly been my experience.

      January 26, 2015 at 7:59 am
      • David Morgan

        I strongly agree with the lowered flats one wider tires/lower psi. After changing my commuter from a racer to an vintage ride with 27″x 1 1/4″ HOOKLESS, (yes, hookless rims where one dares not inflate tires over 70-75 psi) and running them at 55 psi front by 67 psi rear (I got those #s from calculations in a previous BQ article) I found that my commute was a lot more enjoyable, not at all slower (regardless of bike there continued the 2-3 minute per ride variation due to traffic flow and lights, and of course, human variation) but the huge benefit was that I got no more flats on this horrid industrial/ghetto residential area with broken glass aplenty.

        January 26, 2015 at 2:54 pm
  • Jason Marshall (@jmarshall312)

    Taking it one step further I would argue that wider tires are indeed ‘faster’ for long distance cycling like a 1200k brevet or a multi-day tour. The harshness of high-pressure, narrow tires take a toll on one’s body eventually causing a degradation in performance. The wear on one’s body will likely lead to longer breaks at controls and the need for more sleep/recovery time.

    January 26, 2015 at 7:44 am
    • Jan Heine, Editor, Bicycle Quarterly

      You are absolutely right, but that factor is harder to quantify. We can measure power requirements in Watts, but fatigue is harder to measure.

      January 26, 2015 at 7:59 am
      • Andy

        Towards the end of the Cascade 1240 last year, my hands were not very happy on a long chipseal road, and I noticed that the others I was riding with seemed to fare much better on their wide 650b tires (I had 700×32). But I also noticed that 650b is mostly ridden by people with significant rando experience, probably because few people start distance riding on bikes that aren’t readily available in stores. I’d imagine that it’s both experience and tire choice that had them feeling better. Similarly, experienced riders can reduce flats because they tend to stay out of debris-filled shoulders or are just better at scanning the road for glass and debris to avoid. Just another way of saying that these factors are hard to quantify on one variable alone, and while tire size likely helps, so do many other factors.

        January 26, 2015 at 10:06 am
        • Jan Heine, Editor, Bicycle Quarterly

          Isolating factors is always the challenge in scientific research. Among my friends, some get twice as many flats as others, even though we ride on the same road and with the same tires. Some of it is weight difference, some of it lane positioning/avoiding debris, and some of it is luck.

          January 26, 2015 at 10:11 am
  • Luis Bernhardt

    Another “myth” you might want to look into is our proclivity for putting less air in the front tire just because there’s less static weight on it. Yes, it’s true that my front tire (Vredestein 23mm) will get about 7,000 km of useful life, while the rear gets 3,500 to 5,000, so there’s obviously less weight on it. And yes, you might want to put as little pressure as you can get away with in each tire (if less pressure is good, then can even less pressure be better?). But I like to inflate my tires for worst case. Under braking, the front wheel gets more of the weight than the rear. Same thing when you’re out of the saddle climbing. So, same pressure front and rear.
    I think the biggest reason for racers not using wider tires is due to wider performance tires not being generally available until recently. The typical wider tire back then had stiff, thick sidewalls and very thick tread and were way heavier than even heavy 23mm clinchers. Racers also wanted tires with profiles similar to the tubulars they sometimes used in criteriums, or on the track.
    I will keep using 23mm tires just because I like the rigid feel and the way the bike vibrates on the road. The problem with using less pressure is that when you weigh 175 lbs, you pretty well have to inflate to 120 psi to avoid pinch flats, which I have gotten at 110 psi. I also like sidepull brakes (or I should say, I just hate the look of cantilevers) and a 68mm bottom bracket width, and it’s hard to fit anything bigger than 25mm into a released sidepull or between the chainstays of a standard road bike. Greater use of disc brakes will alleviate one of these problems, while bringing in other issues. But I will make sure that my next bike will accommodate maybe 28’s. Seven does a good job of flaring out the stays coming off a 68mm bb,

    January 26, 2015 at 8:44 am
  • Arsenalfan

    I wish I’d known about this before my trip from London to Istanbul. I weigh 138, bike 25, gear 26, totaling 189. I ran the 700×35 at maximum recommended pressure of 100. I felt every single bump going up my spine for the next 2,400 miles. Ugh.

    January 26, 2015 at 8:54 am
    • Jan Heine, Editor, Bicycle Quarterly

      I wish I had known this when I toured the Baltic coast of East Germany. I was running 28 mm tires on a fully-loaded touring bike, inflated to 110 psi or so. And most of the roads were still paved with huge cobblestones…

      January 26, 2015 at 9:04 am
  • Max Beach

    Great article. Currently, what is your thinking as to the best tire pressure chart? I have been using this one (from BQ Volume 5, number 4):

    January 26, 2015 at 9:00 am
    • Jan Heine, Editor, Bicycle Quarterly

      That chart still is a great starting point, but really, it doesn’t matter all that much. Just ride what feels good, and you’ll be fine. When in doubt, let out some air.

      January 26, 2015 at 9:05 am
      • Matt Sallman

        Any thought to update the chart since you now sell 42mm tires and the chart stops at 37mm?

        January 28, 2015 at 8:38 am
  • R. Jones

    Am I wrong or do both of the CX’s show a 10watt difference? That is not what I would call insignificant numbers.

    January 26, 2015 at 9:00 am
  • somervillebikes

    Jan, since tire pressures don’t matter that much and the take-home message is that we should inflate them to what ‘feels’ right, can you comment on the best compromise between comfort and cornering ability? For rough gravel rides, I often deflate my 650x42B tires to about 30 psi, but then sometimes worry that on a tight high-speed paved turn, the front tire will collapse laterally. Is this a myth? If not, what’s a good way to judge how much is enough pressure (without actually going into a fast, tight turn to experiment)?

    January 26, 2015 at 10:00 am
    • Jan Heine, Editor, Bicycle Quarterly

      Tires can collapse, and supple tires will collapse more readily (and dramatically) than those with stiff sidewalls. You’ll notice it when you ride out of the saddle, so you don’t have to wait until you hit a corner at speed and suddenly find your rim lurching sideways… If your bike feels like it has a suspension fork that is bobbing as you ride out of the saddle, add some air.
      We are working on a series of tests that explore all these variables, so that we can come up with new tire pressure recommendations based on what we have learned.

      January 26, 2015 at 10:05 am
  • Giles Pargiter

    In a good humoured way, probably a bit of a maverick post here, but if you are going to claim “science” let’s make sure you really are using it.
    You don’t report the statistical analysis technique that you have used, never mind given us the “raw” data so that we can conduct our own. You say that “…(100-110psi) that most riders use has the highest resistance…” What evidence have you for those two statements? How do you know most riders inflate their tyres this amount? How do you know that the resistance is statistically significant? i.e. where is the maths?
    I have to say both from your investigation and from others that I have read (as well as experience), I do agree with you – but all the same. . .
    Now we are all (mostly) disabused of all this tyre pressure nonsense perhaps you would be interested in testing another series of hypothesis:
    It goes like this: If you have twice as much metal (like for like) on a wearing component it will last twice as long.
    Unless you are right up in the top echelon of racing where keeping precise pace with a fellow competitor matters, then due to the flexibility of the human body it is not neccessary to have more than, say, six gears on a block/cassette. This will mean that the sprockets can go back to being a “proper” 3/32″ thus lasting considerably longer. This will mean that gearing is less complicated and so cheaper. The effect being you can spend less time buying a bike and more time riding it. Also indexing will again become uneccessary and the satisfaction from exercising skill instead of technology to make a gear change will increase rider enjoyment. So that a “cost benefit” analysis will show that it is far more parsimonious to keep it simple. Also the wheel will be a little stronger due to less “dishing”, so widening the range of things one can do with the same machine.
    What do you think?

