Monday 21 February 2022

Do foam tyre inserts degrade rolling resistance?

Do foam tyre inserts affect the rolling resistance of a tyre?  Based on this testing then the answer is no, probably not.

If there is a penalty at all, then it's very small, at least for the tyre inserts that I tested.  Read on for more details...

Why I've been using tyre inserts

About two years ago I tested the effect of tyre pressure on rolling resistance, using my Cyclocross/Gravel bike ridden on a grass field.  That test used the Virtual Elevation (VE) outdoor method.  The tyres I used were 35mm Schwalbe X-One Allround tyres.  The results were explained in my previous blog post here.

Those results from 2020, shown in the plot on the left, really surprised me because there was no 'optimum' tyre pressure.  The red points continue the downward CRR (rolling resistance coefficient) trend towards the left hand side of the plot, for lower pressures.  I was expecting there to be an optimum pressure somewhere, that gave the minimum rolling resistance.  However, the testing showed that lower tyre pressures were always best, even at pressures so low that the tyre became laterally unstable.  Rolling resistance expert Tom Anhalt commented on a forum that other people had found similar similar results to this in the past, for MTB tyres, finding also that lower pressure was best for rolling resistance.

After this testing, I began to run lower pressures on my cyclocross bike, to exploit these rolling resistance advantages, and in doing so also gained some comfort and grip benefits.  The potential problem with running lower tyre pressures though, as all mountain bike and cyclocross riders will know only too well, is an increased risk of rim damage and pinch flats if you hit a sharp object.

Indeed, several weeks later, while I was out on a ride I double-flatted my tubeless Schwalbe X-One tyres after hitting a concealed rock.  One tyre was punctured so badly that it struggled to hold air, even with tubeless 'bacon' repair strips inserted into the tyre hole.  I just about made it back home.  One of the tyres was badly damaged and had to be thrown away.

My motivation for using foam tyre inserts, then, was to allow me to run lower pressures, to take advantage of the rolling resistance benefits shown in the plot above, but reducing the risk of rim damage and pinch flats.

I have been using foam tyre inserts for almost two years now on my cyclocross/gravel bike.  Although I've been generally very happy with them, I've always wondered in the back of my mind whether I'm paying a rolling resistance penalty by using the foam inserts. If it's a large penalty, it might actually cancel out the benefit of running lower pressures.  That's why I wanted to do this testing.


Available Data

As far as I'm aware, there has been very little testing of rolling resistance done and published for the commercially available tyre inserts, which is a shame, and this is the reason I've done this testing.

Testing of tyre inserts, especially for mountain bike applications, usually involves only a qualitative assessment of the effect of tyre inserts on the ride feel, tyre stability, tyre damping and rim protection.  Some quantitative testing was done by Pink Bike last year, but that testing was limited to evaluating rim impact protection only.

Testing of road tyre inserts has been done by Aerocoach and Bicycle Rolling Resistance, both concluding minimal impact on rolling resistance, but tyre liners for road tyres is not something that interests me, because for road use there isn't the same benefit or desire to run very low pressures.

To be honest, I find it a little frustrating that tyre insert companies don't provide rolling resistance data.  Often they don't even address the rolling resistance question at all.  This is perhaps understandable if downhill mountain biking is the target market, but for gravel and cyclocross applications, like this this one or this one, then I'd like to see rolling resistance data provided by the tyre insert companies.

I'm also really hoping that Bicycle Rolling Resistance will do some tyre insert testing for MTB and gravel inserts soon.  It's on their voting list, but far from the top of the list at the moment.  Tyre inserts aren't cheap, so I can't justify buying a variety of brands and testing them myself.



My tyre and foam tyre insert set-up

I previously used 35mm Schwalbe X-One tyres for my outdoor rolling resistance testing.  However, in 2020 I swapped to 
43mm Panaracer Gravel King SK TLC tyres after I damaged one of those Schwalbe tyres in the double-flat incident. Panaracer Gravel King SKs get good reviews and are a more  summer-focussed tyre, which suits my riding.  Also, they test quite well in terms of rolling resistance, based on measurements made by BicyclingRollingResistance.com.

The foam tyre inserts I bought were from Planet X, called "Barbieri Anaconda Puncture Protection System".  These are 
no longer available to buy from Planet X, but they are budget round-section 'pool noodle' type closed cell foam inserts.  They cost £19.99 including valves.

