Cycling UK Forum

A frequent topic of debate on this forum is the effect rolling resistance of tyres and the related questions:
* Are wide tyres faster than narrow tyres?
* What is the optimum tyre pressure?

There is of course a load of data on the Bicycle Rolling Resistance website, but their test method measures rolling resistance of the tyre in isolation and so is really a measure of "casing losses" only. The additional effect of the road surface roughness upon the bike and rider as a whole, sometimes called "suspension losses", is not measured by the BRR method.

Ever mindfull of Lord Kelvin's "When you can measure what you are speaking about, and express it in numbers, you know something about it..." I have attempted my own "real world" measurements of rolling resistance. After a year of on/off evolutionary development I finally have a working test method.

First, a few notes on the scale of the problem. If you start freewheeling on a level road at an initial speed of 20km/h, the next 100m will take 23.885 seconds (rider + bike = 80kg, Crr = 0.005, Cd.A = 0.400). If the rolling resistance of the tyres is increased by 10% to 0.0055, the time will increase by just 0.390 seconds. It will not take much of a headwind to produce the same decrease in time. Clearly accurate measurements and controlled conditions will be neccessary for meaningful results.

I started by constructing a custom data logger using a microcontroller development board from Texas Instruments. I added a Hall effect sensor and wrote the software for this so that it measures the time for each revolution of the front wheel with a resolution of 30us.



The test method is then to roll down an incline whilst recording the time for each revolution of the front wheel using the data logger.
To find the rolling resistance, a set of equations is constructed which describe the energy of the closed system over time, based on the ideas proposed by Robert Chung. The equations have a number of inputs all of which must be measured, calculated or estimated. The output of the equations is the rolling resistance Crr.

This method is not without its drawbacks, the principle one being the requirement for zero wind. The incline is also important: to measure Crr it has to be just steep enough to keep rolling but not too quickly: about 1% gives good results. A steeper incline is used in the same way to determine a value for aerodynamic drag CdA. The height of the incline was measured with a barometric altimeter: not accurate enough really, but all I have available.

The first results were obtained using Continental Grand Sport Race tyres (28mm front/32mm rear) on a smooth road surface.
First a roll down a steep incline was done to get an unimpressive value for Cd.A of 0.520m^2. To double check this I took a photo of myself on the bike head-on with a metre rule for scale and counted the pixels with a photo editing program: my frontal area came out at 0.593m^2. As the coefficient of drag Cd is typically around 0.8 to 0.9 for a cyclist, my measured Cd.A should be in the range 0.415 to 0.534. I used a value of 0.500 for subsequent calculations.


To measure rolling resistance, two runs were done at four different pressures down a gentle incline on a smooth road surface. The results are shown in the following graph, with the min/max indicated by the error bars. The absolute values cannot be guaranteed, but the shape is indisputable. As can be seen from the graph below, rolling resistance decreases as the pressure increases from 60psi to 80psi but then markedly increases at 90psi. The 80psi runs were done on a different day when the wind was not quite zero. Harder is not always faster!



Gratifyingly, this result replicates the findings first published by Tom Anhalt in 2009, see https://www.slowtwitch.com/Tech/What_s_in_a_tube__1034.html

As a data quality check, the calculated elevation profile for each run was plotted on the following graph. They all line up pretty well.



So, what next? I'd like to test some wider and narrower tyres. However I have a very limited collection of odd tyres to choose from and I
cannot afford to buy a shed-load of tyres purely for testing. Voyager Hypers are popular with some on this forum and I have one in size
37-622. Is anyone interested enough to lend me another?

[Edit: updated graphics]

One factor that never seems to be considered with rolling resistance is that softer tyres bounce more than harder tyres when you pedal hard. Pedalling affects the ride. I'm not convinced that any measure that takes place while freewheeling tells you about what is going on while pedalling.

pedalling also affects the stresses (and therefore the losses) in the carcass too. I don't know of any test that allows for this in a meaningful way, except perhaps if you have a real rider riding up hills whilst using a sufficently accurate power meter and 'control' runs are done at the same time to allow for temperature and wind effects on the day. [I have noted many times that I am much slower up hills on fatter tyres, regardless of the purported Crr value]


BTW there is some allowance made for road roughness effects; the drums used in the tests reported in rolling resistance.com are clad with chequer plate. This appears to give Crr values that are comparable to those found on typical tarmac.

BTW if you think suspension losses are a signficant proportion of the whole, then you may get errors arising from other things (say) small variations in how you hold the handlebars.

Nice electronics you have assembled though!

cheers

Congratulations on the electronics and attempting to measure this sort of thing in real world riding.

However, I wonder if a different approach could be taken, using standard (if expensive) bike devices. If the power input from rider is constant (measure with power meter) and the road/conditions are the same, riding position the same, then the primary difference is the tyres. So, ride different tyres at the same power meter reading over the same road and measure the time to complete the distance. The YouTube channel "Global cycle network" has done some reports this way.

