Anemometer Anomaly Detection
An uncorrected nacelle anemometer can make a genuinely underperforming turbine look contractually compliant.
The challenge
Contractual availability and power curve warranties are settled against measured wind speed. If the nacelle anemometer reads high, the turbine appears to be producing what it should at a wind speed that never occurred — and a genuinely underperforming machine passes its test.
A claim never made is revenue never recovered. And because the error is in the reference instrument, no amount of scrutiny of the power data will reveal it.
The approach
The turbine's own output is the truth-teller. Power is a function of wind speed, and that relationship is stable enough that a sustained divergence between measured wind and the wind implied by production is evidence about the sensor, not the machine.
The method grades severity rather than raising a binary flag, and states the direction of the error — over-reading or under-reading — because the two have opposite commercial consequences.
- Validate measured wind against the wind implied by the turbine's own output
- Grade severity by impact rather than flagging pass/fail
- Identify direction — over-reading inflates compliance, under-reading hides degradation
- Correct the record without masking genuine underperformance
The outcome
A corrected wind reference, a severity-ranked list of suspect sensors, and a defensible position at the warranty conversation.
- Sensor error separated from turbine underperformance
- Severity graded and ranked, not flagged binary
- Direction of error stated explicitly
- Underpins the availability claim rather than undermining it
Curious what this would find in your portfolio?
A diagnostic runs across a representative slice and establishes the baseline before any build commitment.