    January 26, 2015 at 11:01 am
    • Jan Heine, Editor, Bicycle Quarterly

      Scientific studies aren’t published in blogs. There is way too much material for a blog post. Even Bicycle Quarterly isn’t a scientific journal, so we report as much as you need to understand and verify our conclusions, but you won’t get pages of tables with reams of data. We’ll do the statistical analysis to make sure we aren’t reporting noise in the data (which, by the way, none of the other sources on bicycling science seem to do), but we usually don’t bore you with the details of that analysis.
      This blog post presents some of the results. For more details and data, I recommend that you read the Bicycle Quarterly article. It won’t have all the data that you’d find in a scientific journal, but it will have plenty of interesting details and data.

      January 26, 2015 at 11:13 am
      • Giles Pargiter

        Thankyou for that reply Jan, it was a bit of a maverick post – must have been the mood I was in. After all I do read it because I find it interesting and the way you go about your tests useful and free of the usual fashion influence. So I fully take your point about including all the details of the data and the way it was handled.
        I would like to see an investigation into the best effectiveness of gear trains though. I find in the hilly country I ride (N. Wales, UK.) that the range of gears is far more important than the number. I also find that undue number of gears equals short life over the older 3/32″ sprockets and chain rings, with no particular advantage (to me as a rider).

        January 26, 2015 at 4:09 pm
        • Jan Heine, Editor, Bicycle Quarterly

          Measuring the wear and tear on components is involved. Ideally, you do it in a controlled lab setting, where you can quickly rack up 5000 km while spraying the cogs with muddy water…
          However, pure physics tells us that all other things being equal, the larger bearing surfaces of wider cogs and chains wear less rapidly. Maybe that is all we need to know.

          January 26, 2015 at 4:18 pm
      • Francisco

        Wear is a funny thing. I get twelve thousand kilometres out of my ultra-narrow 11-speed chains and change them (for fear of fatigue) well before their ‘stretch’ limits. I stand a lot on the pedals too. Good maintenance? Perhaps, but I used to get only half that distance from the wider chains I used in the past.

        January 30, 2015 at 12:15 pm
    • Conrad

      I like your idea about the number of cogs in back. I think a lot about what a truly optimized drivetrain looks like and I’m pretty sure it doesn’t involve an 11 speed cassette with a super skinny chain. I was initially interested in the new SRAM 1X11 setup, that looks to have an optimized chainring tooth profile and rear derailleur. Then I saw what their 11 speed cassette retails for. Guess I’ll be sticking with my 8 speed cassettes. I would like to see a system built around a reasonable number of gears in back, with the ability to get a lighter, stronger wheel in return.

      January 26, 2015 at 1:59 pm
      • Susan

        I think optimal number of gears has a lot to do with rider physiology. IME, some riders have wide “power bands” and others quite narrow ones, just as internal combustion engines do. I expect it’s related to the individual rider’s proportion of fast-twitch to slow-twitch muscles, and probably also to the tibia-femur ratio, though that’s an untested hypothesis. It’s been 25 years since I raced, and I don’t ever intend to take it up again, but I’m still not comfortable without very close ratio gearing. Others I know seem like they can ride all day equally well at 85 rpm as at 100.

        January 29, 2015 at 5:40 am
  • Jock

    These tests seem so full of variables, aren’t they ‘real world’ tests. Especially when you guys switch bikes for your uphill timing. How can you be confident that one ‘ride’ to the next precisely duplicates the other. I think it’s because you’ve lightened your Waterford Herse thanks to ‘no rubber brake-hood covers’! (Issue #50) : )
    Couldn’t resist. But do like your tires anyway, regardless of ‘science.’

    January 26, 2015 at 1:01 pm
    • Jan Heine, Editor, Bicycle Quarterly

      Your concerns about controlling variables is totally valid. The test of tire pressure was done on the track with a power meter. We measured wind speed (<1 m/s) and temperature (constant) – obviously, we had to pick our test days very carefully. Each tire was run for 4 laps, and each lap was within 3 seconds/1% of the next. So conditions were very carefully controlled. Also, by measuring so many different tires and pressures, if you had one outlier, it would show up pretty quickly. The data all matches – even the higher resistance at moderately high tire pressures, so we are very confident in the results.
      Hillclimb tests always will have more variables, but doing numerous repeat runs helps to get enough datapoints that the noise is less of a problem. The real world is messy, and it's hard to simplify it enough to get good results, and not so much that you no longer have meaningful tests.
      An example of the former is riding your bike on a familiar course by yourself. There simply are too many variables, from wind direction, temperature (which affects rolling resistance) or traffic lights to give you meaningful results.
      An example of the latter is testing tires without a rider on the bike. Even if you use a bumpy roller, you cannot measure suspension losses, so you will miss half of the equation.

      January 26, 2015 at 2:30 pm
      • David Morgan

        Easy-have Hondas robots with set outputs (say 300 watts) bike on 2 bikes identical in all but tire psi; and see how far they ride on a mixed course over a given time! (Now simply have Honda ship some to BQ).

        January 26, 2015 at 3:03 pm
        • Jan Heine, Editor, Bicycle Quarterly

          Robots will have very different suspension losses than riders. The problem isn’t so much the rider, and the idea to have two very similar riders ride side-by-side eliminates variables like wind, temperature, etc., since they are the same for both.

          January 26, 2015 at 4:01 pm
  • Conrad

    I took a cyclocross clinic from a local pro when I first decided to try cyclocross. I was on my mountain bike with 60 or 70 psi in the tires because that is what was printed on the wall of the tire and it was what I had always done. The guy basically berated me and let most of the air out of my tires. Initially I was taken aback and thought what a jerk. Then after one lap I realized I had been doing it all wrong and cringed to think of all the miles I had ridden on trails with that pressure. It remains the single best piece of advise I have ever received for cycling. Spread the word!

    January 26, 2015 at 2:08 pm
    • Jan Heine, Editor, Bicycle Quarterly

      Yes, cyclocross is interesting. In the old days, on relatively stiff 28 mm Barums, I just used a pressure that had me bottom out slightly on the biggest bump every lap. When I was getting used to my FMB tubulars, that didn’t work, because the super-supple sidewalls collapsed when I cornered or braked. I had to experiment a lot with tire pressures to get it right, but now I rarely change it any longer. Unless there is a paved section with corners – my normal ‘cross setup feels very scary in those. Then you have to think about how much you lose by having to go slow around those corners vs. how much you gain on the rough stuff. It’s all part of the fun, and it makes a huge difference.
      One nice thing, and this is where this research is applicable even for ‘cross, is that we now know that even at very low pressures, the tires roll as well as they do at higher pressures. So there is no need to run your mountain bike tires at 60 or 70 psi…

      January 26, 2015 at 2:35 pm
  • specialist

    Tom Anhalt’s testing suggests that tyre pressure is significant.

    January 26, 2015 at 2:46 pm
    • Jan Heine, Editor, Bicycle Quarterly

      His results don’t match what we found in multiple tests, both in rolldowns and on the track with a power meter. I am not familiar with his methodology, but I suspect that his testing was done without the rider on board. That means you don’t measure the suspension losses, and thus measure only half the equation.

      January 26, 2015 at 4:00 pm
      • Tom Anhalt

        Nope. Rider on board, using the same methodogy outlined in the blog post linked below, except varying tire pressure and solving for Crr instead of holding Crr constant and solving for CdA changes. As you can see, it’s can be a fairly sensitive test:
        BTW, what speeds were you testing at? Obviously, energy input due to road roughness is a function of velocity, and the test course I use presents a wide range of speeds in each test lap.

