A wind turbine power curve relates wind speed to electrical output under specified conditions. It is a central part of converting a wind forecast into a power forecast. It is also a simplification: the same reported wind speed can be associated with different production because of atmospheric conditions, turbine control, measurement location and the state of the plant.
Read the main regions of the curve
Below the turbine's operating threshold, wind may be insufficient for generation. In the rising region, output increases as the machine extracts more energy from stronger wind. Around rated operation, control limits power near the turbine's rating. At high winds, protective control can reduce or stop generation according to the machine's operating behaviour.
The relevant thresholds and control behaviour are turbine-specific. Do not substitute a generic diagram for the manufacturer's documentation or assume every machine stops and restarts in the same way. A forecast must represent the fleet it serves, including any configuration or operating modes that affect output.
Why small wind errors can matter
In the rising part of the curve, a change in wind speed can produce a substantial change in power. Around the rated plateau, the same speed error may have less effect on production. Near a high-wind operating transition, timing and control behaviour may dominate. A uniform wind-speed score therefore does not fully describe the quality of the resulting power forecast.
Evaluate wind and power together when investigating errors. Group cases by operating region and examine transitions. A model that improves average wind speed can still mistime a power ramp if the relevant weather change arrives too early or too late. The target for scheduling remains power or energy at the agreed boundary.
Atmospheric conditions change the relationship
Air density affects the energy in the wind. Turbulence, wind shear and wind direction can also influence turbine behaviour and the wind experienced across the rotor. A single speed measurement at one point is not a complete description of the inflow. Site adaptation must consider which measurements and model variables are available.
Avoid applying corrections without checking their assumptions. A method designed for one measurement height or reference condition may not transfer unchanged to another turbine or site. Keep the reference density, measurement source and transformation method documented. Otherwise, an apparent model improvement may simply compensate for a mismatch that returns when conditions change.
A farm is more than a sum of ideal turbines
Wake effects can reduce the wind available to downstream machines, depending on direction and layout. Electrical losses, equipment availability, environmental operating modes and export restrictions can further change plant output. Multiplying one ideal curve by the turbine count will not automatically reproduce the export meter.
For a plant forecast, identify the measurement boundary and represent the fleet's current status. If turbines are unavailable, distinguish that fact from weak resource. If operation is curtailed, retain the instruction information rather than allowing the constrained measurements to define the unconstrained conversion curve.
Learn from measurements without learning the wrong target
A measured power curve can reveal how a site behaves, but its data needs quality control. Nacelle measurements may differ from free-stream conditions, and low production may reflect an outage rather than the turbine's normal aerodynamic response. Capacity changes or control updates can make historical observations describe a different asset.
Use status flags and appropriate filtering, while recording what was excluded. Fit on one dataset and assess on another. Examine direction-dependent behaviour and residual patterns instead of relying only on an overall fit. A flexible statistical curve can reproduce training data while failing to generalize to new weather conditions.
Ask for evidence at the power boundary
A provider evaluation should show the complete conversion chain from wind input to the delivered plant forecast. Request results by lead time, operating region and significant ramp event. Confirm how the system handles unavailable turbines, repowering and high-wind transitions. Ask whether its target is available production or expected metered export.
The power curve is best treated as an identifiable model component rather than an unquestioned truth. When its reference conditions, asset state and limitations are explicit, it becomes easier to separate weather errors from plant-conversion errors. That distinction supports better forecasting decisions and more useful conversations between asset, data and trading teams.