Insights / Field note · WindBy Robert Sills CEng MIMechE30 September 2026

The loss hiding below cut in

OEM power curves show zero output below cut in. Six years of data from a 21 turbine wind farm show the real figure is negative, and over the hours a turbine sits idle it becomes a loss worth reporting.

← All insights

Monthly performance reports often close the gap between budget and actual production with a large bucket of other or unknown losses. Asset managers can do little with a number that has no cause attached to it.

Breaking that bucket down starts with a detail that is easy to miss. A wind turbine does not stop using power when the wind drops. Below cut in it sits idle, drawing power from the grid for its controller, yaw system, heaters and pumps.

Measured power curve for Hill of Towie from 2020 to 2025, density corrected, with 10 minute records in orange, the 0.5 m/s binned mean in navy and the OEM curve as a red dashed line. An inset zooms into 0 to 4.5 m/s and shows the measured curve sitting between minus 10 and minus 14 kW below cut in, where the OEM curve shows zero. — Density corrected measured power curve for all 21 turbines at Hill of Towie, 2020 to 2025, 10 minute SCADA. Normal operation and low wind idle records are kept. The inset shows the power below cut in.

Below cut in the power curve goes negative

The measured curve sits at roughly minus 10 to minus 14 kW below cut in, where the OEM curve shows zero.

OEM power curves are often used for energy yield assessments and for the expected power in monthly reporting. Both assume zero output below cut in. The difference per turbine is small, but it applies for a large share of the year.

Small numbers add up over idle hours

At Hill of Towie the turbines were idle in low wind for 18% of the time across six years.

The average draw while idle was about 11.5 kW per turbine.
Across 21 turbines this adds up to about 385 MWh a year, roughly 0.4% of production.
Low wind idle accounted for 87% of all the energy the site imported, with the rest drawn during fault downtime and site outages.
The idle share of time varied from about 15% to 21% between years, so the loss moves with the wind resource.

If reporting assumes zero power below cut in, this loss has no category. It ends up in the other losses bucket, where it cannot be separated from real underperformance.

Every 10 minute record needs a category

Putting a number on self consumption is one step towards a loss breakdown with no unexplained bucket at all.

The second chart shows the 2025 losses at Hill of Towie split into categories. Each 10 minute record is assigned to one cause, such as downtime, curtailment, idle or a site outage, and the loss against expected power is calculated for that record. Underperformance is then split further into anemometer bias, degradation and a small residual.

Pie chart of Hill of Towie production losses in 2025 as a share of total losses, largest to smallest. Site outage 33.3%, grid curtailment 26.0%, downtime 19.5%, anemometer bias 6.8%, degradation 5.7%, data gaps 3.3%, idle below cut in 2.7%, performance residual 1.9%, high wind 0.3%, internal curtailment 0.2% and icing 0.2%. — Hill of Towie production losses in 2025, each shown as a share of total losses. Underperformance is split into anemometer bias, degradation and a residual.

Site outages, grid curtailment and downtime make up about 79% of the losses. Idle below cut in is 2.7%, larger than high wind, icing and internal curtailment combined. Once underperformance is split, the residual that remains unexplained is 1.9% of the total.

Method notes. The analysis uses the open Hill of Towie SCADA dataset, 21 turbines at 10 minute resolution from 2020 to 2025. Wind speed is density corrected to the OEM reference density of 1.206 kg/m³. The power curve keeps normal operation and low wind idle records and excludes fault downtime, curtailment, icing, high wind events, site outages, start and stop intervals, bad data and power curve outliers. Keeping the idle records is not standard power curve practice, a test to IEC 61400-12-1 would exclude them, and it is done here to show the power below cut in. Idle is booked from turbines that are offline in low wind without a fault, so low wind hours with an active fault are counted as downtime. The degradation share uses a degradation rate measured from the site’s own historical data.

PowerVeritas categorises every 10 minute record and calculates the loss for each cause, including self consumption, as part of its Performance Reviews. This lets asset managers and owners see where production is lost and which losses they can control.

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