How we calibrate a nacelle anemometer in the field

Every turbine we instrument gets the same 40-minute check before its data counts towards anything. This is what happens in those 40 minutes, and why a bias of 0.3 metres per second is worth a climb.

The setup

A nacelle anemometer lives in the worst possible place to measure wind: directly behind the rotor, in flow that has just been slowed, twisted and turned turbulent by the blades. Manufacturers correct for this with a transfer function, but the function assumes the sensor is where the drawing says it is, mounted level, with the boom pointing the way it should.

Our engineer carries a reference anemometer that was calibrated in a wind tunnel within the last 90 days. It goes on a temporary boom two metres upwind of the installed unit, at the same height, and the two are logged side by side across at least three wind-speed bins.

The reference boom is mounted two metres upwind of the installed anemometer, at the same height.

What the 40 minutes record

The check produces a short, boring document, which is the point. It is the same document for every turbine, and it is the first thing an auditor asks for when a wind figure is questioned.

• Boom alignment and level, photographed before and after

• Side-by-side readings across at least three wind-speed bins

• Bias and scatter against the reference, with the reference's own calibration certificate

• The turbine identifier and the date, so the certificate history shows when the measurement changed

Why 0.3 m/s matters

Power goes with the cube of wind speed. At a typical site, a 0.3 m/s bias in the measured resource moves the expected annual energy by roughly 6%. That is the difference between a portfolio that appears to be under-performing and one that is quietly being curtailed.

If the installed sensor is outside tolerance, we do not correct it in software. We replace it, log the swap against the turbine identifier, and the certificate history records the day the measurement changed.