A turbine that is stopped is stopped: rotor locked, brake applied, and the anemometer on the nacelle reading whatever the wind is doing without it. The meter on turbine 11 goes on reading eighty kilowatts into the grid, and the grid pays for them.
I am a control room operator for an onshore wind farm. Twelve turbines, and I do four nights on and four off.
What the screens actually show, because it is more detailed than people expect.
Each turbine reports rotor speed, generator speed, pitch angle, nacelle position, wind speed from its own anemometer, brake status, and power output in kilowatts, once a second.
The power output on my screen is not the settlement figure. The settlement figure comes off a revenue meter at the substation, which is sealed, calibrated and read by the network operator, because that meter is what the money is paid on.
A stopped turbine draws a small amount from the grid. Maybe five or ten kilowatts for the controller, the heaters and the yaw system. It goes the other way on the meter, and that is normal and expected and in the monthly figures.
Turbine 11, stopped, brake applied, rotor locked, exports eighty kilowatts.
The state it is in is the whole of this, so here it is off the log, field by field, the way it comes up on the screen in front of me at three in the morning with eleven other turbines reporting alongside it.
Brake status: applied. Rotor speed: zero point zero. Generator speed: zero. Pitch: feathered, which means the blades are turned edge on so the wind does not push them.
And when I say stopped, I mean I can stand at the base of it. We do a manual stop for maintenance with a lock out and a tag on the panel, and I have walked out to 11 in the dark with the supervisor and looked up at three blades not moving.
It does it on about one night in five, between about one and four in the morning, and only when it is stopped.
When it is running it produces what the wind says it should, right along the power curve, and the curve for 11 is as good as any of them.
What has been done, and the owner has spent a lot because eighty kilowatts of unexplained export is a settlement problem before it is anything else.
The turbine's own instrumentation, checked and recalibrated twice, and the converter replaced in 2022 for an unrelated fault.
The current transformers at the turbine, which are what measure its output, swapped.
The communications, rebuilt, because a misaddressed data point is the sort of thing that makes one turbine report another's figures. The manufacturer's engineer spent three days on it and signed a report saying the data path is correct end to end.
And the revenue meter, which is the one that matters, because you could dismiss all the rest as software.
The revenue meter at the substation reads the whole farm. So the test was to stop every turbine except 11, and then stop 11 too, on a night in November 2022, with the network operator told in advance.
With twelve turbines stopped, the farm should import a hundred or so kilowatts.
It exported sixty eight.
Where that leaves the paperwork.
There is a formal dispute record from 2022 and 2023, and the outcome is the thing I find hardest.
The network operator's position is that the meter is correct, calibrated and sealed, and they are right: it has been independently verified twice.
The owner's position was that the farm cannot export with every turbine braked, which is also right.
The settlement stands. We are paid for that energy, and the sums are small, a few thousand a year, and the commercial team treated it as a metering quirk and moved on because the cost of pursuing it exceeds the money.
So a wind farm in this country sells about eleven megawatt hours a year of electricity that nothing in it generates, and the accounts balance, and the energy goes out onto a real network and lights real houses.
What I do in the room.
I log it. Turbine 11, stopped, exporting, and the figure and the duration, and it goes in the shift handover.
There is nothing to do. A turbine that is stopped and exporting is not a safety condition, nothing is hot, nothing is turning, and the protection has never operated.
The one thing I do not do any more.
The job includes a physical check on anything reporting an anomaly, and for two years I drove the site road and looked at 11 with a torch, which is the right thing to do and which I would tell a new operator to do.
In January 2023 I was there at about half two with the export reading seventy four kilowatts on the handheld.
A stopped turbine is loud in a way people do not expect. The tower hums, because the wind goes past a hundred metres of steel tube. You hear that from a long way off.
Standing at the door of 11 I could hear the hum of the tower and, under it, from up inside, the sound the nacelle makes when the generator is turning, which I have heard a thousand times from the same place with the blades going round over my head.
The blades were not going round. I stood there and watched them against the cloud for long enough to be certain, and they did not move, and the noise went on, and the figure on the handheld did not change.
I do the check from the gate now, with binoculars, and I write down that the rotor is confirmed stationary, which is true.