Technical FAQ: Tire rolling resistance testing methodology

Published December 14, 2021 06:35AM

Have a question for Lennard? Please email him at veloqna@comcast.net to be included in Technical FAQ.

Dear Lennard,
I recently upgraded the road wheel-set on my gravel bike so my days on tarmac are a bit more enjoyable. I keep two wheel-sets for now until I procure a sick dedicated climber or lightweight road bike. The LBS manager who is a bit of a wheel/tire head was quite passionate that I try the Rene Herse 44mm extralight citing all sorts of real-world benefits especially on a gravel bike designed for wider tires. While I definitely don’t think I was upsold, I’ve been left feeling strange and wondering if I’d be better off with 28s or 32s (I ordered some Conti GT5000 to compare) but he recently sent me this rather compelling article to calm my anxieties.
— Evan

Dear Evan,
This is a very good question, and one about which there is still room for debate. At least we’re having the debate; until maybe 20 years ago, riders only believed that narrower tires at high pressure were faster, because they feel fast, thanks to a lively (and energy-robbing) bounce on every little pebble in the asphalt. For your road-specific wheelset, I think you might want to trust your intuition.

In our rolling-resistance test of gravel tires on a simulated cobblestone road, bigger tires generally were not faster than narrower ones. In the six cases where we had multiple tires of the same model in multiple widths, the narrower tire was faster in four of those cases. In one of those cases, we even had three widths of one tire (the WTB Byway TCS tubeless tire), and its rolling resistance increased with each increase in width.

Of the two tire models where the larger width was faster, one result could be questioned based on construction differences, and the other the difference was slight. The Specialized Tracer Pro TLR 42C was faster than the 38C width of the same model. However, the 42C Tracer Pro TLR has no puncture-protection belt, while the 38C Tracer Pro TLR does; the belt adds stiffness that will slow the tire down relative to the one without. The other tire where wider was faster was the Panaracer Gravel King SK TLC, but the 38C width required only 0.1W more than the 43C to roll it on the faux-cobblestone test drum at 35kph; that’s essentially dead even.

On the much narrower tires in our Paris-Roubaix rolling-resistance test, tested at the same speed, with the same load, on the same simulated-cobblestone surface on the same test drum as the gravel tires, the wider tire was faster in all three instances where we also had a narrower width of the same tire model.

Since all the tires in both tests were run identically, I combined the two sets of results. The fastest tire was the 28mm tire that won the Paris-Roubaix test, while the second-fastest tire was the 36mm tire that won the gravel test. I find it amazing how narrower tires from the Paris-Roubaix test mix it up with the wider tires from the gravel test on a very rough surface. Maybe I’ll do an article sometime combining them…

And if there is a question about whether a fatter tire is faster than a skinnier one on a rough road, where the bigger tire should have the advantage, then on a smooth road, the narrower tire will likely come out ahead. And even if there is little or no difference in rolling resistance, the advantage will go to the narrower tire overall, due to its lighter weight and lower aerodynamic drag. Contrary to that René Herse blog you sent me, bigger tires are slower aerodynamically, except when the rim is wider than the tire.

As for the René Herse tire rolling-resistance results that your LBS manager turned you on to, those must be taken with a grain of salt. This is the methodology the author (Jan Heine) employed for those René Herse tire tests. I respect the enormous amount of work, time and effort that went into those tests. On the other hand, you can’t accurately quantify a small friction difference between tires, tire widths, or tire pressures when the main thing you’re actually measuring, namely aerodynamic drag, dwarfs those tiny differences.

When riding a bike on flat ground at 40kph or higher speeds, wind resistance accounts for over 80% of the drag forces the rider must overcome to maintain that speed. Of that remaining 20% or less, tire rolling resistance is the largest component, the others being friction in the drivetrain and bearings. If you’re trying to measure a small change in a piece of that remaining 20% of the friction, and you’re doing it by measuring how fast the rider rolls from point A to point B when wind resistance is the vast majority of the friction, it’s a very blunt measurement. It’s akin to trying to measure how much a layer of fingernail polish adds to the thickness of your fingernail by measuring the thickness of your entire fingertip and fingernail with and without the polish on it—with a ruler.

Many years ago, I tried to eliminate some of the variability with wind and temperature inherent in outdoor roll-down tests of rolling resistance. Two riders started at the same time, far enough separated that one was not drafting the other, and I measured the time difference between them at the bottom of the hill with a super-accurate laser beam timer. That’s a decent method if you’re trying to measure differences in aerodynamic drag, as I did in this test. But for measuring small differences in rolling friction with different tire pressures, it’s not. Even though I may have taken out some variability, I’m still left with the fact that the main thing determining the riders’ speeds was wind resistance, since rolling resistance is so much smaller of an effect. That’s why, if we really wanted to be able to quantify what tire pressure, width, or model was faster than another, we needed to do it in a lab, and it couldn’t be just any lab, as I explain in this answer to Tim.

Heine claims that his tire test results are statistically significant because he does multiple runs, averages the times, and then does statistical analysis on the measurements. Problem is, using statistical analysis works only if the errors in each test measurement are random. If the errors are built into the measuring method, statistical analysis isn’t going to fix it.

For your purposes in road riding, I think a 44mm tire width might be overkill, but maybe not too much. I have three sets of wheels for my gravel bike and generally ride 36mm Challenge Strada Biancas on the road. They are both fast-rolling and comfortable, and they don’t penalize me if I head off onto rough roads. If I were road racing, I would be on 28mm. And for a bit more comfort, similar rolling speed with minor weight and aerodynamic penalties, I’d go with 32mm. That’s a long way from the 23mm tires I raced on at 130psi in the 1970s and 1980s.
― Lennard