Brucey wrote:pedalling also affects the stresses (and therefore the losses) in the carcass too. I don't know of any test that allows for this in a meaningful way, except perhaps if you have a real rider riding up hills whilst using a sufficently accurate power meter and 'control' runs are done at the same time to allow for temperature and wind effects on the day. [I have noted many times that I am much slower up hills on fatter tyres, regardless of the purported Crr value]

Nigel wrote:However, I wonder if a different approach could be taken, using standard (if expensive) bike devices. If the power input from rider is constant (measure with power meter) and the road/conditions are the same, riding position the same, then the primary difference is the tyres. So, ride different tyres at the same power meter reading over the same road and measure the time to complete the distance. The YouTube channel "Global cycle network" has done some reports this way.

Yes, rolling resistance can be measured by cycling laps of a course using a power meter, this was first proposed by Robert Chung almost 10 years ago. However, I am not about to "invest" (!) in a power meter any time soon.

Brucey wrote:BTW there is some allowance made for road roughness effects; the drums used in the tests reported in rolling resistance.com are clad with chequer plate. This appears to give Crr values that are comparable to those found on typical tarmac.

BTW if you think suspension losses are a signficant proportion of the whole, then you may get errors arising from other things (say) small variations in how you hold the handlebars.

Nice electronics you have assembled though!

cheers

The problem with the drum tests is not the roughness of the drum, it's the load. A big lump of steel is stiff, but elastic, whereas a human body is neither stiff nor elastic and therefore absorbs much of the energy from vibrations. This does not happen in a drum test with a steel load.

There is a method which is not scientific but takes everything into account in one. Average time over a known course. I cycle relatively easy when I cycle alone. More or less the same average effort. This shows by the fact that my average mph on a particular course will be fairly standard on a particular bike with particular tyres at a particular pressure.

It's when you change something and there's a change in the average mph that you notice.

I cycle a fair bit in Spain with Spanish friends. I noticed that on a 26in wheeled MTB that I had problems going downhill with them on road in that they effortlessly went away from me on their 29ers. When I got a 29er, the problem disappeared. What amazed me was that my 29er had knobbly tyres, as their had, BUT the tyres had a very flexible carcass, unlike my 26in tyres. So, the old saw that tyre flexibility is very important was borne out.
Here in Ireland I have a 29er with the narrower Schwalbe Big Apple fairly slick tyres and they are slower than the flexible knobblies the bike came with. I use them for puncture resistance. How do I know? Average speed.

Scunnered wrote:

Brucey wrote:BTW there is some allowance made for road roughness effects; the drums used in the tests reported in rolling resistance.com are clad with chequer plate. This appears to give Crr values that are comparable to those found on typical tarmac.

BTW if you think suspension losses are a signficant proportion of the whole, then you may get errors arising from other things (say) small variations in how you hold the handlebars.

The problem with the drum tests is not the roughness of the drum, it's the load. A big lump of steel is stiff, but elastic, whereas a human body is neither stiff nor elastic and therefore absorbs much of the energy from vibrations. This does not happen in a drum test with a steel load.

I agree, but the mass of a real rider is variable and the coupling/damping to the bike is variable too (hence my comment about holding the handlebars differently).

I expect that if you imposed a particular (specified/controlled) rider mass/coupling into tests, you might end up choosing tyres around things like their resonant qualities etc

cheers

mad thought; the rolling resistance forces on the tyre are certainly not insignificant at reasonable load, and ought to be easy enough to measure, provided these forces can be isolated. The main problem is that the aero forces on the bike are also large (and getting larger quickly) at speed. They are also very sensitive to wind speed.

Having given this a little thought, I wonder if there is a better scheme.... The idea that I'm thinking of at present is that you have a trailer with a heavy weight on it, running on the test tyre(s). The (instrumented part of the) trailer has a very low Cd value, and indeed could be sheltered by a special fairing on the test bicycle that is towing the trailer, or the trailer itself, so that (over a modest range of wind conditions) the aero drag on the trailer is roughly zero, or at least consistently low. The data to be collected would be an average load in the towing coupling, or the axle coupling within the trailer.

One method of doing this test would be to ride on a road with a slowly changing gradient; at some point (probably about 0.5% downhill gradient) the load in the coupling will fall to zero. The coupling could be set to give a clear indication of this state, perhaps with a damper in the coupling to ensure that the indication doesn't vary with every tiny bump in the road.

Testing logistics would be great with this setup; all that would be required is that you go out with a trailer full of test wheels; the test wheels can stay in the trailer (regardless of the wheels being tested) since this would keep the total load constant. The trailer could be configured to use a single front wheel or a pair of wheels perhaps.

This method would allow tests at different temperatures, road surfaces, tyre pressures, loads and load coupling. Any set of measurements can be quickly compared to a 'control tyre' at any time.