At the time of installing them, I was conscious that the rolling resistance might suffer, so I trimmed the top of the round section off, as shown in the photo to the left.  The aim was to provide more of a gap between the outside of the foam insert and the inside of the tyre, trying to prevent the tyre from touching the foam insert at the contact patch.  Through some tyre drop measurements, I worked out that the inside of the tyre would touch the insert only at pressures below 14-16 psi on a flat surface, which corresponds to a tyre drop of around 15mm.  However, it should be noted that the same 14-16 psi tyre pressure should result in a larger tyre drop when it's on the rollers though, because their small roller diameter exerts a higher localised load on the tyre at the contact patch.

The photo below shows some quick (rather crude) measurements of the tyre external depth and also the depth of the foam tyre insert.



The width of the tyre and the foam insert are shown below. The foam insert is fairly narrow, at 30.4mm, considering the rim has an internal width of 23mm and an external width of 28mm.  Nevertheless, when riding over rocks and other obstacles, I felt qualitatively that the foam inserts were being compressed whenever I hit obstacle that would have otherwise caused the tyre to bottom-out.  It didn't feel like the usual tyre-on-rim contact that you get without tyre inserts installed.  



Test method selection

My test method was the same roller testing, as I described in my previous blog post.  Why roller testing though?  Virtual elevation (VE) testing would have been suitable too, based on my previous experiences, but VE testing is significantly less convenient:
  • VE testing takes much longer to perform, several hours as opposed to about one hour for roller testing.
  • VE testing needs good conditions: Low wind days are needed, and obviously also daylight, which is more difficult in winter.
  • For VE testing outdoors, the ground needs to be fairly robust, so that multiple runs don't degrade the ground and lead to a drift in the rolling resistance being measured. This generally means that for off-road applications, the ground needs to be dry.  Again, that's difficult or impossible in winter (in the UK).
In its favour, VE testing does however provide the complete picture of all three rolling resistance losses: (1) Tyre hysteretic losses; (2) suspension/impedance losses; (3) ground deformation hysteretic losses.  Roller testing on the other hand will only provide an indication of #1, the tyre hysteretic losses.  However, for foam tyre inserts, I felt that this was the primary mechanism by which inserts might cause additional losses. Therefore, I decide that roller testing was the best method for this situation.


Test method description

I started with the foam insert installed in the tyre, performed the tests with various pressures, removed the foam insert, then repeated the tests with various pressures again without the insert in place.

The rear wheel weighing was done using my front fork mount, by lifting the bike off to the side of the rollers and onto the bathroom scales.  As before, timber blocks were use to ensure the bike was horizontal while on the scales, thus ensuring the scales weighed the same rear wheel weight as the weight that's on the rear wheel for the rollers.

For the roller testing, I performed a few repeats throughout the test, three specifically, which helped give me a feel for the repeatability of the results from this testing. The repeat runs are included in the results shown below. 


Results

The test results are shown in the plot below, with the green and blue symbols.  The dashed lines are best fit lines put through the data.  Data from Bicycle Rolling Resistance and from Tom Anhalt's testing is also shown in the plots, although it should be noted that for those tests there are differences in the inner tube being used and also the tyre width.  

Bike tyre rolling resistance coefficient (CRR) testing using rollers and effect of foam tyre liners on rolling resistance


The raw data is shown in the spreadsheet screenshot further below:




The plot above shows that the effect of the foam tyre insert is small and possibly nothing at all.  If there is an effect of the foam insert on rolling resistance, it's close to the precision and the repeatability of what can be tested I think, so less than about 2-3 Watts at 25 kph.

I was expecting a larger difference at low pressure, where the foam insert might have been getting compressed.  However, even at 15 psi, the differences are small.  The best fit lines suggest there is a larger difference at 15 psi, but the green and blue data points aren't a lot further apart at 15psi than they are than elsewhere.  Additional testing to collect more points (repeats) would be needed if I wanted to really confirm that.


Conclusion and caveats

In any case, the conclusion is that the effect of these foam tyre inserts is small at the most, and possibly nothing.