        January 26, 2015 at 10:17 pm
        • Jan Heine, Editor, Bicycle Quarterly

          Tom, thank you for chiming in. If I understand correctly, you are testing on the open road, over a course where your speed varies? I’d be surprised if you could determine small changes in equipment, like tire pressure, among all the noise you’ll inevitably get with that approach. Did you run the same setup twice to compare the results? (Ideally, they’d be the same.) And did you do a statistical analysis to check whether the results you got are just noise, or whether they actually reflect differences in performance.
          We did both those things, multiple times, and found that our results were highly statistically significant. We also tested with two different methods, one a rolldown test, the other the track test described above. During the track test, we held the speed constant at 27.9 km/h (17.3 mph). Trying to keep the power output constant is hard, since the power meter numbers jump around a lot. We found that we could keep the speed within 3 seconds per 400 m lap, which is within 1%. So we kept every factor the same – speed, rider position, wind, temperature, tire, bike – and varied only one: Tire pressure. We also did repeat runs of several pressure settings to make sure that we weren’t measuring some other extraneous change (say it got a little warmer, then a little colder, then warmer again, which would make the tire faster, slower and faster again). Beyond the statistical analysis, it would be unlikely for this warm-cold-warm cycles to happen three times, just as we tested each tire in the “moderately high” pressure range.
          So in conclusion, I am very confident that our results are reflecting what is happening in the real world. We’ve satisfied the standard scientific protocol (repeat runs, statistical analysis) to confirm this.

          January 27, 2015 at 5:59 am
      • Tom Anhalt

        In fact, you may want to subject your methodology to Andy Coggan’s “Tom Compton Challenge”, as mentioned in my blog post to get a gauge on your method’s sensitivity

        January 26, 2015 at 10:27 pm
        • Jan Heine, Editor, Bicycle Quarterly

          I looked at your post, and you have four datapoints that line up nicely. However, any statistician will tell you that four datapoints isn’t very much. What I’d like to see is repeat runs with each setup, so we can gauge how much variation you get with the same setup. And then do a statistical analysis to see how much variation is noise, and at what point you can be confident that you are actually measuring differences in performance. As mentioned above, that is standard scientific procedure.
          Perhaps you did this, and I missed it. If you need help with the statistics, we can provide that. I am very interested in seeing other methods test the same things. That is the best way to ensure that the models we use actually reflect reality.

          January 27, 2015 at 6:03 am
      • Tom Anhalt

        Since you had stated that my own observations were an “outlier”, I’d like to point to this study published in a peer reviewed journal which shows a significant difference in measured rolling resistance at 60 psi vs. 120psi (5.575N vs. 4.215N) and which matches my data better than what you’ve presented. You may do well to get a copy of that paper to take a look at the methodology.

        January 28, 2015 at 7:46 am
        • Jan Heine, Editor, Bicycle Quarterly

          Most tires that can handle 120 psi will be seriously underinflated at 60 psi. I have no doubt that it will roll slower – and it’s probably not safe to ride, either. (That matches our rolldown tests, where we found that below a very low threshold, resistance went up by a lot.)
          In fact, if you look at our curve, you see that between 80 and 65 psi, rolling resistance went up considerably. But those ultra-low pressures aren’t useful on the road.
          From what I understand, the goal of the study was to see whether power meters were sensitive enough to pick up differences in equipment. So they went with very large differences… For our purposes, it would be more interesting to see whether the power output would be different at 90 psi and 120 psi.

          January 28, 2015 at 8:28 am
      • Tom Anhalt

        You’re missing part of the point…the methodology used in that study allowed for the separation of the aero and rolling drag force values so that they can be evaluated separately. As Alex pointed out below, not doing so means that your overall power reported is being greatly determined by the aero drag force. That makes it difficult to “tease out” the rolling drag force. Additionally, that methodology varies the speed over a fairly wide range, something your testing does not do. The regression of power/velocity vs. velocity^2 is a very common method for determining aero drag AND rolling resistance in field testing since a simple linear fit will garner not only the estimated CdA (the slope of the linear fit), but also the Crr (the y-intercept of the linear fit).

        January 28, 2015 at 8:59 am
        • Jan Heine, Editor, Bicycle Quarterly

          What we care about is how to make our bikes faster. So when we find that switching from one tire model to another saves 5% in power, but that inflation pressures don’t make a significant difference, then that is useful data.
          If you are suggesting that what we see is just noise in the data, then the statistical analysis shows otherwise.
          The question on whether higher pressures result in lower rolling resistance has been settled – not just our testing, but also that of all the professional teams who are running wider tires at lower pressures now. And the evidence in the field corroborates this – otherwise, those pros who are still on 23s and super-high pressures would win most races, and the others would quickly go back to the narrower, harder tires.

          January 28, 2015 at 9:18 am
      • Tom Anhalt

        Wait a minute…in that chart above, the differences in power between 65 and 80psi are LESS than other differences in that same chart which you say there is “no difference”…which is it? I’m confused.

        January 28, 2015 at 9:04 am
        • Jan Heine, Editor, Bicycle Quarterly

          Look, if you just want to prove us wrong instead of engaging in a constructive discussion, what is the point? Don’t we all want to understand bikes better?
          Nobody said “no difference”, but there is no clear trend. So yes, it’s probably better to err on the lower side of tire pressures… or go super-high, but not in the middle. But in reality, for most riders, it’s hard to know where “the middle” is. In any case, the main point here is that without this data, wide, supple tires wouldn’t make sense. And that is the true revolution, not whether you should let out 15 psi from your tires to make them a little faster.

          January 28, 2015 at 9:22 am
      • Tom Anhalt

        You state there’s no clear trend and imply that that means there’s no reason to worry about lower tire pressures making one slower. Well…if the difference between 65 and 80 is significant, then so are the other differences you show (assuming they are from rolling resistance differences, that is).
        Yes, I too want to figure out how to best make bikes faster. I’m not trying to prove you wrong…in fact, in general we are in pretty much agreement on the subject of tire pressures (“Tis far better to err on the side of too little pressure than too much”). I’m just trying to get you to see that despite all of your analysis of the data, it may not be showing you what you think it’s showing you. There ARE more insightful methods available which you could be using, especially with an on-board power meter. These other methods would allow you to separate out the aero and rolling drags so that you can not only discern differences in the rolling drag, but also ensure the aero drag hasn’t unknowingly changed. Both of those things are readily apparent in the linear regression method in the Lim paper, and also in the VE method I pointed out to you. The interesting thing about at least one of those methods (VE) is that it may be possible to use it to evaluate the data you’ve already taken to either add confidence in your results, or perhaps indicate where the measurements didn’t account for unplanned changes. If you can post those power files on line it would be greatly appreciated. I think you may be surprised at the level of detail that can be gleaned from a simple power file.
        As far as the comment that “In any case, the main point here is that without this data, wide, supple tires wouldn’t make sense.” I’m not sure why you say that. For certain applications, wider tires will always make sense…and the habitual over-pressuring of tires in the past (regardless of width) is something I’m glad we’re finally convincing folks is folly. The data plot you show above isn’t needed to change anything about that. I would caution to not let your love for all things “650B and wide” color your evaluation of your data.

        January 28, 2015 at 10:07 am
        • Jan Heine, Editor, Bicycle Quarterly

          My love for “all things 650B and wide” came after the tests. I only started to realize the potential of wide, supple tires after we started testing tires. Before, I had planned to order a new randonneur bike with 28 mm tires…

          January 29, 2015 at 4:57 am
    • Tom Anhalt

      Hi Jan,
      Rather than attempt to get you up to speed on the VE methodology in the obviously constraining media of the comments of a blog post, I highly recommend you review the .pdf put together by Robert Chung on the ins and outs of the method, which was linked to early on in my blog post. In case you missed it, here it is:
      On the subject of statistical precision, I will point to you slide 38 of that .pdf. The CV estimate of as low as 0.3% is based on a typical file of my own testing. Instead of using an overall average as a single data point, one of the advantages of the method is that every “second by second” data record is used to create the VE profile and thus the totality of the data is “leveraged” due to the fact that they are correlated within and across laps.
      I highly encourage you to review Robert’s presentation and after digesting that, if you have any additional questions I would be more than happy to attempt to address them.
      The “beauty” of the VE methodology is that it is unnecessary to control for speed, power, or even elevation differences…in fact, it’s helpful if they DO vary, which is one of the main reasons I do my testing on what is affectionately dubbed a “half pipe” course. Varying the speed allows one to more easily “pry apart” the Crr and CdA values due to the fact that the Crr drag force is constant while the aero drag force varies with the square of the velocity.
      Lastly, and yes, for the tire pressure testing shown in that chart above, I did repeat at least one pressure (the 8bar results IIRC…it was quite a few years ago when I did that).
      Again, I’d recommend you guys attempt a “Tom Compton Challenge” of your own methodology to see what sort of differences you can reliably detect. In this case, since the interest is in Crr values, one thing you could do is have the rider wear a backpack with varying masses in it and see if that’s reflected in the estimate of the rolling resistance force (i.e. do the analysis assuming the mass is constant and see what resulting Crr difference would need to be to match…since the Crr force is merely Crr x m x g, that’s a reasonable approach.)

