Hi Joe,
Tour's wind tunnel photos show many sources of error. Most are "consistent" to the casual eye so are easy to dismiss as nit picking, but all of them can change the flow in unnatural (and often non-obvious) ways. The general effect is that the bike or wheel is being "tested" in some (invisible!) swirling air that, yes, produces a drag force, however precisely measured, but that has potentially weak (or no) relation to the drag you pedal against when riding out doors.
Here are a few examples off the top of my head:
- Swinging safety chain in the wake has upstream effects that are not present when riding. Might not seem obvious that this is important but users of CFD will recognize how a good wake is critical to getting stable test results.
- Struts are unnecessarily large, creating unnatural changes in flow.
- Large open holes in the floor can affect flow in an unnatural way as well. In both cases the turbulent flow shouldn't be there, but it is , and it can be picked up by the tires and thrown into the (supposedly) clean flow around the wheels and frame.
- Component groups were not the same on different frames (Red, Di2, etc.)
- Clothing. If the question is about clothing, you need clothing. If the question is not about clothing, you want NO clothing. Even "the same" clothing is never the same aerodynamically. Drag forces are very sensitive to even the smallest shift in fabric in even a very tight fitting skin suit. Cervelo wanted to clothe foam Dave (for lots of reasons, hides the joints and gives advertising value to the pro team's sponsors, etc.) but found that even with the awareness we'd gained through years in the tunnel we could not get the consistently precise results we wanted with clothes on.
Below are a few more thoughts I wrote earlier.
http://forum.slowtwitch.com/...post=3746899#3746899 A few comments, again with a grain of salt without having seen the article. In no particular order:
· Clothing. justkeeppedaling nailed it. We learned years ago that clothing, although it's realistic (no one rides naked, well almost no one!), it does add noise to the measured drag data, potentially swamping small differences in frames. Wrinkles matter, to an athlete for sure, but especially when the question isn't about the effects of clothing but about the effects of different frame designs.
· Right or left foot forward: Although our DZ mannequin doesn't pedal, we've played around in the wind tunnel with pedaling mannequins. There are still challenges to doing complete testing with moving legs, but one of the things we've experimented with is fixing the mannequin's legs at different points around the pedal circle. It's perhaps surprising that the main bike interaction is with "legs", and drag isn't so sensitive to "legs here or legs there," other than asymmetries some of you have already noticed. DZ's fixed legs are in the position that matches the same average drag as pedaling legs.
· Bottles. I didn't communicate personally with TOUR about bottles, but we did design the S5 with bottles in mind, since road riding is rarely done without at least one. Tom A. has already quoted the typical aero drag penalty for a round bottle, and also the improvement with the S5. The S5 was built with bottles in mind, and TOUR's test did capture that it seems.
· Mannequin position: Posture makes a difference, but if it's the same it's fair. John Cobb has even named "A" backs and "B" backs, for example. Cervelos work with both, so I don't think this is a big flaw with TOUR's mannequin. Could be closer to the average back shape perhaps but I'd put it farther down the list.
· Arms: Yes, our mannequin has triathlon arms. We chose that because triathlon is important to Cervelo, and our tri bikes need DZ's arms on the aero bars. That said, we've also done some other experiments and found that arms on tri bars have a very small effect on the frame’s aero drag, since the arms are a little above most of the frame elements that affect drag. There's less vertical component to air flow than some might imagine.
· Lasers: TOUR is smart to use precision positioning. I assume they've also positioned the bike in a smart way: For example, when the rear axle is fixed in the same position on the tunnel balance, then changes in chain stay length move the BB (thus the rider) forward or back, so the rider's position is different with respect to the wind tunnel. Same with BB height. We learned this years ago and account for it now.
· Open or closed wind tunnel: Both can work well.
· Cable housings: To accurately compare frames, cable housings should be included when the cable routing is different, since cable routing is a design choice that can really affect aero drag. In addition to different personal judgments on lengths and loops, with the "same" bike and cables we've seen that cables add turbulence and drag as expected, but they also add variability due to vortex shedding. It's typical in a wind tunnel test that the inner cables might not be fully connected to the derailleurs (or even present), and without the normal spring tension on the inner wires the outer housings create large drag variability because they'd be more flexible than in real life. We insert rigid inner wires (weld wire, for example) to address this. Otherwise you get unintentional drag variation that can be wrongly attributed to some other design aspect. Di2 is another variable – if some bikes are mechanically equipped then are you really comparing frames?
These are just a few of the thoughts off the top of my head and I'm sure the TOUR guys will only get better at wind tunnel testing. They've shown when they updated their stiffness test protocol that they're willing to improve as new knowledge comes, and after a few more trips to the tunnel I'm sure we'll have really good data from them.
You probably know Cervelo has spent a lot of time in various wind tunnels. We've learned there are a lot of details that can make surprising differences. This doesn't mean wind tunnels aren't any good, it just means that to find the differences you’re looking for you have to control all the other variables. There are many variables that at first glance don't seem to be, in fact, variable. It takes a lot of experience to accumulate the "mistakes" that lead to this knowledge. If done well, the difference in performance due to decisions in frame design is also there for the rider in real life, even when the other variables are there as well.
(End quote) Wind tunnel testing can be a good, valid way to measure the biggest performance aspect of bicycles. At Cervelo we were in the wind tunnel before the company was even started. We know the mistakes we've made and learned from in the past and it drives me nuts to see the latecomers (and not only TOUR Magazine) making many of the same mistakes. We've even offered to help but it takes money and that's understandably hard to come by when "bad" wind tunnel testing sells just as many bikes and magazines.
Okay, sorry for the rant. Suggestions welcome on how to move toward a better wind tunnel test protocol for the industry including companies and magazines.
:-)
Cheers,
Damon Rinard
Engineering Manager,
CSG Road Engineering Department
Cannondale & GT Bicycles
(ex-Cervelo, ex-Trek, ex-Velomax, ex-Kestrel)