Back this past fall I got the chance to spend a day in the Specialized Win Tunnel to test all sorts of gadgets out. Yes, it’s actually called the ‘Win’ Tunnel, not the ‘Wind’ Tunnel. But of course, a wind tunnel is what it is.
I’ve long wanted to test a wide variety of sports tech gadgets to see how they handled from an aerodynamic standpoint. But doing so without a repeatable testing methodology is tricky. I did some of that a while back at the ERO facility using Alphamantis. That worked well for testing positions and components. But it would have been harder for the sheer quantity of items I wanted to test. I needed something that would allow me to exclude all other variables (human or otherwise) and repeat for hours on end, while also being sensitive enough to detect the difference of a GoPro thumb screw. Thus, the need for a true wind tunnel.
The Specialized guys have long posted videos to their YouTube channel with various wind tunnel tests, from the popular shaving of legs to women’s hair-do’s. But in this case, I wanted to cover things they hadn’t done before in their videos. So I spent a full day there with Cam & Chris (of Specialized), and a suitcase full of gadgets to get to the bottom of things.
An Overview of the Tunnel:
First up is a primer on the tunnel itself. The Win Tunnel is located in Morgan Hill, California, as part of the Specialized campus. It’s not in the main building with all the designers, but rather is a separate building a few blocks away. In an ode to skunk works, the otherwise reddish brick was painted a jet black.
When you step into the building you realize the entire building was essentially gutted, leaving just this external shell of a structure. Inside is where they then built the tunnel itself. From a size standpoint, just imagine four double-length city buses stacked (2×2) on top of each other, sitting inside a high school gym. That’s kinda like the Win Tunnel building.
The tunnel basically moves air from front the front to back. Contrary to popular assumption, air is not pushed, but rather pulled. So in the above picture, we’re looking at the front of the tunnel where air would go through (ultimately hitting me on the head on the bike). If we walk closer you see this honeycomb structure. That enables the air to be normalized as it gets pulled into the tunnel section.
The area in front of the tunnel (where the air is pulled from) is quite vast, ensuring that there’s enough air to be pulled towards the tunnel as required at various speeds. It also serves as a great spot for makeshift parties and events. The remnants of one from a few days earlier was still present, including Crowie’s Ironman World Championship bike:
You’ll head up a short set of stairs (seen above) to the control room. There’s also a ramp around back too for rolling objects up to it.
The control room is a quieter area that allows them to control the wind tunnel, but it also acts as a classroom and staging area. Be it for teams, test setups, or so on. Across the way there’s another room that’s designed for storing secret stuff.
Before talking about the insides of the tunnel, let’s circle around the back. This is where you’ll find the giant fans that pull the air through the tunnel. You can see the wall to the left (below) is angled to re-distribute air back towards the front of the building.
The design of the tunnel is distinctly different from a typical automotive wind tunnel in that it’s designed to be low-speed (for cycling). With most automotive focused tunnels (or even ones for aviation), the speeds are far higher. This makes it tricky to get down to various speed thresholds that are more common to cyclists. Said differently: It’s not as useful to test at 80KPH, if you’re actually racing at 40-50KPH.
Heading back into the tunnel we’ve got a slightly raised platform that air can pass both above and below.
The bike is situated forward in the tunnel, and then there is a projector in the ceiling that projects real-time information down onto the ground:
The bike is fitted to a modified trainer to ensure you are pedaling. It’s technically a Wahoo KICKR below the paneling, albeit somewhat torn apart. You can see it here once the panels are removed.
The mechanical components under the paneled portion drive both the front and rear wheels, allowing a specified speed to be set (i.e. a ground speed that may or may not match the wind speed).
Even cooler though, is that the entire circle paneled portion actually rotates like a giant merry-go-round.
This enables them to test different angles to the wind, for example a wind hitting at 10° to the front (yaw). Everything is automated too, so that each test is done at various yaw angles. Here’s a quick video showing a bike (and me) on it, as well as a bike rotating autonomously on the platform.
However, the real key here is that those metal plates act kinda like a giant power meter sensor for wind. Essentially changes to pressure is what’s being measured. It’s not the wind directly. There is no colored streams of wind here, or any other fancy stuff somewhere in the fans. Instead, it’s the changes in pressure on the plate that determine how aerodynamic something is.
