PowerVeritas
Insights / Field note · WindBy Robert Sills CEng MIMechE

Are wind turbines affected by the time of day?

The wind a turbine sees at midday is not the wind it sees at midnight, even at the same speed. A year of data from one turbine shows the night power curve sitting 8.8% lower between 5 and 9 m/s.

← All insights

Most power curve analysis corrects for air density and stops there. The wind a turbine sees in the day and at night is quite different, and not just in speed and temperature.

Splitting a year of data from one Hill of Towie turbine between the middle of the day (11:00 to 15:00 UTC) and the middle of the night (23:00 to 03:00 UTC) shows the difference clearly. Both windows fall on every day of the year, so season and air density are balanced between them.

Density corrected power curve for one Hill of Towie turbine in 2025, with day records from 11:00 to 15:00 UTC in orange and night records from 23:00 to 03:00 UTC in blue. The night binned mean sits below the day binned mean between about 4 and 12 m/s, crosses it near the knee of the curve and matches it above rated power.
Density corrected power curve for one Hill of Towie turbine in 2025, split between 11:00 to 15:00 UTC (day) and 23:00 to 03:00 UTC (night). Points are ten minute records and lines are 0.5 m/s binned means.

The night curve sits lower below rated

At 5 to 9 m/s the night curve sits 8.8% lower than the day curve. The gap closes towards rated power, and around the knee of the curve the night curve sits slightly higher. Above rated the two curves are the same, because the turbine is limiting its own output.

Density correction does not remove this. Night air is denser, and the wind speed has already been normalised to air density using the temperature measured on the nacelle.

Three secondary effects change between day and night

In the day the sun heats the ground and the air mixes. At night the ground cools, the lowest layer of air becomes stable and the wind behaves differently at the same speed. A ground LIDAR on the site shows three effects that follow this daily cycle.

Turbulence Intensity (TI), how gusty the wind is, is higher in the day. Higher TI increases power at lower wind speeds and reduces it around the knee of the power curve, because power rises faster than wind speed over the lower part of the curve. This is the largest of the three effects reducing performance at low wind speeds at night.

Shear, how wind speed changes with height, is stronger at night. Slower wind across the bottom of the rotor outweighs faster wind at the top, so the stronger night shear adds a small loss.

Veer, how wind direction changes with height, changes by only a couple of degrees between day and night, so its effect is small.

Three stacked panels against hour of day in UTC from the Hill of Towie ground LIDAR. Turbulence Intensity at hub height peaks at about 14% around midday and falls to about 10% overnight. The shear exponent falls to about 0.12 around midday and sits near 0.22 overnight. Veer is close to zero around midday and about 2° overnight. Shaded bands show the interquartile range and grey columns mark the day and night windows.
Turbulence Intensity at hub height (top), shear exponent between 38 and 99 m (middle) and veer over the same heights (bottom) by hour of day, from the site ground LIDAR, August 2025 to April 2026. Lines are medians and bands the interquartile range. Grey bands mark the day and night windows used in the power curve.

What the data shows

  • The night power curve sits 8.8% below the day curve at 5 to 9 m/s, after density correction.
  • Turbulence Intensity at hub height has a median of 10.1% at night and 14.3% in the day.
  • The shear exponent between 38 and 99 m has a median of 0.22 at night and 0.13 in the day.
  • Veer across the same heights is about 2° at night and close to zero in the day.

Why this matters for performance monitoring

A power curve built mostly from night data will look worse than one built mostly from day data, with nothing wrong with the turbine. The same applies when comparing seasons, sites or periods with a different mix of stable and unstable conditions. Environmental effects like these need to be ruled out before a turbine is flagged as underperforming.

Method notes. Turbine T07 at Hill of Towie, ten minute SCADA from January to December 2025, filtered to normal operation with fault downtime, curtailment, icing, start and stop, site outages and flagged sensor data removed. 1 to 4 December 2025 is excluded, when the turbine was briefly part of a wake steering trial. Wind speed is the nacelle anemometer normalised to air density using the nacelle temperature, because reanalysis temperature at 2 m misses the warmer air above the ground at night. The 8.8% weights each wind speed bin by how often it occurs. The LIDAR is a ZX300 about 1 km from the turbine, so its values describe the site rather than this turbine exactly, and it only covers August 2025 to April 2026. Only undisturbed wind directions are used. A scanning LIDAR reads turbulence differently from a cup anemometer, so its absolute values are approximate, but turbulence is lower at night in every month. The three effects are not quantified separately here, and together they are unlikely to explain the whole gap.

At PowerVeritas we specialise in understanding the factors that affect turbine performance, which helps our Performance Reviews identify genuine operational issues more accurately.

Charts and analysis by PowerVeritas. Where open datasets are used, sources are credited on the attributions page.