      January 27, 2015 at 7:44 am
  • Gideon Glass

    As a (relatively) happy owner of the Compass 700×38 Barlow Pass tires, I don’t think I disagree with any of the above. However, the one thing they leave me wondering about is hard efforts out of the saddle (> 400 Watts say), especially going uphill where there is a lot of weight over the front tire. This seems to improve with higher pressure, to some extent, but there is still a bit of squirm and deflection and I can’t help but wonder if that is consuming energy and slowing me down.
    Let me suggest the following experiment. Find a hill, or a set of them, suitably short to permit high intensity efforts, say 50-200 feet of elevation gain (or 50, 100, 200). With a calibrated power meter, and changing only the tires (ideally on separate wheels to make changes faster), measure the total energy expenditure (kJ) required to get up the hill at hard effort, along with, of course, time. Maybe do a few riders; each person takes a turn, then swap the wheel & tire; do another run for each test rider; then swap wheel & tire back; do, say, 4 runs on each tire, so each rider goes 8 times (at least for the short hill; 6 is probably fine for the longer hill!). Plot time and kJ on the Y axis and interval number on the X axis. If you get a curve like \/\/\/\ and it’s statistically significant, then it indicates the second tire is more efficient.
    In any case, thanks for all of your efforts in these areas over the years.

    January 26, 2015 at 3:35 pm
    • Jan Heine, Editor, Bicycle Quarterly

      We could do the uphill experiment, but I don’t think it would tell us much. After our initial rolldown tests, we were concerned that pedaling might introduce more squirm of the tires, and thus increase resistance at lower pressures. So we tested on the track, and found that it didn’t matter enough to make a difference.
      On hard uphill efforts, the front wheel is unloaded, both from the inclination (front wheel higher than rear) and the torque of the pedaling. The rear tire will see a bit more shear, but I doubt it’s enough to consume significant energy.

      January 26, 2015 at 4:05 pm
  • exmaschine

    Great article! I always thought that as well, that pressure didn’t have much of an effect on speed.
    Though at first thought, it doesn’t add up. Because of the friction and rolling coefficient. You would think it would slow the tire speed down…right? But I suppose it has to do more with the rubber compound.
    I fiddle with pressure strictly for the purposes of traction (friction) I typically run 100-105 in the warm and hot weather and 85-90 in the cool and cold weather. Never had a pinch flat either.
    I adopted this notion from when I raced motorcycles and the tire reps would always recommend lower than normal pressures when track temps were cool or cold.

    January 26, 2015 at 6:46 pm
    • Jan Heine, Editor, Bicycle Quarterly

      Generally speaking, for bicycles, lower pressures result in better traction – less bouncing of the tire – up to the point where the tire sidewalls starts to collapse. So you might try running 85-90 in warm weather, too.

      January 26, 2015 at 7:04 pm
      • exmaschine

        Will do Jan! Thanks.

        January 26, 2015 at 8:36 pm
  • marmotte27

    Could one downside of wider tires with lower pressures be that they throw up much more water from the road? I’ve mounted the Compass Loup Loup on my bike, and running through (really) small puddles or water in ruts on the street so much of it is thrown up, that it splashes around the mudflap (which goes down to 5cm above the road) and onto my feet. Maybe the thread of the tires is also a factor in this?

    January 26, 2015 at 11:29 pm
    • Jan Heine, Editor, Bicycle Quarterly

      It makes sense that a wider tire would pick up more water. As far as tire tread goes, I haven’t noticed much difference between a shaved Hetre (no tread), a Hetre (longitudinal ribs) and a Babyshoe Pass (chevron or file tread). All are the same 42 mm width.

      January 27, 2015 at 5:31 am
  • John Duval

    I have read several product introductions lately for 28mm tires with the same construction as the makers premium tires, usually including the high levels of puncture protection. More consumers seem to be demanding products that are different than racers are using, including clencher tires, disk brakes, wider tires, and tubeless. In spite of this, light weight is still credited with mythical levels of performance enhancement, especially when it comes to wheels and tires. These things do give a sense of speed, even where the difference over any distance is insignificant.

    January 27, 2015 at 12:20 am
  • Alex Simmons

    Can you please provide details on your test protocol and analysis methodologies and post the power meter and environmental data for all the runs?

    January 27, 2015 at 1:11 am
    • Jan Heine, Editor, Bicycle Quarterly

      Protocol: Track at Marymoor Park in Redmond, WA. We used the inner apron, which is paved with smooth asphalt, not the track, which is concrete. No wind, small temperature variations (for which we corrected based on our reference tires). Four laps per setup, with a power meter at constant speed. (Times for individual 1/4-mile laps were within 3 seconds/1% of each other.) Repeat runs of a reference tire at the beginning, middle and end of each test day, to make sure conditions hadn’t changed. Rigorous statistical analysis of the results showed that the results are highly significant. Even the local maximum of tire pressures around 100 psi was statistically significant. It’s just not actionable, since it probably depends on road surface…
      The Power Meter data were published in Bicycle Quarterly Vol. 11, No. 3.

      January 27, 2015 at 5:36 am
      • Alex Simmons

        Rigorous examination of your test protocol suggests it is inadequate for the stated purpose and your conclusions are not valid.
        Since rolling resistance in your scenario represents ~15% of the total power demand, then you need very strong protocols and analysis methodology to account for factors that affect the other 85% of power demand, before drawing any conclusion about Crr variances.
        Were the constant speed runs for each set up performed at a variety of speeds? e.g. 10km/h, 20km/h, 30km.h, 40km/h? If you are only using one speed and not making a change to some other controlled variable (e.g. a large known mass difference), then you are kidding if you think you can say anything about Crr from such tests.
        How did you control for the different power-speed of the rider from run to run?
        Sounds like speed was based on lap times and lap distance.
        How did you control for the distance travelled per lap?
        A 3 second variance in lap time on a 400m track at speeds of around 30km/h = 15%-19% variance in power.
        So, how did you account for the power-speed variance?
        Did you account for the kinetic energy difference between the start and end of each run?
        How did you measure or control for CdA variations between runs?
        Did you calculate a estimate of the CdA and Crr for each set up?
        Was this all on same day of different days?
        Tyre temperature is not the only variable. Air density matters as it affects the speed-power relationship quite a bit. That’s also a function of more than temperature. Did you measure air density, or at least the barometric pressure and temperature at time of each run (we can assume altitude doesn’t vary, and impact of humidity is negligible)?
        No measurable wind? It doesn’t take much. Only need 0.34m/s wind variance (less than say a Kestrel can measure and barely noticeable to a human) to result in a 10W difference in power at same speed when riding at around 30km/h.

        January 27, 2015 at 12:31 pm
        • Jan Heine, Editor, Bicycle Quarterly

          Rigorous examination of your test protocol suggests it is inadequate for the stated purpose and your conclusions are not valid.

