Insights / Field note · Wind23 July 2026

Wind turbines feel the heat

It is not just humans that struggle with the heat. Two years of open SCADA data from a French wind farm show the same wind speed producing around 13% more power in cold air than in hot.

← All insights

High temperatures have a real impact on how much power a wind turbine can produce. The hotter the air, the lower its density. Less mass passing through the rotor means less energy at the same wind speed, roughly 3.4% less density for every 10 °C rise.

To see the effect, we plotted two years of 10-minute SCADA from ENGIE’s open dataset for the La Haute Borne wind farm in France, four Senvion MM82 turbines, with every point coloured by air temperature. The site saw everything from −6 °C to +38 °C.

Scatter plot of power against wind speed for four Senvion MM82 turbines, points coloured by air temperature from deep blue below zero to deep red above 30 degrees Celsius. Hot red points sit clearly below cold blue points through the steep part of the power curve. — Two years of 10-minute SCADA at La Haute Borne, France, coloured by air temperature. At the same wind speed, the hottest air produces visibly less power than the coldest.

What the data shows

The hot data points sit clearly below the cold ones through the steep part of the power curve, exactly where production is most sensitive to wind speed.
At 9 m/s the median output below 5 °C was around 1,150 kW against 1,010 kW above 25 °C, roughly 13% more power for the same measured wind speed.
The separation largely closes at rated power, where output is limited by the turbine’s rated-power control rather than by the energy in the wind.

Correct for it before judging performance

A density correction should be applied when assessing how a turbine is performing. The IEC 61400-12-1 approach adjusts the measured wind speed to a chosen reference density, here the standard 1.225 kg/m³, V_c = V × (ρ/1.225)^⅓, so that data from different air densities can be compared on one curve. Without it, a hot summer month can look like underperformance when much of the shortfall is expected atmospheric behaviour.

Applying that correction to this dataset brings the hot and cold data much closer together.

The same power curve scatter plot after density correction of the wind speed, with hot and cold coloured points now largely overlapping and only a small residual band of hot points below the cold ones. — The same data with wind speed corrected to 1.225 kg/m³, using SCADA temperature and ERA5 surface pressure adjusted to hub height. Most, but not all, of the separation disappears.

Method notes. Points shown are a temperature-balanced sample of the operational data with obvious curtailment and outliers removed. Density is computed from SCADA nacelle temperature with ERA5 surface pressure and dewpoint, corrected to hub height. The correction roughly halves the hot-cold separation rather than removing it entirely. The remaining separation may arise from seasonal differences in turbulence, shear and atmospheric stability, along with nacelle wind-speed measurement effects, a reminder that density is necessary but not sufficient when normalising a power curve.

The bigger heatwave risk, derating

On affected turbines, an often larger source of heat-related production loss is high temperature derating, where the control system reduces power below rated to protect thermally constrained components such as the generator from overheating.

Investigating repeated or excessive derating during a performance assessment can identify cooling system degradation or developing component faults early, and recover otherwise avoidable production losses.

PowerVeritas quantifies temperature effects, derating and the other real causes of lost production from operational SCADA data as part of its Performance Reviews.

Data, ENGIE La Haute Borne open SCADA, published under the Etalab Open Licence 2.0. ERA5 pressure via the Open-Meteo archive. Analysis and charts by PowerVeritas.