So if you were to hold up your hands like the YMCA song, it’d create more drag which in turn creates more pressure pushing down onto the plates (through your bike). Thus, less aerodynamic. Obviously, the entire plate system is incredibly sensitive.
All of this is then recorded in real-time into the control room and exported onto spreadsheets and data plots for deeper analysis. Additionally, the software stores photos of all the setups from multiple angles too.
The tunnel is big enough that they can (and have) done lineups of multiple riders, trying to figure out the best possible scenarios for team trials and lead-outs. Numerous pro teams and pro riders have visited the tunnel for testing. On the flip side, you can also do much smaller scale testing, of say just a tire, as done by Tom Anhalt lately.
Which brings us around to what I was testing…
What We Tested:
Like the last time I did aerodynamic testing, this wasn’t really about testing me or my position. In fact, I didn’t even bring my bike this time. Rather, they supplied a bike (the Venge). I just brought a suitcase full of gadgets and accessories, and an entire day of free time.
The goal here was to test a few categories of items:
A) Bike computers
B) Action cams
C) Bike computer mounts
D) Heads up display (HUD)
The last category was sorta a dumb-luck type of thing, since I happened to have it in my bag and there was some curiosity as to whether it’d be more or less aerodynamic than juts having a normal bike computer.
Here’s the exact lineup:
1) [Action Cam] GoPro Hero4 Silver with waterproof case
2) [Action Cam] GoPro Hero4 Silver with just frame case
3) [Action Cam] GoPro Hero4 Session
4) [Action Cam] Garmin VIRB XE
5) [Action Cam] Sony 1000V with waterproof case
6) [Action Cam] Sony 1000V without waterproof case
7) [GPS] Garmin Edge 520
8) [GPS] Garmin Edge 1000
9) [GPS] Garmin Edge 810
10) [Mount] K-Edge forward mount (combo Garmin/GoPro) without bike computer
11) [Mount] K-Edge forward mount (combo Garmin/GoPro) with bike computer
12) [Mount] Barfly SLi Mount (combo Garmin/GoPro) without bike computer
13) [Mount] Barfly SLi Mount (combo Garmin/GoPro) with bike computer
14) [HUD] Recon Jet HUD with full pods
15) [HUD]Recon Jet HUD without pods (just glasses)
16) [DCR] A typical DCR test bike setup (multiple GPS/cams/etc…)
As for why I didn’t test other items…well, it’s just what I figured there would be the most interest in. We did this back in September, so some things now seen weren’t available then of course, but in general I thought the above was probably the most likely items to be of interest to the most cyclists.
The bike used was a 58cm S-Works Venge 10R frame with CLX60 wheels, 24mm S-Works Turbo Tires, and a SRAM RED group set. Obviously and most importantly, it was also painted black. Because black is faster (duh).
For the testing it was generally without a rider (except the Recon Jet HUD and DCR setups). The reason is that in this case everything we were testing was forward of the human, and forward of the bars. So the human aspect wouldn’t be likely to influence test results.
Whereas if we were testing helmets, shoes, or apparel, you’d definitely want a human involved. With helmets for example, they are incredible person specific. What may be a faster helmet on Jane Doe, is a slower helmet on John Doe. But in my case, everything being forward on the bike it wasn’t an issue.
Further, with 6-8 hours of testing gadgets, it’d be a lot for any person to hold the same precise position over and over again. So this ensured that we got very repeatable results each time.
For each of the tests the automated system went through a yaw sweep of –10° (cross-wind) to 0° to +10°, in 5° increments. Each increment was held for 30 seconds of data collection. All of this was automated. The speed of the tunnel for tests was set at 50KPH (31MPH), and the wheels were also spun up to 50KPH for each test.
For the tests involving me (the Recon JET tests), the same protocol was followed, except that I was in the tunnel on the bike.
Action Cam Results:
Ok, let’s dig into some goodness. Now that I’ve explained all the basics, we’ll dive immediately into the results by category. Note that Cam also put together a full PDF report of the day too – which you can download here for a bit more detail.
And now – the results!
In this case, I was most curious if there would be any difference between the smaller Hero4 Session, and the larger Hero4 Silver/Black (identical external shell). Also, whether or not the larger waterproof case impacted aerodynamics. After that, I wanted to see how the Garmin VIRB X/XE performed (also identical external shells), and the Sony 1000V.
