I was offered the opportunity to check out the new CCNS wind tunnel in Middletown, CT last week. The tunnel, which opened for business on September 1st, is the brainchild of brothers Aidan and Foster Charles. I called up Anthony Accardi and we took the two hour drive from Manhattan to visit the tunnel.

The CCNS tunnel, the latest component of the 3 year old performance and physiological testing center, began innocently enough when pro cyclist and coach Aidan told IT tech Foster that he thought they could build a wind tunnel. The two tested the concept with whatever they had on hand (rubber bands, fish scales) and soon realized that it was indeed possible. Wayne Kirk, fellow cyclist and engineer with Pratt & Whitney, stepped in to provide calculations. Kirk, whose work included building and testing wind tunnels, taught the boys everything he could about aerodynamics. He also worked out the minimum size for a workable cyclist tunnel, and a reasonable wind speed suitable for extrapolation to typical TT speeds.

The rest was up to Aidan and Foster. What they couldn’t build themselves, they sourced out to electricians and metal and machine shops. Aidan did much of the building, and Foster wrote the programs that gathers and analyzes the data. The tunnel itself is a steel and wood structure, rounded off with laminate corners. Two fans sit behind the rider and pull air through the tunnel, while other fans help complete the circuit around the room. The front end of the tunnel is covered by a wall of straightening louvers that resemble a honeycomb of drinking straws. There are cameras in front of and above the rider, and a monitor sits on the floor before the front wheel.

The bike mount is the most complex and most secret part of the tunnel. It’s an interchangeable module – different modules sit at different yaw angles. The mount is motorized, shifting fore and aft so the rider can mount the bike while the delicate sensors underneath aren’t engaged. The sensors take 10,000 data points per minute, and are incredibly sensitive. The slightest twitch is registered. Other modules spin wheels for gear tests, while another allows the rider to pedal inside the tunnel.

Ok, but…
As much as I’m in awe of this amazing achievement, I’d be remiss to not point out some negatives. The only other wind tunnel I’ve visited is the San Diego Low Speed Wind Tunnel, widely considered the state of the art tunnel for cyclists. The SD LSWT’s cross section is about five times bigger than the CCNS tunnel. The presence of the rider effectively reduces the cross section of the tunnel, forcing the airflow to accelerate. A bigger tunnel minimizes this effect (since the rider will be a smaller percentage of the total area).

The LSWT has a variable speed fan, and an anemometer inside the tunnel to track wind speed. It can easily exceed 100 mph, and measurements on cyclists are made at actual riding speeds. The CCNS tunnel runs at a fixed speed – 14 mph – and results are extrapolated up to riding speed. Lower speeds mean lower drag forces, and friction and other noise would probably have a greater effect on the data.

The LSWT has a Computrainer-like setup in its bike mount. You can dial in a wattage and ride during the measuring runs. A drum spins the front wheel at the corresponding speed. You have to be able to pedal the bike at that wattage to be able do a run. If you adopt a ridiculously aero but unrideable position, you’ll find out. Also, the mount rotates for quick changes of yaw angle. Measurements are taken statically in the CCNS. Your right crank is zip tied to the chainstay and you’re required to stay as still as possible during each 20-30 second run. Neither wheel spins.

Finally, the LSWT has enormous windows on its sides so that a video camera can shoot the rider from outside the tunnel. This view is used to compare rider position for different runs. The tight confines of the CCNS tunnel make a side view impossible for now.

Accentuate the positive
I brought up all of these points to Aidan and Foster, and they had an answer for all of them. I didn’t necessarily understand all of it, but nevertheless… They say they’ve compensated for the size of the cross section, and that 14 mph is plenty fast enough. Extrapolation is a legitimate and accepted process in wind tunnel testing. While CCNS has a platform that allows pedaling, they prefer the accuracy of the static platform. Aidan can always take you out of the tunnel and test the viability of a position on a CompuTrainer.

Ultimately, they say, their tunnel is both consistent and accurate when calibrated against reference objects and riders who have visited other tunnels. While I was there I was quite impressed with the amount of time the boys spent on calibration. Clearly they take accuracy seriously and are undaunted by the tedium of being precise.

And let’s not forget that for us east coasters, the CCNS tunnel is much more accessible, and at $475/hour, less than half the price of the SD LSWT.

Hopping In
I was only an observer during my trip to the SD LSWT (Kristin Armstrong and Bernard Van Ulden were being tested), so this was my first time being tested at a tunnel. I was quite surprised at how quickly my attempted detached coolness was replaced by pure giddy schoolboy glee. Perhaps Accardi and I are just hopeless geeks, but we had a great time being tested. This being a quick demo, we only tested three positions each (depending on the extent of equipment changes, Aidan says he can get up to ten runs in an hour).

My first run was a baseline measurement (all our runs were conducted at 5 degree yaw). My CdA was .248, which is pretty good. We didn’t have enough time to make equipment changes, so we only explored a couple other options. We put my hands up Floyd Landis style and I got slower. Aidan has found that the Praying Landis is generally quicker for shorter riders. He thinks that the greater distance between the hands and face on a taller person means the wind treats the two as separate objects instead of a single unit. Next he checked the airflow with a ribbon on a stick (smoke is also an option), and found a dead spot under my helmet. Turns out that my Spiuk, which normally tests well, doesn’t mesh well with my back. We tried a LAS and knocked off a second and a half per mile, reducing my CdA to .24.

Accardi already had his hands up, and he turned out to be slower with his hands level. Interestingly, he’s 2 inches shorter and his arms are and inch and a half shorter than mine. His baseline CdA was .286. He also tried the LAS helmet, but only gained a half second per mile.

Verification
The reason I brought along Accardi, aside from the fact that he’s charming and smells like a field of wild flowers, is that I wanted to test the accuracy of the CCNS tunnel by comparing their numbers against our data from the Floyd Bennett Field time trial from June 8th. I wanted to compare the ratio of my CdA vs Accardi’s CdA in both instances. My hope was that by comparing numbers from the two venues I could eliminate the discrepancy between air density, yaw angle, and wind speed. In other words, my real life CdA may not match the tunnel CdA because conditions were different, but it should be a similar percentage of Accardi’s in both instances. I’d be comparing apples to apples and oranges to oranges.

My average speed on that day was 25.9 mph at 279 watts, Accardi did 26.3 at 326 watts. A rough estimate of our CdA’s from those numbers (ignoring wind and pacing strategies but factoring in 45 watts of rolling resistance) has me at .246 and Accardi at .282. The ratio of our CdA’s is .872. At Floyd Bennett, Accardi’s hands were somewhere between his first two runs at CNSS, so if you split the difference and use .287 as his CdA, then the ratio of our CNSS test is .867. That means there’s a 1% difference between the two ratios. Pretty amazing, even keeping in mind there are enough holes in those calculations to drive a truck through.

Conclusion
This experience just reinforces the notion that it’s folly to try to perfect one’s position intuitively. You can get pretty far, but without real data you’re just groping around in the dark. Aidan tells me that people can sometimes get faster by moving their head, raising their bars, or shifting fore/aft – things that you just don’t expect would work. I experimented with the Floyd Landis position thinking it made me faster, not knowing I was slowing myself down in reality. And if you think that this is frivolous, a .008 reduction in my CdA would’ve dropped my time at the Floyd Bennett race from 26:05 to 25:27. If Aidan had more time and took .02 off my CdA (there’s no guarantee that this could be done, of course), my time would’ve been 25:00, vaulting me from 25th place overall to 9th!