Sunday, March 6, 2016

Win Tunnel Playtime - Part 3 (The "After Party")

About a week after the fun session in the Specialized Win Tunnel (outlined in Part 1 and Part 2), I received an email from Cam Piper which included "wheel only" data for the Roval CLX64 wheel and each model size of both the S-Works Turbo (22C, 24C, 26C, and 28C) and Turbo Cotton (24C and 26C) tires. At the time I visited I had asked about this data set, and although they had all the data, it wasn't easily collated into a single file. So, on the following Friday, Cam took on the (large) task of running and procuring a dataset for all of those tires in a single session. That data is summarized in the CdA vs. yaw angle plot below:

Roval CLX64 Wheel

As one can see, that's a really nice data set, especially the symmetry. Also, one can clearly see the effect of tire size on drag, especially with the S-Works Turbo sizes. In regards to size, it's important to note that the listed tire sizes typically "grow" by ~2mm when mounted to a wheel with such a wide interior bead width, such as the Roval CLX64 (nearly 21mm, if I recall correctly). For example, the 22C S-Works Turbo actually measured slightly >24mm wide when installed on that wheel.

Not long after receiving this data, I also received a box from Specialized containing samples of each of the listed tires for me to roller test personally for Crr.  Wow...that's a lot of tires to test, plus at the time I didn't have a bike capable of testing the 28C S-Works Turbo when mounted on a wide rim. Luckily, I was in the process of building up a "gravel bike" (based on a 26" rigid MTB frame...I digress...) and so eventually I could get the Crr data for that one as well. 

When the tires arrived, I realized that it may be time to revisit my tire testing protocol and tweak things to make the process more amenable to testing wider tires. When I started roller testing tires, I was mainly looking at TT tires and wheels of the time were still significantly narrower than they are today, so testing at 120 psi seemed reasonable. However, as rims and tires have gotten wider, I've been uncomfortable with pumping tires like 26C and 28C models up to 120 psi to test...and in fact, have had a couple cases where I couldn't get tires that wide to stay on the rims when inflated! Knowing that, while testing this batch of Specialized tires I decided to do some narrow (Mavic Open Pro) vs. wide (Hed Jet+) rim testing along with tests at 120 psi and 100 psi. To make a long story short, I found that in general, a tire will roll approximately the same on my rollers on a narrow Open Pro rim at 120 psi as it will on the extra wide Jet+ rim at 100 psi. That was a valuable thing to discover, and it means that my future tests will be run at 100 psi on a 20.5-21mm internal width diameter rim. I especially wanted to test all of these tires on a wide internal width rim for Crr so that it would best match the tires as mounted on the Roval CLX64 wheels used in the aero testing.

The Crr results for the 6 Specialized tires are as follows:

Specialized S-Works Turbo 22C = .0041, 38W for pair @ 40kph
Specialized S-Works Turbo 24C = .0036, 33W for pair @ 40kph
Specialized S-Works Turbo 26C = .0035, 32W for pair @ 40kph
Specialized S-Works Turbo 28C = .0035, 32W for pair @ 40kph (note: AC101 disc wheel)

Specialized Turbo Cotton 24C = .0029, 27W for pair @ 40kph
Specialized Turbo Cotton 26C = .0028, 26W for pair @ 40kph

By comparison, here's the results for the "benchmark" GP4000S (tested on Open Pro @120psi):

Continental GP4000S 23C = .0034,  31W for pair @ 40kph

There are a couple of interesting takeaways in those results. First, I was somewhat surprised at the seemingly non-proportional "hit" in Crr the 22C S-Works Turbo took in comparison to the 24 and 26C sizes. It was also interesting that 28C size basically rolled the same as the 26C size, although that may be attributable to the fact the 28C was tested on a low profile, 32 round spoke disc brake wheel, while the 26C was tested on a Hed Jet6+ deep wheel with bladed spokes. One other interesting result is that in the Turbo Cottons, the 26C tire rolled only slightly faster than the 24C. This may mean that the Crr of that tire model is driven more by the tread composition (both tires appear to use the same width tread) than by the casing material properties. Lastly, these results appear to confirm that the S-Works Turbo tires in the 24C and 26C sizes are basically "tied" with the Continental GP4000S in terms of Crr (remember that I consider anything within .0001 of Crr to be "tied")., if any of you have read some other blog posts of mine, you probably know where this is going.  Yep...what does it mean when we combine the Crr and CdA results? Which of the tires above gives the best combination of rolling resistance AND aero for a particular application? 

Well, to truly get at that answer requires some fairly detailed modeling, such as that performed by, for example. However, it's possible to at least get an inkling of which tire may give the best combo using some simple assumptions.  In this case, I made the assumption that the wheel load is 38kg (~45% of my typical "all up" mass of 85kg, typical of a front wheel for me) and a ground speed of 35 kph (~22 mph).  Figuring out the power for each tire at that speed is easy, and is merely Crr x Speed x Mass x gravity.

The expected power for the aero drag is slightly more difficult...and involves using the apparent wind speed expected for the particular ground speed AND yaw angle. It's probably easier to explain with some vector diagrams (and I'll do that if the interest is shown), but suffice to say that if your ground speed is 35 kph AND you have a non-zero yaw angle, then the apparent wind acting on the rider is going to be greater than 35 kph. It's important to remember that the results coming from a wind tunnel are the CdA (or sometimes grams of drag) in the body axis of the wheel, or bike, for a given APPARENT wind speed (typically set to ~30mph for better resolution). So, what that means is that there is some trigonometry that needs to be undertaken for non-zero yaws for the power calculation. In this simplified analysis, it's also an assumption that any sidewind is a "pure" crosswind, or oriented 90 degrees to the riders travelling path. Thankfully, the data acquisition setup at the Specialized Win Tunnel already does that trigonometry for us. This is sometimes referred to a "beta correction" in wind tunnel parlance.