          Your questions indicate that you haven’t read the study, so I am surprised that you draw these conclusions.
          We did runs at various speeds for other tires. The control for distance per lap was easy: We rode on the white line. Any variation in distance is going to be tiny. I’d be more concerned about slowing down due to wobbling, but the bike had a very stable geometry, so that wasn’t a problem, either. We did each run with a flying start, so no acceleration needed. The drag coefficient we measured in the wind tunnel. We found that the rider could easily assume the same position time and again, so there isn’t much variation there. All runs were consecutive on the same day, so air pressure didn’t change much.
          Finally, we did not calculate Crr (coefficient of rolling resistance). We only compared power outputs with different tires, or different pressures. So we know the Crr is the same for these pressures, but we don’t know its absolute value.
          I hope this answers most of your questions. I suggest you find the back issue – ask your library if you can’t afford it – and read the study before you conclude that it’s no good.

          January 27, 2015 at 2:58 pm
    • Tom Anhalt

      I’m pretty sure Alex is inquiring about the “second by second” PM data files, not any power average summaries you may have published in an article. Those data files would be helpful to see since they could easily be plugged into GoldenCheetah’s Aerolab utility to get an alternative evaluation of what the data “says”.

      January 27, 2015 at 7:48 am
      • Jan Heine, Editor, Bicycle Quarterly

        I believe we have those files somewhere. We did analyze them in the same second-by-second approach that you suggested. We have the advantage that the conditions are carefully controlled. There really isn’t an alternative explanation of the data, especially with so many test runs – 22 alone shown in the table, but we also ran a Grand Bois 25 mm from 120 to 200 psi (4 runs), and again found no difference in power output for the same given speed. Then you have the roll-down tests, where we tested 7 different tires and found no great variation in speed (for the same energy input) at different pressures for each tire.
        The fact that the professional teams have since done their own testing and also concluded that lower pressures (and wider tires) are fine indicates that the results hold up well.
        So your data really is an outlier. I am concerned that with your method, there are too many variables. I understand that that can be an asset, but only if you are aware of every factor that is changing and consider it. Even a small change in temperature will cause a significant change in rolling resistance… and unless you control for that, your results will be skewed. So if you test pressures one after another, and the day warms up, you’ll get faster and faster times, but they have nothing to do with the tire pressure.

        January 27, 2015 at 8:05 am
      • Tom Anhalt

        Can you make the files available? I’d love to take a look at them. I’m willing to share my raw PT files as well…I’m pretty sure I still have them on my ancient desktop PC.
        Yes, I’m well aware of the affects of temperature, which is why I have temperature compensation terms built into my spreadsheet for both air density AND tire Crr.
        To be honest, I’ve seen your coastdown data previously…and I have concerns about the hand-timed nature of the testing and the ability to reliably discern differences…so, it doesn’t surprise me you had a hard time “seeing” those pressure differences.
        I also want to re-emphasize that with any tire testing of this type, it’s important to make sure that the test encompasses the full range of speeds one would expect in actual riding. Again, energy input due to road surface roughness is a function of velocity.

        January 27, 2015 at 9:10 am
        • Jan Heine, Editor, Bicycle Quarterly

          I’ll dig around for the files.
          Your concern about hand-timing is valid, which is why for each tire and pressure, we did three runs. That gives us a good idea about how much variability we have.
          We found that hand-timing is very accurate, at least within 0.2 seconds. They used to time the Olympics that way… and with our runs taking 25-30 seconds, and significant differences between tires being on the order of 2-5 seconds, depending on the tires, an error of 0.2 seconds makes little difference.
          That is why the statistical analysis is important – it allows you to see whether the noise in the multiple runs of the same tire is small enough that you can discern differences between different tires. We found that for very similar tires, even though we got different numbers, they weren’t statistically significant. (For exmaple, speed difference between Michelin Pro2 Race and Continental Ultra Gator Skin was not significant.) Others were statistically significant (for example, Panaracer Pasela vs. Rivendell Rolly-Poly).
          So you are right, based on the track test, we can say only for speeds of 27.8 km/h that tire pressure doesn’t impact speed. The roll-down tests were at lower speeds, and again, found none of the great variations with tire pressure that your data shows. Furthermore, there doesn’t seem to be a mechanism that would indicate that tire pressure doesn’t matter at 27.8 km/h, but that it does matter at other speeds.

          January 27, 2015 at 9:35 am
        • Jan Heine, Editor, Bicycle Quarterly

          My concern with your method is that you have a huge number of variables: Speed, gearing, temperature, wind, rider heart rate, power output, tire pressure, position on the bike, the line you pick on the course, traffic passing, dirt on the road… and many others we may not even be thinking about.
          You record as many variables as you can, and hope the others don’t matter. To show that the others don’t matter, you’d have to do a large number of repeat runs and compare them, yet you’ve done only one repeat…
          Compare that with our tests, where we a) keep the system as simple as possible, so there are fewer variables; b) test successively on the same day, again to keep the variables down; and c) do multiple runs with the same setup to show that we really are keeping the variables low.
          Science is about isolating variables, and two of our editorial team have spent many years learning how to do this on their way to obtaining Ph.D.’s.

          January 27, 2015 at 10:28 am
      • Tom Anhalt

        You agree that the response of the tire+rider “suspension” system varies with the power spectral density of the road surface roughness energy input, correct? Well, it can be shown analytically that the input PSD is proportional to the vehicle velocity. Additionally, the tire pressure varies the stiffness of the dominant “spring” in the system, and thus one would logically expect the tire pressure to affect those same responses. I would suggest you find a copy of the paper “Effects of Road Roughness on Vehicular Rolling Resistance”, by Xiau-Pei Lu for a really in-depth technical discussion on how these elements are interrelated.

        January 27, 2015 at 9:59 am
        • Jan Heine, Editor, Bicycle Quarterly

          I totally understand that, but if pressure doesn’t matter at two speeds, it probably doesn’t matter in between, unless you can propose a mechanism why there is a U-shaped curve of “influence of tire pressure on speed”.

          January 27, 2015 at 10:02 am
      • Tom Anhalt

        The mechanism is, if I understand what you’ve written correctly, what you have previously described as “suspension losses”. Basically speaking, once the tire has been stiffened up to a point that it is no longer acting effectively as the primary “suspension”, then road energy is transmitted through the frame and dissipated in the “squishy bits” of the rider at the contact points. The same effect occurs at a given pressure and road roughness when the vehicle velocity is increased.
        As that paper I mentioned above describes, the rolling resistance losses can be broken into 3 main components:
        a.) smooth surface rolling losses
        b.) energy dissipation in the tire due to road roughness, and
        c.) losses in the suspension system due to relative motion between sprung and unsprung masses.
        In the case of road bicycles, that “suspension” is represented by the spring constant and damping coefficient of human flesh. When the pressure gets too high for a given road roughness and velocity…or the velocity gets too high for a given pressure and road roughness…or, the road roughness gets too high for a given pressure and velocity…then at that point the increasing losses in the c.) term begin to “overwhelm” the a.) and b.) terms, which actually decrease with continuing increase it pressure due to less tire deflection. You actually observed this exact same thing in your “rumble strip” testing.
        A more simplified explanation is described in the “Tire Pressure” section of this article I wrote back in 2009:

        January 27, 2015 at 10:33 am
        • Jan Heine, Editor, Bicycle Quarterly

          You are right about suspension losses, and that is the reason for the non-linearity of the power vs. inflation curves. Clearly, the harshness of the tires goes up quicker than the reduced hysteresis, hence the higher resistance at medium pressures. Then, as you suggest, the harshness doesn’t increase much (it’s already very high), while the hysteresis still decreases. Overall, the two factors (suspension losses vs. hysteretic losses) more or less cancel, hence resistance doesn’t change much with tire pressure.

          January 27, 2015 at 2:52 pm
      • Alex Simmons

        While an Aerolab of the power file on the face of it removes the lap distance and timing issues, we are left with two main problems:
        1. We have to assume:
        i. the power data is good (so need more information on that) and
        ii. the wheel speed data is good (e.g. wheel circumference errors due to pressure or changing tyres)
        2. Aerolab will tell us is the size of the net difference in resistance forces between runs (which we normally all lump into an equivalent impact to CdA), but it won’t tell us the individual contribution of each resistance force that makes up that net change. For that you need stronger test protocols and analysis methodology (which you are are of course well aware of).

