Wind tunnel testing with John Cobb at Texas A&M

As I was driving to Denver International Airport yesterday I was talking to my wife, Sarah, on my cell phone, and telling her that I was going to miss a day of riding while I was doing my wind tunnel testing at the Texas A&M wind tunnel with John Cobb. I didn’t realize that I’d be doing 4 hours of five-minute intervals. I’m pretty sure I did 17 five-minute intervals inside that tunnel, and some of those were in pretty uncomfortable positions. Near the end of the testing session I asked that they lower the resistance, which they did by putting me in a lower gear. (I couldn’t shift into a lower gear myself because I didn’t have cables on the bike.) I should have asked for that sooner, but live and learn.

My overall impression of this wind tunnel testing was that it was much better than my first testing session in Ft. Collins with Colorado Premier Training. In Ft. Collins, while the staff (Mark Cote in particular, who now works for Specialized) were quite knowledgeable and experienced, I felt like the equipment was duct taped together, the tunnel was “blustery”, and the results weren’t repeatable. In Ft. Collins they only tested me straight-on (zero degrees of yaw) and they asked me not to pedal. Each run lasted about 30 seconds. With John Cobb, at the Texas A&M tunnel, each run lasted about 5 minutes, they had me pedaling pretty hard the entire time, and they rotated me through a range of yaw angles that went from zero to 20-degrees. And I was very happy to see that the numbers were repeatable. Two or three times during the four-hour testing session they switched me back to a previous setup (same equipment and same position) to see if they got the same drag numbers as before. I was quite pleased, and I gained confidence in the testing methodology, because repeated tests were pretty much spot on with the numbers from before. In defense of the Ft. Collins testing program, I was one of the first riders tested there and it’s quite probable that the equipment and the testing methodology have improved since I was there. For example, they did have a rotating platter to test various yaw angels, but it was broken on my testing day so we didn’t use it. I’m assuming that it’s working now and that they do test multiple wind yaw angles.

Similar to my first testing session, I got to witness the wind tunnel test before mine, and the guy “came out 2 minutes faster” than he went in. That really means that “the guy” was way too high to begin with, and John Cobb lowered his bars by about 10 cm. I’m not exaggerating, it really was something in that range. I mentioned to the testing staff that given my Ft. Collins testing session, where they couldn’t make my position any faster than when I came in, that I wouldn’t be surprised if John Cobb and his team also couldn’t get me any more aero. To bolster this fear, when I came out of the tunnel after my baseline tests, several people in the computer filled room said things like “wow, you’re one hell of an aero kid” or “you’re the most aero person we’ve seen so far. It’s no wonder that you can go fast.” These are not the words you want to hear. You’d much rather they say “Holy cow! Your position is like a parachute….” They also tested my 2009 Cervelo P4 alone and said it was measurably faster than the three Cervelo P4s they tested earlier in the week. They attributed it to my soon-to-be-illegal 3T Ventus LTD bars, and my Zipp VumaChrono cranks, and my one-of-a-kind center pull brake. The bike is also fairly small at 54 cm.

After the initial baseline tests they swapped out my non-adjustable, sleek and beautiful 3T bars with a very ugly, not-so-sleek, set of uber-adjustable bars. These bars consisted of a Look Ergo Stem (courtesy of Al Morrison, best know for his tire rolling resistance research) and basically two base bars. It’s hard to describe why it was essentially two base bars, but there was a real base bar, and then the clip on Profile bars extended not only forward, as you would expect, but they also went horizontally pretty far, in effect creating a double-decker base bar. The ugly duckling setup also had rather large, and squarish, clamps on the bottom and were literally the antithesis of my 3T bars. John Cobb and his team then adjusted the elbow pads and the extensions to match my initial position with the 3T bars and ran a test again. I was significantly less aero with this setup, which was due to the far from ideal bar setup. It surprised me how much less aero I was with the same position, but with really ugly bars. I guess equipment does matter.