Probably there is something obvious that I have not thought of....?

cheers

Brucey wrote:mad thought; the rolling resistance forces on the tyre are certainly not insignificant at reasonable load, and ought to be easy enough to measure, provided these forces can be isolated. The main problem is that the aero forces on the bike are also large (and getting larger quickly) at speed. They are also very sensitive to wind speed.

Having given this a little thought, I wonder if there is a better scheme.... The idea that I'm thinking of at present is that you have a trailer with a heavy weight on it, running on the test tyre(s). The (instrumented part of the) trailer has a very low Cd value, and indeed could be sheltered by a special fairing on the test bicycle that is towing the trailer, or the trailer itself, so that (over a modest range of wind conditions) the aero drag on the trailer is roughly zero, or at least consistently low. The data to be collected would be an average load in the towing coupling, or the axle coupling within the trailer.

Not mad at all and I have seen papers describing such a setup, usually university projects. As I recall, the main issue is the signal to noise ratio of the measurement of towing force. A real road always has lumps and bumps and there is always turbulence from the towing vehicle which has non-constant speed. Couple that with the tiny force being measured and you can see the problem.

Scunnered wrote:So, what next? I'd like to test some wider and narrower tyres. However I have a very limited collection of odd tyres to choose from and I
cannot afford to buy a shed-load of tyres purely for testing. Voyager Hypers are popular with some on this forum and I have one in size
37-622. Is anyone interested enough to lend me another?

I've got various used tyres from 20mm up (I did have more but fitted some of the narrow ones as rim protectors when selling wheels), whereabouts are you?

Trailers "tug" as you pedal to a greater of lesser degree.
It would be better if the trailer was towed smoothly behind a motorised vehicle. Gurney perhaps?

The utility cyclist wrote:

Scunnered wrote:So, what next? I'd like to test some wider and narrower tyres. However I have a very limited collection of odd tyres to choose from and I
cannot afford to buy a shed-load of tyres purely for testing. Voyager Hypers are popular with some on this forum and I have one in size
37-622. Is anyone interested enough to lend me another?

I've got various used tyres from 20mm up (I did have more but fitted some of the narrow ones as rim protectors when selling wheels), whereabouts are you?

Hi, thanks for your offer. This thread is a year old and I convinced myself that wider tyres ~32mm at moderate pressure were at no significant dissadvantage to narrow tyres. I have moved on to other projects now.

Mick F wrote:Trailers "tug" as you pedal to a greater of lesser degree.
It would be better if the trailer was towed smoothly behind a motorised vehicle. Gurney perhaps?

yes, I've got some ideas about that; one of them is that if the tests are carried out on sloping roads, at some point the 'gravity engine' will be just right to overcome rolling resistance. Under these conditions the only pertubation in the system will be caused by the road surface, and damping ought to be possible so that the measurement can be taken accurately.

Regarding tyre choice; I agree in general terms that the rolling resistance of wider tyres is not greatly increased vs narrower ones. If the roads are bad enough you are better off with wider tyres for most forms of 'normal riding'. Years ago you couldn't easily get wider tyres that were built well; they usually had heavy coarse carcasses and thick, squishy tread; those tyres were slow. However now you can more easily source wider tyres that are built better than that.

The penalties for wider tyres are a (small) weight increase and an increase in the aero drag. The proportion of total drag that is aero drag varies with road speed, so the balance of tyre choice will change with envisaged speed; in extremis if you are TTing at 25-30mph it may be worth taking a significant hit on Crr if it gives any measurable benefit in aero drag. At low speeds any Crr benefit is more likely to outweigh an aero deficit. However there is a lot of territory inbetween and that is where much cycling is done; it is very much a live question here.

cheers

Scunnered wrote:

The utility cyclist wrote:

Scunnered wrote:So, what next? I'd like to test some wider and narrower tyres. However I have a very limited collection of odd tyres to choose from and I
cannot afford to buy a shed-load of tyres purely for testing. Voyager Hypers are popular with some on this forum and I have one in size
37-622. Is anyone interested enough to lend me another?

I've got various used tyres from 20mm up (I did have more but fitted some of the narrow ones as rim protectors when selling wheels), whereabouts are you?

Hi, thanks for your offer. This thread is a year old and I convinced myself that wider tyres ~32mm at moderate pressure were at no significant dissadvantage to narrow tyres. I have moved on to other projects now.

oops, didn't see that

Brucey wrote:Probably there is something obvious that I have not thought of....?

You are still neglecting "suspension losses", which would render any rolling resistance vs tyre pressure relationship unrealistic.

You want a trailer load that will absorb significant amounts of energy as it is vibrated, and which can be standardised (which rules out using an actual person for comparions over a longer period).
How about a tank of water mounted on a vertical track and supported by stiffish springs?
You could add flat damper plates inside the tank (from above to avoid seal problems) if you want extra energy absorption. Test run comparisons with a real person should get you in the right ball park.