        January 27, 2015 at 12:43 pm
      • Tom Anhalt

        BTW, I think you may misunderstand the methodology I used since you say “You record as many variables as you can, and hope the others don’t matter. ” Aside from the weather conditions (which is necessary to correct for air density effects) the only things necessary to be recorded are speed and power for a VE analysis. I too test under calm conditions and early mornings to minimize weather variations within a test.

        January 27, 2015 at 1:35 pm
        • Jan Heine, Editor, Bicycle Quarterly

          My concern is that you don’t even know which variables don’t matter until you have shown that they don’t matter. Simply assuming this isn’t enough. And even the weather conditions are hard to measure for a long ride. In our testing, we measure wind at the beginning and end of each run that takes about 4 minutes. If there is any wind during the run, we throw it out without even looking at the data. Over the course of a 4-hour test session, even on a day with absolutely still air, we have a good handful of runs where there is a slight gust of wind. The time keeper is also watching for wind… which is easily done by looking at the leaves on the trees surrounding the track. (The leaves actually are a more reliable indicator than a wind speed meter, but we use both to satisfy critics who like to see numbers.)
          If we used a road course, there would be no accounting for this…

          January 27, 2015 at 3:10 pm
        • Jan Heine, Editor, Bicycle Quarterly

          I can see a few potential problems: What if your bike is getting faster from test run to test run, for example, because your new bottom bracket has seals that free up. Or you are testing new tires, and the chords of the casing are loosening a bit, making the tires faster. (We always run tires for 50 miles on the road before we test them.)
          If you ran the same setup at the beginning, in the middle and at the end of the test, you’d discern those trends. Repeatability is one major requirement for scientific studies, and yours unfortunately doesn’t meet it. (I also liked your study of crank length, but again, from two runs, one with each setup, you cannot draw any conclusions that are statistically significant. I would run at least five tests with each setup, so you have more datapoints to work with.)
          This doesn’t mean that your results are incorrect. Perhaps there are some tires that will get faster at higher pressures. We’ve only tested tires with relatively supple sidewalls. I could see a tire that doesn’t flex much – thus has high suspension losses at any pressure – to benefit from higher pressures that reduce the hysteretic losses. I haven’t been able to find your test – can you provide a link?
          All this is speculation until you can repeat your tests and show that the trends still hold.

          January 28, 2015 at 7:32 am
      • Tom Anhalt

        You really need to read and understand Robert’s .pdf on the Virtual Elevation methodology before commenting further about what is and isn’t taken into account. Any variances in external conditions are easily seen in the VE trace and can be dealt with accordingly. Each of my test runs takes less than 10 minutes. I rarely have a test session last longer than an hour.
        Alex brings up a good point. How did you account for changes in wheel rollout for each run? I’m assuming speed was measured using a wheel magnet, correct?

        January 27, 2015 at 4:26 pm
        • Jan Heine, Editor, Bicycle Quarterly

          We accounted for the difference in wheel rollout by adjusting for the distance traveled. We know a lap is 400 meters and a few centimeters – don’t remember the exact value, but it was measured since it’s a track used for championship events. So it was easy to correct for this.

          January 27, 2015 at 4:29 pm
      • Tom Anhalt

        Also, my test course is in a sheltered, remote neighborhood. As I said, I do my testing early in the morning to help avoid whatever little traffic there is in and out of that neighborhood in the first place. Vehicles on course will abort a run…luckily the runs are short 😉
        However, in the end the “proof of the pudding is in the tasting”, as they say…I think it’s time for the BQ crew to do their version of the “Tom Compton Challenge”. Do a test of KNOWN changes in either rolling resistance force (e.g. using the mass changes I suggested earlier) or know changes in CdA (using something such as the spheres I described in my blog post) and see to what level your methodology can discern.

        January 27, 2015 at 4:31 pm
        • Jan Heine, Editor, Bicycle Quarterly

          Hi Tom,
          I am glad you are running your tests under more carefully controlled conditions. I’ve poked around your web site, and I like your approach. For example, your test of crank lengths was pretty cool. Even there, you need more runs, though, to get statistical significance. It’s tedious, but crucial…
          Anyhow, I’ll think more about your data. While I have total confidence in ours – repeated dozens of times with two methodologies – your data raises some interesting points. In the mean time, I can tell you that on my current tires, I roll as fast at 30 psi as I do at 50, and as fast as my bike with narrower tires did at 80 psi. Whether it’s my time in brevets or riding with friends who ride at my speed, there is no difference. And the pros who switched to wider tires at lower pressures also didn’t start losing races immediately after the switch…
          I hope we can reconcile our data somehow in the future.

          January 27, 2015 at 4:39 pm
      • Tom Anhalt

        That rollout correction could be problematic…first, as you well know, the 400m measurement is ONLY at the lower track line. Since you guys were riding on the apron, obviously the distance is shorter…and, most likely variable. See Alex’s blog post on the possible effects of taking just slightly different lines around a track on the distance traveled:

        January 27, 2015 at 4:38 pm
        • Jan Heine, Editor, Bicycle Quarterly

          Correcting for the shorter distance was simple – it’s a simple geometry problem. Anyhow, the statistical analysis would show if we were getting a lot of noise. The fact is that our runs are extremely repeatable – meaning that if we ride the same setup later during the day, we get exactly the same results. Then we do statistical analyses to make sure the repeatability wasn’t due to dumb luck.
          So instead of theorizing where there might be problems, you’ll have to accept that the statistics show that there aren’t problems. The variables are well-controlled. Again, that is standard scientific methodology, not something we made up. We only report results that meet statistical significance, i.e., that fall within the 95% confidence interval.
          That leaves us with the possibility that we have inherent problems in our methods. This is the reason we used two different methods – rolldown and power meter on the track – to confirm the results.
          Let’s move the discussion of this blog, since your studies aren’t really part of this post, and only came up in the comments.

          January 27, 2015 at 4:45 pm
  • aquilaaudax1

    Jan how much do you think frame geometry effects tyre performance?
    I have 2 bikes that I regularly ride. One is a Australian built Geoff Scott race bike. Tubing I believe is Tange Champion no.1. Geometry is classic eighties, parallel 74 degree head and seat tube angles. The frame is 24″ centre to top with a top tube of 57cm, centre to centre. The other is a Japanese built early 80’s sports tourer. Frame tubing is Tange Champion no.2. Geometry looks to be 72 degrees parallel. Seat tube is 25″ and top tube is 56cm. Both bikes are setup to fit the same although the sports tourer has the handlebars only an 1″ below saddle height. The sports tourer weighs about 3 kg more then the Geoff Scott due to mudguards, lights, higher spoke count wheels etc.
    To each of these bikes I have mounted Challenge Parigi Roubaix tyres (These have picked up very cheaply of the ebay and from various English online stores. Compass tyres are way out of my price range with the collapse of the Aussie dollar). The tyres are inflated on both bikes to 80psi rear and 60 psi front.
    Where I live there is a very fast, highly technical descent which I do at least once a week. I am an average climber but I am certainly up at the top end of the field when it comes to descending and I think I know how to handle a bike.
    My best time on the GEFSCO –
    My best on the sports tourer –
    As you can see I am not really holding back going down. Strangely though on the GEFSCO I can really lean the bike into the bends and it wont drift or move offline. However on the limit the sports tourer has rear end drift and what feels to me to be the rear tyre rolling over on the sidewalls. On my last descent on the tourer this drifting/rolling actually caused a 50kmhr front end shimmy, mid corner. A hair raising moment to say the least. Thank goodness I know how to deal with them.
    What would you say is the cause? It can’t be tyre pressure as I have no issues with it on the other bike. It can’t be the weight, as 3kg difference between the two bikes couldn’t have a huge influence when factoring overall weight of bike and rider. All I can think is that it’s frame geometry is influencing the way the tyres perform.