The bulk of the afternoon was spent trying to reduce my drag by changing my position on the bike. They lowered me 5 cm and noticed a slight improvement. Then they lowered me another 2 cm and got the opposite effect — more drag. So they raised me back up those 2 cm and called that my ideal drop. Then they tried moving my arms closer together and this is actually where they found the most improvement with my position. I have fairly flexible shoulders and in the end they had me riding with my forearms touching each other. Then then tried angling my forearms up, in a slight Floyd Landis “Praying Mantis” style position, and with my body type this was worse than a level extension. I guess that’s good as the Mantis position is no longer legal, either. Oh, and I shouldn’t forget that we did try some crazy positions, including a few where I wore my prism glasses and kept my head entirely down. The prisms change your angle of view by 90 degrees and allow you to look forward with your nose pointing directly down. It turned out that I was indeed pretty aero in this position, especially with my borrowed short-tailed retro Bell Vortex aero helmet (courtesy of Kevin Nicol) but the difference was too small to make it worth the risk of driving a bike at 30 MPH while looking through little mirrors. I also have a suspicion that in a real race those little mirrors would get covered with sweat in about 10 minutes and become useless. Nevertheless, it was a fun experiment. Anyone want to buy a set of prism glasses at a bargain?

Then we moved on to trying different helmets and skinsuits. In a nutshell, the fastest helmet for me was the Bell Meteor II that I’ve been letting Kevin use this past season.And the fastest skinsuit was my Discovery Team Nike Swiftspin skinsuit. My team Excel skinsuit was a very close second, close enough that I’ll continue to use it out of loyalty to my team. And oddly the long-sleeve Garmin-Slipstream speedsuit was the worst. John Cobb thought it looked like a fast suit, and was questioning the data, so we ran a few tests comparing the Garmin-Slipstream speedsuit with the others and it consistently came out significantly worse. This is the skinsuit I let Kevin Nicol use at Moriarty for the National Record Challenge this year and maybe I slowed him down by doing so. Sorry Kevin. Actually, this is probably a good time to mention that John Cobb said, particularly in regards to clothing and helmets, that the optimal equipment choices are super idiosyncratic, and what’s great on one rider might be horrible on another. So in reality the Garmin-Slipstream speedsuit might have been great on Kevin.

Then to wrap things up we went all the way back to my starting position (on the ugly-duckling bars, not the 3T bars) to confirm the overall improvements we’d made. I was happy to see the drag numbers go up, and a bunch of nodding heads indicating some sort of satisfied approval, as the little dots on the computer monitors showed my drag to be back where we started.

The math is complicated. Despite their best efforts to limit what variables change from one test to another, things other than the rider position and equipment do change with every run. As an example, they were shooting  for 30 MPH of wind with me, as that’s the speed I went at Moriarty this year, but the actual speed of the wind appeared to vary maybe a few tenths of a mile per hour from run to run. And the temperature, humidity, and barometric pressure, can all change during the 4 hour testing session, too. So to get really accurate data they need to factor all of those minute changes into the drag calculation equations. It’s possible that John Cobb’s software was factoring those variables into the numbers in real time — I’m not sure. But they were in a hurry to pack things up, and let the Texas A&M staff go home, so our post test debriefing wasn’t very extensive. They said that they would send me the complete report after Thanksgiving, once they had time to compile and review it. As a quick parking lot assessment, they guesstimated that my new position was about 25 seconds faster over 40k than my old position, assuming a good aerobar rather than the ugly duckling.

So was it worth the money and time? Absolutely. There were at least half a dozen races for me this year alone where I either lost a place by a second or beat the guy ahead of me by a second. I had the second fastest time at the Cherry Creek Time Trial, on the final race of the series, by less than one second. Taylor Phinney beat me in one of the 10-mile Lyons to Boulder time trials by 1 second. I got third at Masters Nationals and was only 4 seconds ahead of 4th place. And Kevin Nicol, my great friend and teammate, beat me at the National Record Challenge by a scant 3 seconds. So seconds do count.

I’ve spent my testing budget for this year, but eventually I’d like to go test with Kraig Willett of in the San Diego tunnel and see what they can do.

Here are some photos:

3 thoughts on “Wind tunnel testing with John Cobb at Texas A&M

  1. Pingback: Wind Tunnel Test Results - Cycling Technology Blog

  2. Hey Nico, I’ve had the idea to use the prism glasses for the head-down position on aero bars too! Great find, the prism glasses with reversible position. I’d be interested in buying them from you, shoot me an e-mail. ~~Brad

  3. Pingback: World’s most aero UCI legal bars? - Cycling Technology Blog

Leave a Reply

Your email address will not be published. Required fields are marked *


You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>