So, I set up a spreadsheet to handle all of that, and here's the expected combined power for a single wheel with a 38kg load, travelling at 35 kph.

Roval CLX64, 38kg load, 35 kph ground speed

So, what can we take from that plot? As I've said previously, "Low rolling resistance can make up for a LOT of 'aero sins' " (Here, and here)...and that plot above helps bring that home. Although the S-Works Turbo 22C tire was the clear leader in the CdA plot, when combined with the expected Crr of the tires, it actually isn't as good as any of the other tires at 10 degrees and below of yaw angle.  In my view, the Turbo Cotton tires are the clear winners in the combined power plot, with the edge going to the 24C version, at least in my mind. The total power at 0 and 5 degrees are basically identical, but the 24C has a slight edge at 10 degrees. At 15 degrees, the 2 tires are tied again.

In looking at the S-Works Turbo tires only, it appears that the best overall of the bunch is the 24C model, with the 26C nearly tied with it.

Now obviously, that's the results for that given load and ground speed. For different wheel loads, the Crr contribution is going to vary proportionally up and down relative to the load, and for different ground speeds, the aero proportion is going to vary with the cube of the ground speed up and down. So, for lighter and/or faster riders than what is assumed, the aero effects will be relatively more important, while for heavier and/or slower riders the Crr effects will be relatively more important.

After calculating these results, of course I also applied them to the wheel and tire combos that were taken during my Win Tunnel visit.  To remind everyone of the CdA plot of all of the combos looked, here it is again:

Combining that plot with the Crr results like we did above, results in the following overall expected power plot:

I think I'll just leave that there without further comment...Enjoy!

The spreadsheets containing the data and calculations can be viewed here:

- Roval CLX64 plus Specialized Tires

- All wheels from Win Tunnel Visit

edit 23Jan2017: After roller testing a newer version of the Continental SuperSonic in 2016, I found that it had significantly improved the Crr in comparison to the c2012 version I had previously tested. So, I decided to use the newer value in calculating the total power for the H3/Conti 20C SS combination in the chart above.  With those changes, the chart looks as follows:


  1. Tom, Outstanding, though I dare say you've just opened yourself up to a whole lot of added work. How generalizable do you think the results are to other tires you've tested?

    If/when/over time you can get around to it, it would be wonderful if you could come up with a tool where one could plug in load, speed, tire model, tire width and get combined CdA and Crr plots as you have done with the Specialized ones above. Assuming the tire width is the primary variable of aero performance at a given speed, load, rim width and less dependent on brand (a crude assumption but don't know whether others will be as accommodating as Specialized in providing you aero results) and the Crr differences you will develop with your new wider rim, 100psi testing, then would something like that be doable?

    My only other comment is that going from the Open Pro inside width (13mm?) to the Jet+ (20.6mm) rim for testing is a huge jump. Perhaps another width representative of the 17C alloy and 19C carbon wheelsets that are more standard today might be worth considering in addition or instead of these.

    Best, Steve

  2. Hi Steve, Thanks!'re probably right about the extra work ;-)

    The only downside of a tool like you describe is getting the wheel/tire combination data to pull that off. That's a lot of data and time.

    The Open Pro measures just under 15mm. I had done some tests previously with a "mid" width rim (Zipp 101) and couldn't really tease out any Crr differences from the measurement noise. It seems like it has taken the REALLY wide rim like the 21mm internal width Jet+ to start seeing that effect on the rollers.

  3. I'd be curious to see how you worked out the vectors

  4. This is great stuff. Thank you! Looks like I will be retiring my SWorks turbo 22 after one ride. The supersonic and turbo cotton rule the combined tests. I am curious if your roller tests are adjusted for real world surfaces or not. Do you believe a multiplier of the Crr is appropriate? I suspect that would only widen the gap.

    I am very interested to see if the Vittoria Corsa Speed falls into the fast category. Having a similar construction as the TC, but in a slightly narrower package might make it a fantastic TT tire.

    1. Hi Austin,
      Yes, the Crr calculations include an "on road" factor already. I use 1.5X over the smooth roller results since that is what I've measured for my roads around here. That's also consistent with the 1.3-1.4X found by Andy Coggan.

      I too am curious to see how the new Corsa Speed performs. There seems to be conflicting info on just how good it is...but, the rollers don't lie ;-)

    2. I have them on order. If I get them before you I would be happy to send one in for testing.

  5. Super work thanks so much for sharing
    I've 2 options for front wheel for tt and track pursuit outdoor track
    Hed 3 clincher currently bont aerowing but changing to conti supersonic
    And a enve ses 6 with 24mm spec turbo cottons
    Both latex tubes and riding s5 set up as tt bike speed for tt Apx 42/43kph depending on distance .. Pursuit obviously quicker
    Would you hazard a guess as to which should be my go to front wheel
    Thanks again

  6. Interesting to see Roval with Supersonic...

  7. This comment has been removed by the author.

  8. Perhaps I missed it but which FLO 90 did you use? I know it's been updated in the last year or so. Thanks!

    1. The original Flo 90. The new ones didn't exist at the time.

  9. Have you considered calculating estimated factor that would account for the used tire size (volume, not just width) and effect of tire pressure on CRR. Because most presumably 25mm tire will roll better (on smooth surface) when inflated to 120psi than similar 23mm tire.

    Or what do you think that would be reasonable when testing different sized tires and trying to factor in the effect of tire pressure?

    Though CRR-charts for a tire tested with different inflation pressures don't show that great variance in CRR if the difference in pressure isn't huge.