    January 27, 2015 at 2:38 am
    • Jan Heine, Editor, Bicycle Quarterly

      Frame geometry – especially the front-end geometry – will influence how a bike corners, so that is probably what you feel, and so you hold back on the bike that doesn’t inspire confidence.

      January 27, 2015 at 5:38 am
  • Nathan

    Hi Jan, I really like your posts on tyre science. Is there any credibility to the theory that bigger tyres accelerate or “spin up” more slowly due to them being heavier?
    I couldn’t detect the weight penalty going from 23mm to 25mm, but my 27mm’s do feel a bit slower to accelerate (but they are just as fast). Is this legit or just placebo?

    January 27, 2015 at 4:09 am
    • Jan Heine, Editor, Bicycle Quarterly

      Physics tells us that the tire weight shouldn’t matter. Cyclists just don’t accelerate that fast – even a pro cyclist has a power-to-weight ratio that is 1/10 of the power-to-weight ratio of a sluggish economy car.
      The experience of pro cyclists seems to bear this out. The easiest way to make your wheels lighter (and thus “spin up” faster) is to make them smaller. Bike builders have tried out smaller wheels time and again, whether it was Cino Cinelli in the 1960s, or Ernesto Colnago on the lovely machine he built for Guiseppe Saronni (shown in Bicycle Quarterly‘s 2014 calendar). None of these were raced, probably because the riders found that the advantages of the smaller wheels were too small to outweigh the disadvantages… If a lighter wheel brought a huge advantage, a sprinter like Saronni should use it. Even if they cannot get a spare in case of a flat tire, they’d win every race where they don’t have a flat. Those would be better odds than they currently have!
      By the way, UCI rules require a minimum wheel size of 55 cm, but current 700C wheels are 10 cm over that limit, so there is lots of room for “improvement”.

      January 27, 2015 at 5:43 am
    • Jan Heine, Editor, Bicycle Quarterly

      One thing that does change is how the bike feels as you rock it from side to side. So during an out-of-the-saddle sprint with heavier wheels, the bike will feel different, even if the wheels don’t “spin up” slower.

      January 27, 2015 at 5:43 am
  • Michael

    Is this a safe practice to run tires below/above recommended manufacturer ratings?
    Below 50 psi is under Panaracer’s limit for a Hetre, though I have read reports of 30-45 being used.
    I even remember one Vol. 5 BQ article where it said Cyprus tires performed best at 80 or 85psi(forget which), which is above that tire’s current max pressure rating of 75psi.
    I would expect that the manufacturer would know the limits of their tires. Makes me wonder about the safety of over/under inflation, or if it taxes the structure of the tire, leading to shorter lifespan of the tire? Please comment.

    January 27, 2015 at 6:53 am
    • Jan Heine, Editor, Bicycle Quarterly

      The maximum pressure should not be exceeded. Above that pressure, the tire can blow off the rim, or self-destruct under the forces of the air pressure inside. Of course, there is some leeway in this, since it depends on the tolerances of tire and rim. If you have a slightly oversize tire on an undersize rim, you’ll run into trouble quickly. In the opposite case (undersize tire/oversize rim), your risk of a blowout is much lower – but the tire will be hard to mount. By the way, the maximum inflation of the Grand Bois Cyprès is 95 psi. (75 psi is the “recommended minimum” pressure.)
      The minimum pressure has no real engineering reason. It exists only to show riders that there is a pressure range, and that they don’t have to inflate to the “number that is on the sidewall”. If you run your tires with too little air, individual threads of the casing will start to break. You can see that as a zigzag line that goes along the casing, but it’s not dangerous. Most of all, the tire begins to squirm as the sidewalls collapse in cornering. For me, this happens on a Hetre around 30 psi… which is quite a bit below the range of the 50-75 psi listed on the sidewalls.
      Compass tires don’t list a minimum inflation pressure. We trust that our customers know not to inflate to the maximum, unless they have good reasons to do so (heavy rider/tandem).

      January 27, 2015 at 7:15 am
      • Michael

        Thanks for the info, Jan.
        I’ll feel free to air down if needed. I run my Hetres at 50-55. But lower would be fun to try.
        BTW, the sidewalls of my Cypres tires I bought from Compass this past fall says to keep inflated to 55-75 psi.
        So I am thinking that means 75 is the max. Is this wrong?

        January 27, 2015 at 10:13 pm
        • Jan Heine, Editor, Bicycle Quarterly

          You are right, the 650B version says 75 psi. I suspect that when they made the mold, they weren’t sure yet what pressure a supple, wide tire could accept. The tire is the same as the 700C version, so 90 psi should be safe.

          January 28, 2015 at 9:13 am
      • Michael

        I guess the 75-95 must be the 700c version.
        What’s the sidewall rating for the 650b Compass Loup Loup?
        What’s the lowest you recommend for a 170lb. rider on Loup Loups?

        January 27, 2015 at 11:04 pm
        • Jan Heine, Editor, Bicycle Quarterly

          The maximum pressure for the Compass 38 mm and 42 mm-wide tires is 75 psi. For a 170 lb. rider, I wouldn’t go much lower than 40 psi with 38 mm-wide tires, unless you are on really rough terrain.

          January 28, 2015 at 9:14 am
  • Ted Arnold (Lifer)

    Maybe noteworthy to those on a budget that lower PSI on more supple rubber can mean more cuts which could mean more flats. A lot of years in shops experimenting with road PSI. Allen Lim, spoke of similar results to yours when testing for Powertap Variables.
    For years I’ve aimed for a balance of supple and protection. I run 80-90psi and my two favorites are GP4000, the sleeper tire is the Vittoria Diamonte. It’s often overlooked, but has an affordable 220 TPI nylon casing so it’s not as supple as the magic carpet of the Open Corsa, but more durable and a better daily rider option. I’m not sure thought that they have 25.
    On the pro front I was working with radio shack a little when they came over to the wider tubie rim. The problem with the new rim was all the tubies were 23 and Challenge had the only tubie they could use for a while. Later they got on Schwalbe 25’s. But, picture this problem, a wide carbon tubie rim would hit pavement in hard cornering (ouch) or crack rims as they’d bottom out on hard impacts and the carbon would go to the pavement. The riders preferred the wide rims for sure, but went through a lot of test sets as they got the right tires.

    January 27, 2015 at 7:08 am
    • Jan Heine, Editor, Bicycle Quarterly

      Thanks for the perspective. Interesting to hear about the issues the pros had with finding wider tires – no doubt constrained by sponsorship contracts. (Vittoria and FMB have been making 25 mm tubulars for a while now.)
      However, the “more cuts with lower pressure” doesn’t seem to be true. After all, to cut the rubber, you need the tire to push hard against the object that is trying to cut it. At low pressure, the tire will just deform around the object instead. (It’s like trying to cut something with a knife – the harder you push on the knife, the easier it is to cut.)

      January 27, 2015 at 7:23 am
      • Ted Arnold (Lifer)

        I understand the supple aspect, but feel like a supple casing tire and soft durometer tire at lower psi will have greater deformation/bigger contact patch. More Tire on ground contacts more debris. Thanks for reading.

        January 27, 2015 at 7:36 am
        • Jan Heine, Editor, Bicycle Quarterly

          Yes, you will roll over more debris with a wider tire. But the field experience shows that you reduce the amount of debris that penetrates by so much that you come out far ahead.

          January 27, 2015 at 7:57 am
  • Chris Cleeland

    Can you explain what the graph is actually showing? The axes are “pressure” and “power”, but you talk about “speed”.

    January 27, 2015 at 9:42 am
    • Jan Heine, Editor, Bicycle Quarterly

      It’s the power output required to ride at a constant speed of 27.8 km/h, for different tire pressures. Less power is better.