There is one final caveat to mention though: When I removed the foam insert I noticed that the tape holding the two ends of the foam insert had come apart and so the two ends of the foam insert weren't attached to each other.  I don't know when this happened, but it means that the foam insert was probably 'free floating' inside the tyre cavity, rather than being a continuous foam hoop held in contact with the rim and tyre bead.  In a way, it would have been behaving rather like a Huck Norris foam insert, but possibly with even less contact with the tyre, owing to it's small diameter.  I don't know if this has an effect, but it's important to note.  I didn't have the time or inclination to repair the foam insert (to clean and re-join the ends), re-test all the pressures, and then remove the tyre again to check it was still okay.  A free-floating foam insert might possibly cause lower rolling resistance losses, because it is not contacting the tyre side wall near the rim.  However, at this point I can only speculate. 










Saturday 19 February 2022

Testing of tyre rolling resistance using rollers - Part 2, Setup improvements

Following on from Part 1 of the roller testing described in the previous post, I made a couple of important improvements:

Speed Sensor: In my Part 1 blog post, I explained that I didn't yet have the magnetic Garmin speed sensor that I'd ordered.  I subsequently received the speed sensor through the post and installed it in a similar way to how Tom Anhalt did, setting the 'wheel circumference' on my Garmin head unit to 261mm, which is the measured roller diameter of 83.0mm, multiplied by pi.

I'm still a bit surprised the speed sensor works with the magnet triggering the sensor so quickly, at about 40-50 Hz (every 2-3 hundredths of a second!).   

Front fork mount: Previously I had to prop myself upright using my elbow.  I built a fork mount using a spare piece of chipboard flooring and a few spare bits of timber.

The axle itself is the axle clamp borrowed off my Thule 561 bike carrier (the 561 is now discontinued). 

After making these two improvements, I wanted to check the effect on the rolling resistance measurements by re-doing the runs at 80 psi, using my road bike and Continental GP5000 tyre, as tested in Part 1.  The results below (shown with blue triangle symbols) show that these two set-up improvements have quite a small effect on the results.  It's quite reassuring that one week later, with some tweaks to the setup, I get very similar results to the previous weekend.






Sunday 13 February 2022

Testing of tyre rolling resistance using rollers - Part 1

 

Bicycle tyre rolling resistance coefficient (CRR) testing using rollers
A previous blog post from 2020 described my measurements of tyre rolling resistance that I did using the Virtual Elevation outdoor method.

Tyre rolling resistance coefficients can also be measured using rollers instead, and I've been keen to try that method ever since reading about it a few years ago. However, I've never owed a pair of rollers, so that test method hasn't been possible up until now.

Recently though, there have been a couple of things I really wanted to test, which I'll describe in a future blog post.  That finally gave me the impetus to buy a pair of rollers.  I bought a pair JetBlack R1 rollers in the Evans Cycles sale for £100.


Overview of available rolling resistance testing methods

As a brief introduction, I want to first quickly describe the four methods for measuring bicycle tyre rolling resistance that I'm aware of:

1) Drum testing: This involves turning a tyre and wheel on a large drum and measuring the power required to turn the drum, to overcome the losses due to tyre rolling resistance.  An example of this kind of testing can be found on bicyclerollingresistance.com.

2) Roller testing: This is similar in principle to drum testing, but is more accessible to amateurs that don't have access to specialised equipment necessary for drum testing.  All that's needed is a standard set of rollers, a power meter, and a few small pieces of equipment (which will be described later).  An example of roller testing, which is also the exact method I used, can be found here.

3) Virtual Elevation testing: This is another technique accessible to amateurs, as long your bike has a power meter.  It is more time consuming than roller testing, but has other advantages.  My previous post here describes how I measured the rolling resistance coefficients for my cyclocross tyres on a grass surface, using this method.

4) Roll down testing: This is another technique accessible to amateurs and is also relatively quick to perform.  Unlike methods 2 and 3, it does not need a power meter.  This roll down method can therefore be very appealing to people, especially as it gives a direct speed/time benefit, needing no data processing.  However, many experts express reservations about the sensitivity of roll down testing to detect changes in rolling resistance.  For example, the runs are generally performed on the same hill in one direction and therefore the results are affected by small, indiscernible fluctuations in wind.  VE testing (method 3) mitigates this sensitivity to the wind, partly at least, by performing laps that go against and also with the wind.


