Like many on this forum I have been using Tufo tubulars for sometime. In fact, most of my miles in 2005 have been on a set of Reynolds Cirro SV with Tufo S33 Specials taped with the Tufo Extreme tape. All of the recent talk of poor rolling resistance with Tufo tires was a surprise to me. I was unable to find any solid data on the Tufo rolling resistance on the web with the exception of a German Magazine article that I can’t read. As such, last week I solicited recommendations from this forum for a simple test. In my opinion the best suggestion was a roll down test. On Saturday morning armed with a piece of chalk, 3 sets of wheels, 4 sets of tires, a GPS, and a tape measure I set off to conduct the best experiment I could perform with my limited resources. I am aware that my test has many shortfalls and uncontrolled variables so take it for what it cost you.
The Venue
I selected a quite street 1 block away from my home. The asphalt road was re-paved last year and is in very good condition. I selected this road because it has a slight <1% declined followed by a long flat section. I parked my car at the crest of the road with my wheels and using my chalk I marked the starting line on the road. I then marked X’s every approximately 75 feet to assist in maintaining a constant line. During the hour it took to complete all of the coasting runs the temperature was a constant 62 degrees, overcast with zero wind.
The Bike
I used my Trek 5200 with traditional road bars because it would be easier to handle at the slow starting and finishing speeds. The chain was also removed to eliminate any cassette drag variables. Finally I wore running shoes so that I didn’t have to manage clipping and unclipping at low speeds and to make it easier to run the bike back to the top of the street (remember no chain).
The Tires and Wheels
As mentioned I had 3 sets of wheels and 4 sets of tires. All wheelset were inspected to insure that the bearings were rolling smoothly. All tires were inflated to 115 psi. The wheelsets were: Carbon Reynolds Cirro SV with Tufo S33 special tires mounted with extreme Tufo tape. This is a low profile (23mm) carbon rim with hidden nipples laced to super smooth White Industries LTA hubs with 20 spokes front, 24 spokes rear. (tubular) Mavic Ksyrium SSC with Michelin Pro Race 23mm tires. (clincher) Mavic CXP33 rims with Campagnolo Record 10sp hubs laced with 32 Revolution spokes front and 32 Competition spokes rear. (clincher) Mavic Ksyrium SSC with 23mm Specialized Mario Cipollini tires. (clincher)
The Coasting Runs
A coasting run would start with me leaning against my car with the front wheel on the start line. I would “let go” not “push off” of the car and start down the road careful to hit each chalk mark along the way. My legs stayed steady with the cranks at 12 and 6 o’clock, my elbows where locked and I looked strait ahead. When the bike would no longer move forward I would place a foot on the ground and mark the location of the front wheel. Each wheel and tire was run 3 consecutive times with the exception of the Reynolds/Tufo wheels which were run a fourth and fifth time at the end of the test. Two runs where not marked because a car drove by which could have effected the test results. At the conclusion of the tests I used the GPS to measure to one of the middle finishing spots and then measured all of the other marks with a tape measure from that fixed location. This would reduce any measurement error.
Average Distances (meters)
S33/Rey 233.56
CXP/ProR 263.58
Ksy/Cipo 254.14
Ksy/ProR 265.63
Discussion
Clearly there was a difference between the different tire and wheel combinations. One of the surprising things (or at least to me) was the relative tightness of the distribution for a given tire and wheel combination. The obvious question is “are the differences in the means statistically significant at any reasonable confidence level”.
Well it has been a few years but I dusted of my student copy of SAS and imported the data. The following graphic should do a reasonable job helping everyone visualize the data. The “Green Diamonds” represent the 95% confidence intervals for the mean of any given tire/wheel combination. It is clear that the means of the Pro Race tires mounted on either the CXP33 wheels or the Ksyriums are not statistically different. That is to say the if we conducted enough runs we could well find out that the true mean distance that this tire/wheel would roll is the same. Further more, the “Green Diamonds” of the Ksyrium/Cipo combination are just touching CXP33/ProRace which is to say that the means could be the same within the 95% confidence intervals. I didn’t bother to rerun the model at a lower confidence level but I suspect if we were to drop the confidence level down to 80% the means would comeback statistically different. Finally, it is obvious that the Reynolds/Tufo combination is statically different from any of the other tire/wheel combinations tested at a 95% confidence.
With that out of the way what did I really answer? Well I know that my fancy carbon wheels with my taped Tufos don’t roll for shit compared to my basic 32 spoked Mavic CXP33’s with Pro Race tires. What I don’t know for sure is the cause of the poor rolling. Is it the White Industries LTA hubs? I doubt it because they sure spin smoothly on the truing stand. Is it an aerodynamic difference? Unlikely because the Reynolds have 12 fewer spokes in front and 8 fewer spokes in the rear as well as hidden nipples. Is it the Tufo tires? Is it the Tufo tape? I don’t know for sure but this data does converge with some of the antidotal evidence recently posted on this forum.
Next Steps
I have a new pair of Vittoria EVO CX tubulars that I am going to glue on the Reynolds rims and test. It should be interesting to see if the distance differs from the Tufo runs. When I conduct the next test I will run the CXP33/ProRace combination to provide a baseline for atmospheric differences.
Other Notes
At the end of all the runs I made one last run with the Reynolds/Tufo setup inflated to 150psi. The run fell a good 10 meters short of the previous shortest mark. As such I didn’t even bother to mark and pursue higher-pressure runs.
The Raw Data
Tire/Wh Distance (meters)
S33/Rey 231.49
S33/Rey 231.74
S33/Rey 234.54
S33/Rey 236.47
CXP/ProR 267.43
CXP/ProR 264.00
CXP/ProR 259.30
Ksy/Cipo 251.76
Ksy/Cipo 252.77
Ksy/Cipo 257.88
Ksy/ProR 268.50
Ksy/ProR 264.74
Ksy/ProR 263.64