      January 27, 2015 at 10:00 am
      • Alex Simmons

        Since I can’t reply to your earlier response to me – I’ve no idea where the details of your experiment are published. Please post a link so I and others can read them.
        Or outline your testing protocol, analysis methods, and if possible the actual data.

        January 27, 2015 at 6:01 pm
  • exmaschine

    Wanted to add some fringe relevance to the discussion Jan.
    As the evolution of motorcycles progressed (drums to discs, to advancements in suspensions, traction control, tire technology etc) so goes the MTB and so goes the road bike. (albeit not as much or as quick) Though that will change in the coming future. We know have electronic controlled suspensions in motorbikes and now mountain bikes. That will progress further. I believe we will see ‘active’ suspensions (in some form) on road bikes as well. As cost and weight issues are resolved ( and they will be), this is where technology is headed.
    Eventually, I think all bicycles will have a fair amount, if not all electronically controlled “mechanisms.” So, perhaps, though not definite, the last bastion of the art/science to chassis setup on a bicycle and motorcycle for that matter is and will be tire pressures. It’s great to be a part of the advancement in two wheeled machinery, but at the same time, technology is removing the human engineering aspect. Though I suppose electronics will eventually take care of the pressures as well. With pressure sensors, etc.
    Software is replacing ingenuity (yes, you could argue that humans have to create the software, but, I can argue that software can and does create ideal conditions and remove a significant amount of human trial and error) But to me, figuring out chassis settings whether on a mtb or road bike is becoming a lost art. I’m old school, I love trial and error of setup, I love being able to ‘dial-in’ a bike, to achieve that sweet spot.
    I know technology is for the most part unstoppable and in some ways it is beneficial, but in other ways, I think it is also detrimental imo. C’est la vie!

    January 27, 2015 at 11:20 am
    • Tom Anhalt

      In regards to discs on road bikes, I’ll quote Bill MacReady of Santana Tandems: “Rim brakes have always been disc brakes. When cars and motorcycles were fitted with disc brakes, they caught up to the braking efficiency bicyclists had known for a half-century.”
      In regards to active suspensions on road bikes… due to the lower speeds and relatively limited “motors”, the fastest road bike setup will always end up being the setup which allows the tire to act as the vast majority of the “suspension” in the system, since that energy to deflect the tire can be mostly returned in the trailing half of the contact patch. Additional suspension elements in a road bicycle will only dissipate energy and be heavier.

      January 27, 2015 at 5:06 pm
      • Jan Heine, Editor, Bicycle Quarterly

        I think everybody agrees that the tire acts as the vast majority of the suspension, especially for high-frequency vibrations. Beyond that, the views you quote on disc brakes and suspension are simplistic and leave out a number of important variables.
        Disc brakes have a smaller rotor, thus higher clamping forces, which wipes off water much more quickly when riding in the rain. (They also are further away from the tire.) So in wet weather, they react more quickly, which is a major advantage.
        The downside is that you need stiffer fork blades to deal with the greater forces at the dropout. So you lose comfort, and that is where a well-designed suspension might come in.
        For me, I won’t give up the flexibly fork blades and the better modulation of centerpulls, and I accept that I have to look ahead a bit more when riding in the rain. But it’s not a clear-cut question with only one logical answer.

        January 28, 2015 at 7:20 am
      • exmaschine

        Well like most who are against the idea-implementation of discs, sorry to say but discs are coming and staying buddy. As far as suspensions go, that’s coming too. Seems you’re a tad stuck in the past…hey, I’m old school too, but a denier I am not. 😉

        January 28, 2015 at 8:34 am
      • exmaschine

        And if you read what I posted, I said, once the weight and cost issues are resolved (and they will be) You know the Luddites put up a good fight…but ultimately lost.

        January 28, 2015 at 8:36 am
  • Alastair

    4.5 bar is already a reasonably high pressure for cycle touring – I run mine at 5 bar. Most mountain bike tyres are run at 3 bar or below, and I suspect this would affect performance.
    How was power measured?

    January 27, 2015 at 1:29 pm
    • Jan Heine, Editor, Bicycle Quarterly

      These were 25 mm tires. At the lower pressures, the bike became hard to steer around the corners on the velodrome, because the sidewalls were collapsing. Power was measured with a PowerTap power meter in the rear hub.

      January 27, 2015 at 3:04 pm
    • Jan Heine, Editor, Bicycle Quarterly

      The wider tires are becoming popular not just for Paris-Roubaix. Already in 2014, François Marie from FMB told me that all the pros were ordering 25 mm tubulars, and no longer 23s.

      January 28, 2015 at 7:05 am
  • acoggan

    Why do you think your results differ from this published study?

    January 28, 2015 at 7:12 am
  • Jason Ferrier

    I’ve been running 25-30psi in my Hetres (non extra leger version) for several years now. I’m <150lbs and when I go camping with a heavier front load with two front panniers, I bump the front tire up a bit, but no more than 35psi. The 42c tires are awesome and make my cyclocross bike with 35c feel rough.

    January 28, 2015 at 10:15 pm
  • Michael

    I took the Hetres out today on massively salted but dry/damp roads.
    They looked like powdered donuts at times.
    Was wondering if salt makes for dry-slipperiness, like riding on dusty roads. They seemed to do fine.
    Also, have heard salt can be stick and make for more drag. True?

    January 30, 2015 at 8:52 pm
    • Jan Heine, Editor, Bicycle Quarterly

      I don’t know about salt, but cold rubber doesn’t flex as easily, so you have higher rolling resistance and less grip when the tires (and road) are cold.

      January 31, 2015 at 4:48 am
  • Steve

    “Tire pressure has almost no effect on a tire’s speed.”
    “There is no relationship between tire pressure and performance in the tested range.”
    I am having trouble understanding your statements above and the graph in figure 1. It seems that for every change of pressure you got a change in the power needed to run constantly at 27.9 km/h (17.3 mph). Doesn’t that inherently mean that performance is a function of tire pressure? And the lowest power requirements are at the high pressure end of the scale, what am I missing to understand your conclusion that lower pressures are better?
    I prefer lower pressures myself. A major factor in me moving to lower pressures is the Tire Drop article in BQ Volume 5, number 4. The article reinforced for me the best method of airing up tires when I was a kid in the 60s and most pumps did not have gauges. There were three types of dads in the neighborhood that would show a kid how to inflate a bike tire. The first type cluelessly added air until it popped off the rim. The second type would add some air, poke the tire with their thumb, and repeat until their thumb felt right. Then there was the dad who would add air, lean on the bike and watch for how much the tire dropped, and repeat until the tire dropped some but not too much. Using the tire drop article, I now ignore the pressure marked on the sidewall of the tire. I secure all the bags on the racks, add some air, lean on bike and check the tire drop, and repeat until a SWAG puts it at 15%. I tell myself that this is the best method because it gives a good contact patch and the sidewall is stiff enough not to collapse in normal use, but flexible enough to provide suspension.
    I would find it useful if you added some information for each tire/pressure about the size of the contact patch, and the amount of tire drop.

    January 31, 2015 at 5:59 am
    • Jan Heine, Editor, Bicycle Quarterly

      You are right that tire pressure has a (relatively small) influence on performance, but it’s non-linear and hard to predict. The differences are statistically significant, but they depend on the road surface, rider weight, tire model and maybe some other factors, so optimizing your tire pressure for the best performance will be difficult. For example, the Rubinos are fastest at 6.5 bars, yet at the same pressure, the CX tubulars are slowest!
      The good news is that compared to other factors, it’s not that important. The CX clinchers are always faster than the Rubinos… so choosing your tire well has the biggest impact on performance.
      The “dad” method of looking for tire drop when inflating the tires is a good suggestion.

      January 31, 2015 at 5:02 pm

Comments are closed.

Are you on our list?

Every week, we bring you stories of great rides, new products, and fascinating tech. Sign up and enjoy the ride!

* indicates required