Roller testing method description

One of the pioneers and experts in tyre rolling resistance testing is Tom Anhalt.  For simplicity, I followed Tom's roller method as closely as possible, as described in his blog here.  Since the description on Tom's webpage is already very clear, I won't repeat it here.  Instead, I'll only describe where I deviated from his method, or one or two other interesting points:
  • I used Tom's gravel tyre data recording and processing spreadsheet pretty much as-is.  I saw little value in re-working the spreadsheet, or creating my own version.  If I did so, I would only risk making mistakes.  Only one change was needed, and that was the diameter of the rollers:  Mine were 83mm in diameter whereas Tom's were larger, 114mm in diameter.
  • I didn't have an old-fashioned Garmin magnetic speed sensor when I did this initial testing.  I had already ordered one from a seller on eBay, but I was still waiting for it to be delivered.  Instead I had to use my (newer) Garmin wheel hub based speed sensor, which I think works using an accelerometer to detect rotation frequency, in combination with a prescribed wheel circumference.  This is a simplification, because keeping the wheel circumference fixed isn't quite right when the tyre pressure is changing.  For this initial 'shakedown' testing, though, I think this simplification is fine.
  • I didn't have a front fork mount.  I later constructed one (to be described in a future Part 2 blog post), but for this initial test I had to improvise.  Instead, I positioned the rollers next to a shelf (see photo below) and used my elbow to keep myself upright.  This really wasn't as bad as it sounds!  It's far from ideal though.
  • For this testing, and for everything else, I used a Stages left-hand crank based power meter.  Using a single-sided power meter is inferior to using a dual-sided power meter, but it's all I have unfortunately.  At some point, I will invest in a dual-sided power meter.

Measuring weight on the rear bike wheel
Rear wheel weighing.  I asked my wife to read off the number on the bathroom scales while I sat on the bike.

The wooden block under the front wheel was there to ensure the bike was level, which I checked using a 6ft spirit level spanning the front and rear quick release skewers.  A similar level check was done for bike on the rollers, for the actual roller testing.






Bike tyre rolling resistance coefficient (CRR) testing using rollers
The photo on the left shows my initial roller testing setup.  The setup was improved in subsequent weeks, but this is how it was for these initial tests.











Results

In view of the simplifications explained above, I didn't want to invest too much time doing a lot of testing initially, so I spent just half an hour or so collecting and processing some data.  I followed Tom Anhalt's method exactly, so used a 5-minute warm up period, and a 4-minute 'run' period, then processing the data from the last 2 minutes of those 4 minutes to get average power and speed values.

I tested my 23mm Continental GP5000 tyre at different pressures.  Results are shown below, comparing my results.  I have shown screenshots of the spreadsheet for full transparency of how I arrived at the values in the plot.

Setup and Analysis Constants

Speed, Power and CRR derivation spreadsheet


Bike tyre rolling resistance coefficient (CRR) testing using rollers

Rolling Resistance Coefficient (CRR) plot
 

The results shown in the plot above are quite encouraging I think.  Considering the small simplifications in my test (i.e. the lack of an appropriate speed sensor and lack of a front axle support), I think the results look reasonable.  The trend and values are similar to the Bicycle Rolling Resistance data (orange points), although my CRR values should actually be lower than the BRR values, instead of higher, because my testing was done with a latex inner tube whereas BRR's testing is done with standard butyl tube.

Comparisons versus Tom Anhalt's data point (the black circle in the plot) should be a reasonable like-for-like comparison, with differences coming possibly only from the roller diameter, brand of rollers, and power meter differences.  My CRR values are about 30% higher, which is quite a big difference.  I can imagine a few possible explanations, but as this was the first time I've used my rollers, I wonder if the roller bearings need 'running in', which would reduce their friction losses and would therefore reduce the apparent CRR values. This is something that should become more clear if I do further testing.

On the positive side, my repeated point at 100 psi tyre pressure for the 4th run shows excellent repeatability with the equivalent point for the 1st run.  The two data points at 100 psi are barely discernible on the plot because they are so close to each other.

For my next set of testing, I will probably use my gravel bike and try to measure the effect of foam tyre inserts on the rolling resistance coefficient.  This will be documented in a future blog post.