So what you see above is a graph plotting the drag at different yaw angles (from left to right). The lower the value, the better. For the action cam tests, we used the K-Edge Mount as the baseline for everything. It was our constant.
The results showed us that:
– The Hero4 Session was the fastest overall setup: Of course, the Hero4 Session doesn’t have the highest quality settings, so it’s a bit of a trade-off between speed and quality. This being fast wasn’t a big surprise to me (or the Specialized guys). After all, you’ll remember the Hero4 Session was released last year at the Tour de France and numerous pro teams had it rigged on the bikes as part of a partnership with Velon. Some testing had indeed been done around aerodynamics prior to mounting it on the bikes.
– Hero4 Silver/Black was only 2-4 seconds slower over 40KM: While it’s a tiny bit slower, the quality differences are pretty big here with the Hero4 Silver & Black. So unless you really think that 2-4 seconds is make or break, I’d still go for the Hero4 Silver. Of course, if you’re in the Olympics or racing elsewhere at the absolute front of the pack, then 2-4 seconds per 40KM could indeed be a deciding factor.
– Waterproof housing made no aerodynamic difference: Interestingly, the waterproof housing made no difference compared to the frame housing I often use. Of course, the real reason to use the frame housing is that you can improve image quality (by removing another layer of glass in front of the lens). Also, I just like the smaller look of it. The waterproof housing does spike a little bit more though at 10° yaw however.
– Sony Action Cam about the same as the GoPro Hero4 Silver/Black: While it looks more aerodynamic than the square GoPro’s, it turns out the Sony 1000V is actually about the same. And, even more interesting is that it doesn’t matter whether you use the waterproof case, it’s roughly the same there too.
– Garmin VIRB XE was the slowest: The Garmin VIRB X/XE was the slowest action cam we tested of the group, being 4 seconds slower at baseline 0° yaw. Though it was slightly more middle of the pack at cross-winds. These results won’t come as a surprise to many, after all, the thing looks like a sideways brick with its somewhat larger rectangular design. On the flip side, it has data metrics built-in, so again, if 4 seconds doesn’t matter to your use case – then go forth and enjoy the additional data. YouTube videos with data are more fun to watch anyway.
I know there are other action cams on the market not covered here, but again this is what I had handy and was able to test. I’d point out that having a GoPro compatible mount solution was a key factor in terms of ensuring all tests kept the constant while using the K-Edge mounting system.
We also did a test between a GoPro thumb screw vs a smaller screw as found on/with the K-Edge & Barfly mounts. However, we think there might have been some slippage there – so they believe there’s likely no difference but we’d have to re-test that down the road to be 100% sure.
Finally, I didn’t really want to duplicate tests that Specialized had already done. For example, they did a test around different action cam positions. You can see their video on that here.
Bike Computer Results:
Next up was testing three bike computers. In this case they were mounted to the K-Edge mount as seen above. The mount did include an empty GoPro mount on the bottom of the K-Edge combo mount.
Going into this test we somewhat knew that the variances between these units are super-small. They had done a bit of testing previously in this area. And indeed, you see in the results that at 0° yaw, we’re talking less .001 (m^2) shift in CdA – or roughly about a second over 40KM.
What you see here is that the Edge 520 was the fastest across all yaw angles, and the Edge 810 was definitely slower. However, we aren’t quite sure 100% on the Edge 1000. We believe the mount might have shifted slightly in this test – thus causing a small increase. If per the test, it was 4-8 seconds slower over 40KM.
But that’s because analysis of the imagery afterwards show a very slight increase upwards in the mount. Which, is actually a great learning point. If your out-front mount isn’t flat, it’s like having a parachute up there. So in this case, if I mounted it just a degree or two non-flat, it’s quite substantial in terms of aerodynamics. This is barely notable to the human eye, the photo above is actually what was tested, and isn’t truly perfectly flat.
Next up was the individual mounts. We had tested both the K-Edge Garmin/GoPro combo out-front mount, as well as the Barfly SLi Garmin/GoPro out-front mount.
Without a bike computer, we saw the K-Edge being the fastest, and with a bike computer, it was equal or the fastest as well at most yaw angles. The challenge was that we think the Barfly may have been raised slightly upwards as noted above. So you don’t notice it much without a bike computer, but once we tossed on the Edge 1000 it spiked. It was only while analyzing the photos/data later on that the Specialized guys noticed the slight difference.
