One of the reasons that cycling aerodynamics can be so difficult to get to grips with is that the various ways you can measure them often don’t completely agree with each other. Different methods and protocols suit different purposes, budgets and levels of expertise.
So even if you want to take the matter a bit more seriously than just guessing, you have choices to make about your method, how much you’re going to spend in terms of money and time, and how accurate you want the result to be.
I thought it might be interesting to take a quick tour.Wind tunnels are often taken as the “gold-standard” for aero-testing. They probably aren’t, but only because there’s not really a “gold-standard” in this area at all. The main advantage to a wind tunnel is that it can measure very, very accurately. If you stick to the same tunnel each time, the results are usually highly consistent from session to session.
Wind tunnels are often taken as the “gold-standard” for aero-testing. They probably aren’t, but only because there’s not really a “gold-standard” in this area at all. The main advantage to a wind tunnel is that it can measure very, very accurately. If you stick to the same tunnel each time, the results are usually highly consistent from session to session.
The biggest problem is that the thing they measure so accurately and consistently it isn’t always very representative of the real world. It’s fine if you’re looking at hardware like frames or wheels. But if you’re looking at rider position, it’s on a bolted-down bike, the rider probably pedalling at a fairly low intensity, and not having to deal with riding in the real world, in real conditions.
In a tunnel, for example, you don’t have to look where you’re going, you can hang on the front of the saddle with no risk of being bounced off by a rough road and you can achieve contortions that are impossible to sustain under full power. It’s easier than you’d expect to perfect a tunnel position that is practically useless in real-world cycling.
There are also choices to make about the speeds you’re going to test at, and what angle of yaw (crosswind) you want to find out about. So while you can measure very accurately, you quickly run into doubts about what it exactly is that you’re measuring, and how it relates to life outside the tunnel. Across the whole industry, tunnels are still very important, but they’re becoming a little less so for testing riding positions, suits and helmets.
At the moment, the main alternative to tunnel testing is track testing. You ride a few laps of a (usually indoor) track at a representative speed while your power and speed are recorded and processed through some software that corrects for the geometry of the track and produces a CdA (drag) number; just as in the tunnel.
More advanced ways of doing this will measure your speed through the air as well as over the track, since even a solo rider will start to stir the air in the track and create a slight vortex effect, meaning the air speed is a little lower than the ground speed.
It’s not quite as accurate as a wind tunnel. There’s more noise in the system, the slight air currents that are inevitable in a large open building will have an influence, and you’re ultimately relying on a power meter - and most of them are a little inconsistent. (Though you can spend your way past this shortcoming, inevitably.)
On the other hand, you’re testing riders in a more realistic environment, they’re working fairly hard and (hopefully) looking where they’re going. If a particular helmet is going to get pushed into a sub-optimal position as a rider moves on the bike, you can see that more easily on the track than in the tunnel, and more importantly, do something about it.
The biggest limitation to the track is that you can only test at track yaw-angles – it’s impossible to test for the wider angles you see with crosswinds on the road. On the track the typical yaw averages out at about 2.5 degrees – on the road it can be up to ten, and that makes quite a difference.
Computer modelling using CFD (computational fluid dynamics) is a familiar concept. You scan a rider and bike, run some simulations, and get lovely multi-coloured images that show you what the air is doing as it flows over a rider and their bike. It’s hugely controllable, and it doesn’t tire out a bike rider so you can try iteration after iteration.
It’s very good for hardware that’s in clean air – forks, wheels, handlebars, and other aerodynamically refined objects. It’s much less good for riders – riders are profoundly un-refined aerodynamically, and the bluff shapes of torsos, legs and arms produce very dirty, complicated turbulence, and at the moment the software isn’t quite developed enough. Even if you do use CFD to try to work out what’s going on, you have to validate it all on the track or in the tunnel, so at the current state of the technology it doesn’t really help riders (as opposed to equipment makers) all that much.
Field testing is the final option available to most of us. Conceptually this is the same as the track – see how fast you go and how much power it takes. If you’re a road rider, field testing has the big advantage that you’re getting data in the exact environment that you’re going to race in. It will replicate the yaw angles, the speeds, even the turbulence in the air produced by the air flowing over nearby buildings and fences before it hits you.
Of course, the equally huge problem is the sheer inconsistency that real world conditions produce. That’s why we started into wind tunnels in the first place, and it’s a difficult circle to square. And it’s not just the wind conditions. Anything that affects your speed becomes a problem, because a rough road or a dirty chain are impossible for power meters and speedos to differentiate from an aerodynamic issue.
It is possible to field test, though, and produce decent results. You just need enough data. Unlike a tunnel test or a track test, a day or two of dedicated work isn’t enough; you need to gather up many hours’ worth of data in order to factor out the variables. With a conventional power meter, lots of riding up and down a slightly hilly out-and-back route with minimal traffic and enough time and work, you can produce valid results using relatively user-friendly virtual-elevation software.
But what we’re beginning to see now are systems that will reduce the sheer time and effort needed. If you use a pitot to measure air speed as well as ground speed, that helps, though you still have the issues of road surfaces and rolling resistance to deal with. Or you can (as with Body Rocket) factor those issues out entirely and just measure the drag on the rider themselves. Capabilities like this mean you’re less limited to specific “test” rides – you can accumulate data during regular racing and training sessions.
The other plus to increasing the accuracy is that it’s realistic to use real-time drag – so you can see, as you’re riding, whether your drag number is drifting up, suggesting you’re not holding the right position on the bike. (Or down, suggesting your initial position wasn’t as good as you thought it was.)
To put it another way, we’re getting closer to aerodynamic drag becoming a “normal” number for a cyclist or a coach to work with, much the way power came out of the lab and into the real world through the 2000s.
