The Chung method: aero testing with your power meter

The short answer: the Chung method turns a power meter into a wind tunnel. Ride laps of a quiet loop without braking, and work out which drag area (CdA) makes every lap end at the height it started. It is good at telling two setups apart on the same day and less good at an absolute number, and only as good as your testing.

How it works

Your power goes into four things: rolling resistance, climbing, speeding up, and pushing air. Power and speed are measured, mass you weigh, and air density follows from the weather. Take a guess at CdA and the power equation gives the slope of the road at every moment. Add those slopes up and you get a “virtual” elevation profile ¹.

On a loop, each lap ends where it began. With the wrong CdA the virtual laps drift up or down; with the right one they line up. That is the whole trick, and analysis software such as Golden Cheetah does the sums.

Unlike the classic method, which needs constant speeds on a flat, windless road ², the road need not be flat and your speed need not be steady. Varying both actually helps ¹.

How to run a test

Find a loop of half a minute to a few minutes that you can ride without braking and without traffic. A slight dip or rise is fine.
Ride several laps per setup, at different speeds, and hold your position exactly.
Don’t brake. Braking shows up as a sudden step in the profile, because energy disappears that the model knows nothing about.
Pick a calm day. Wind is the biggest source of error. It shows up as laps whose shape changes with your speed.
Know your numbers: total mass of rider and bike, and air density from temperature, air pressure and altitude. In Chung’s example a 1% error in air density moved the CdA by about 1.5%; a 1% error in rolling resistance only by about 0.3% ¹.

What it can tell you

Small differences, if you test well. When rolling resistance is known, Chung reports agreement with wind-tunnel and classic field tests within about 1%. Experienced users report a spread between repeated estimates as low as 0.3%. These are reported results from users, not a controlled study, so take them as what is possible, not what you will get on your first try ¹.

That precision is real enough to matter. At 40 km/h, a change of 0.01 m² in CdA is about 8 watts; at 30 km/h about 3.5 watts. A field study with ordinary power meters found 11% less aerodynamic drag on the drops than on the hoods, and could also see the rolling resistance of tyres at half pressure ².

Your absolute CdA is harder. Drag area and rolling resistance trade off against each other, so unless you know your rolling resistance, you get changes more reliably than the true number. With the same tyres at the same pressure on the same road, rolling resistance stays put and the changes are what you measure ¹.

Testing well, whatever you test

The same rules make any test trustworthy, whether it is a position, a helmet, tyres or a new pacing plan:

Change one thing at a time. Two changes at once tell you nothing about either.
Compare on the same day. Chung advises against comparing results across days, because the conditions are never the same ¹.
Go back and forth: A, B, A, B. If A is the same both times, the day was stable and you can trust the difference with B. If A drifted, so did your conditions.
Repeat before you believe. One run is an anecdote.
Look at the data, not just the number. The virtual elevation plot shows where a run went wrong: a brake, a gust, a shift in position. Throw out the run, not the whole test.
Write down what you did: pressure, clothing, position, weather. A result you cannot repeat is not a result.

Use what you measured

The rider profile estimates your CdA from your position and weight. If you have measured it, enter your own number instead and every tool uses it. How the estimate works is in What is CdA.