AI quick summary
- Aerodynamic drag dominates cycling resistance above ~24 km/h, making it the highest-leverage variable for free speed.
- Virtual elevation (the Chung method) lets you solve for CdA from power, speed, and elevation data you already collect.
- Field testing is cheap but unforgiving — wind shifts, position drift, and bad Crr assumptions will wreck your results.
- A-B-A protocols with one variable changed per run separate real effects from measurement noise.
/ 01
Why Aerodynamics Dominates Performance
Once you're moving faster than about 15 mph (24 km/h), aerodynamic drag becomes the largest resistive force you have to overcome. By race speed it accounts for 70–90% of total resistance. That is why a rider with average fitness but a clean position can drop a stronger rider who sits up into the wind.
The catch: aerodynamics is highly individual. Two riders with identical height, weight, and equipment can have very different drag coefficients because of shoulder width, head position, hand placement, and how their kit fits. Catalogue claims from manufacturers are useful for relative comparisons but rarely match your actual numbers on the road.
/ 02
The Virtual Elevation Method
The most accessible form of field aero testing is the virtual elevation (VE) or Chung method, popularised by Robert Chung. You ride a known course while logging speed, power, wind, and elevation. A script in GoldenCheetah, AeroTune, or your own spreadsheet inverts the power-balance equation to solve for the CdA (drag coefficient × frontal area) that makes the reconstructed elevation profile match the real one.
If your virtual elevation shows a flat or looping shape when the real road is flat, your assumed CdA is close to correct. Bumps in the VE plot hint at wind gusts, brake rub, or position changes. With a clean run you can resolve CdA to within ±0.005 m² — meaningful enough to detect the difference between two helmets or two hand positions.
/ 03
Field Test Protocols
Run tests on a flat, quiet, sheltered road — ideally a closed circuit or out-and-back with no stops. Six to ten laps at steady power (tempo, ~75% FTP, in the drops or target position) is the standard. Hold cadence and posture constant: shifting your hands or looking down changes CdA more than the equipment you are testing.
Test one variable at a time. Do not change helmet, jersey, and bottle cage in the same run. A-B-A testing — baseline, change, baseline again — catches drift in wind or rolling resistance. Record air density from a barometer and humidity sensor; small pressure shifts change the drag calculation by a few percent across a session.
/ 04
Aero Testing Methods Compared
Common aero testing methods, by cost and resolution
| Method | Cost | Resolution | Best for |
|---|---|---|---|
| Wind tunnel | $500–2000 / session | ±0.002 m² CdA | Final position validation, R&D |
| Track with aero sensor | $200–800 rental | ±0.005 m² CdA | Helmet and wheel comparisons |
| Virtual elevation (field) | Free (power meter) | ±0.005–0.010 m² CdA | Position and apparel changes |
| On-bike pressure sensors | $300–800 | Qualitative | Real-time aero feedback |
/ 05
Common Errors and Fixes
Most failed VE runs trace back to bad inputs. A mis-calibrated power meter, wrong Crr (rolling resistance) assumption, or unmeasured wind shift will sink the reconstruction. Use the same tyres at the same pressure across test days, and check your power meter's zero-offset each lap.
Body position drift is the other killer. Even a few millimetres of shoulder drop or wrist angle change shifts CdA more than the apparel you are testing. Use handlebar-mounted video or a second rider with a camera to verify posture across laps. The data is only as good as the discipline that produced it.
/ Sources