In their prior testing of these mounts, they found them pretty close to the same.
So this one might have to go back to another round of testing sometime in the future. However, the very clear message there: Keep your mounts flat/horizontal to the ground!
Recon JET & DCR Test Setup:
Last up I had in my bag the Recon JET Heads Up Display. This unit includes GPS in the pods, as well as can connect to ANT+ or Bluetooth Smart sensors, projecting your metrics on a small screen attached to the glasses.
The way the Recon JET works is that you can remove the pods and just use them as sunglasses (no tech at that point), or, with the pods they are the full HUD experience. So we tried both. Additionally, we tested out the entire DCR testing config, just for the fun of it. But more on that in a moment.
As you can see above, the Recon Jet in a normal (pod-inclusive) configuration was quite a bit slower. Some 19 seconds over 40KM, pretty significant actually.
For the DCR setup, this was the last item we tested before I sprinted to the airport. It was just sorta a ‘for fun’ test to see how bad an aero penalty I take when going out and doing various testing, mostly power meter or action cam testing (training, not racing). In this case I had mounted:
– Four Garmin units: 2xEdge 1000’s on the handlebars, 1xEdge 810 on the stem, and 1xEdge520 on an out-front mount.
– Two GoPro cams: 1xHero4 Session, and 1xHero4 Silver.
– Recon Jet: Just because.
Overall, this was some 33 seconds slower over 40KM.
I don’t know how much slower it would have been if I didn’t have the Recon Jet on. Also, Specialized measured my rider repeatability as +/- 8 seconds over 40KM. Basically, how precisely I could replicate my position each time I ‘reset’ for a new test.
Final Odds & Ends:
Since I was there, I got a bit of a tour around the place. It’s fairly impressive. A large portion of the building is dedicated to museum space, classrooms, and related Specialized University areas. They’ve got numerous fit rooms setup to demonstrate and complete bike fits to dealers.
There are also conference rooms named and modeled after individual sponsored athletes.
For example Sam Hill:
Or, this one for Mario Cipollini, which is decorated in a much more opulent style reflective of Mario’s tastes.
A massive number of employees ride to work each day, or, go on the daily lunchtime rides. You can see the bike storage in the hallways here:
I went on one of the lunchtime rides, which would best be described as ‘rather brisk’ (but a blast). It’s definitely a very fit crew there, and they’ve got beautiful roads in their backyard to go out and ride on.
The main building even has a bike maintenance shop area at the back:
And of course, a pool table and lounge area:
There are two museums of sorts, one upstairs and one downstairs, covering both actual retail bikes, as well as some designs that never quite made it to retail.
And there’s even a fireman’s pole to connect to the two floors (also known as an exotic dancing pole for rowdy company parties). Stairs (or supreme fitness on the pole) are required for the ascent back to the 2nd floor.
Beyond these areas, it’s largely just normal office space of cubes. Water coolers and all, pretty much like most other offices. Except with sports pictures of sponsored athletes (or, once sponsored) instead of lame motivational sayings.
Actually, when you arrive they give you a water bottle instead of using paper cups. So you can refill that anywhere on campus. That does tie-in quite a bit to the Specialized Purist water bottles that the company makes onsite.
No doubt Specialized benefits from building out their own wind tunnel in terms of product testing, sponsored athlete testing, and of course marketing. But I think it’s these sorts of fun little tests that are the most useful to athletes day to day. Sure, the difference between the GoPro Hero4 Session and Hero4 Silver may not be a deal breaker for everyone. But at the same time, you’ve gotta start somewhere with aerodynamics. Be it shaving your legs, putting your hair in a braid, or using the right mount.
Every second does matter somewhere. Or at least that’s what I tell myself as I mindlessly sprint the last 50 meters of a triathlon for that extra 2-4 seconds. This post might just save me that sprinting (or, get me another few seconds back).
I’m sure down the road I’ll head back to the Win Tunnel to test out new gadgets as it makes sense (especially because it’s a blast). So if there are things folks would like to see tested, definitely drop them in the comments below.
And finally, a huge thanks to Cam Piper & Chris Yu (of Specialized) for spending the day in the tunnel playing around with sports tech.
Thanks for reading!