In wind energy we measure energy well, but we measure wind badly
The natural turbulence of the wind means that it changes quickly over time and space, which makes it hard to measure. But I would take a wind measurement that was consistently bad, but we don’t even have that: the wind is measured inconsistently badly! What do I mean by that?
A modern turbine typically measures wind speed at a single point in space from an anemometer on top of the nacelle, and due to its location behind the rotor it requires some kind of calibration. Calibration means it has calibration parameters which can be changed. In practice these parameters change quite frequently, and often without the asset owner being aware of it.
How should you be measuring performance?
So if your primary method of wind turbine performance analysis or monitoring is to look at monthly (nacelle anemometer) power curves, you’ve either got a serial false positive generator or are just missing performance issues all together, probably both. At Bitbloom, we do track power curves in our performance and condition monitoring software platform Sift (the precise metric is actually “power curve efficiency in partial load” but ultimately comes down to the same dependency on wind speed measurements) and in addition:
- We back this up by an independent method to calibrate the wind speed measurements
- We validate with side-by-side tests, but most importantly
- This is a small part of what we do in performance monitoring
Are you happy with up to 5% energy yield loss going undetected?
Wind turbine performance monitoring is far more than just tracking power curves. It is a multi faceted approach of looking at many metrics covering turbine control systems, aerodynamic efficiency, curtailment losses and more. Many of the issues picked up by these analyses are costing multi-percentage point drops in your energy yield that would have gone undetected due to the uncertainty of power curve tracking.
You’re probably not losing that much on a fleet-wide average, but in our experience it usually is around 1%. And even if everything appears to be running optimally, wouldn’t you like the peace-of-mind that it actually is?
Is your farm performing optimally?
Bitbloom‘s experience shows that few farms perform optimally. Looking at power curves alone is not enough to capture most performance issues.
