Wind farms: older turbines, wake losses and lifetime limits

The controller is not adapted after blade or converter changes

What happens. A turbine gets new blades or a new converter, and the controller keeps its old settings. The machine runs, but its response to wind, to gusts and to the grid is no longer what the settings were made for.

What it costs. Yield is lost, and loads on the structure are higher than they need to be. A mismatch of this kind is often found only when a fault, a noise or a worse result than expected turns up.

Why the usual fix fails. The original supplier may not support the old controller any more, and a general retuning by trial costs production time. Taking the settings from a sister turbine ignores the differences in the machine.

How a fixed-price remote diagnosis checks it. From operating data (power, rotor speed, pitch angle, loads where measured, grid events) and the data sheets of the new parts I build a turbine model and check how far the controller deviates from what it should be. The fixed-price remote diagnosis tells you whether retuning with a model is worth it for your turbines, and what it would involve.

Background: Control that adapts to drift and wear

Structures may see only limited loads if they are to run longer

What happens. An older turbine can run beyond its design life only if the structure has reserves left. To keep the permission, the loads it sees have to stay within a limit, and nobody knows how much of the reserve each year of operation has used.

What it costs. The choice is between output and lifetime. Run it as before, and the reserve runs out early; derate it to be safe, and the yield falls year after year.

Why the usual fix fails. A blanket derating takes the same percentage off every turbine, whatever site it stands on. A strain gauge on the tower tells you about one point, and many gauges fail.

How a fixed-price remote diagnosis checks it. From the turbine’s operating data (wind, power, rotor speed, pitch, accelerations) I look at whether the loads can be computed over the past years and for the next ones. The fixed-price remote diagnosis tells you whether load-limiting control, which reduces loads exactly where and when they matter, can keep a turbine in operation longer.

Background: Control that adapts to drift and wear

Upstream turbines take yield from those behind them

What happens. Every turbine leaves a wake of slower, turbulent air. The turbines behind it work in that wake and produce less, and each one still tries to optimise itself alone.

What it costs. The farm yields less than the sum of its machines, and the downstream turbines see more turbulence and more load. The loss depends on the wind direction and is largest in the directions where turbines stand in a row.

Why the usual fix fails. Every turbine running at its own optimum is not the optimum of the farm. Wake steering by hand-set offsets works on paper and can only be checked by long measurement series.

How a fixed-price remote diagnosis checks it. From the farm’s operating data (power per turbine, wind speed and direction, yaw angles) I look at how much power the wake costs and whether a different alignment of the front turbines would raise the total. The fixed-price remote diagnosis tells you whether wake steering is worth testing on your farm, and what the checks need.

Background: Economic predictive control

Curtailment for noise and bats costs output

What happens. Permits for many turbines require a curtailment at night, in certain wind speeds and temperatures, to protect neighbours or bats. The rule is usually a fixed table of wind speed, time and season.

What it costs. Output is given up on every night the rule applies, including nights when no bat is flying and no neighbour is affected. The rule is conservative on purpose, and the loss repeats every year.

Why the usual fix fails. Changing the table without evidence is a risk to the permit. Adding sensors or surveys is a cost of its own and does not tell the control what to do in the next hour.

How a fixed-price remote diagnosis checks it. From your curtailment records, the wind and temperature series and the permit conditions I look at how much output the rule costs and when. The fixed-price remote diagnosis tells you whether an optimised curtailment, which meets the permit’s protection aim with less lost yield, is feasible, and what proof the authority would need.

Background: Optimal paths, schedules and setpoint plans

Ice on the blades is detected late

What happens. Ice forms on the blades and changes the shape and the weight. The turbine loses power, runs with unbalance and may throw ice. Many turbines detect it late, by a power curve that falls away or by a sensor that is itself iced up.

What it costs. Downtime and lost yield, and safety risk for people near the turbine. A turbine that stops on suspicion also stops on days when it could have run.

Why the usual fix fails. A dedicated ice sensor is another part that fails in exactly the weather it is meant for. A rule on temperature and humidity stops turbines that have no ice and misses those that do.

How a fixed-price remote diagnosis checks it. From operating data (power against wind, rotor speed, pitch, vibrations, temperature) I look at whether ice can be told from other causes of a power deficit early enough to act. The fixed-price remote diagnosis tells you whether an estimate of ice from signals the turbine already has is reliable on your site.

Background: Computing unmeasured states for control

Faults are found only at inspection

What happens. Many faults in a turbine, such as a pitch drive that is slowly wearing, a sensor that drifts or a gearbox that gets rougher, are seen at the next inspection or after the turbine has stopped.

What it costs. Unplanned downtime, crane and crew called out at short notice, and consequential damage to parts that could have been saved. In older farms the monitoring is thin.

Why the usual fix fails. More inspections cost money and miss what happens between them. Alarm limits on single signals either ring all the time or only after the damage.

How a fixed-price remote diagnosis checks it. From the turbine’s operating logs (signals, alarms, stops, service reports) I look at which faults leave a trace in the data before they stop the machine. The fixed-price remote diagnosis tells you whether diagnosis from operating data, without new hardware, finds enough of them early to plan repairs.

Background: Control that adapts to drift and wear

A battery is added to the farm

What happens. A battery is installed at the wind farm, but the farm and the battery are run as two separate things. The battery follows its own price rules and the farm its own forecast.

What it costs. Value lost: the battery charges when the wind is already curtailed or discharges when the price is low, and the grid connection is not used fully.

Why the usual fix fails. Simple rules (“charge in surplus, discharge at the evening peak”) do not take the forecast error, the limits of the connection and the battery’s ageing into account at once. Buying a bigger battery is an expensive fix for a control problem.

How a fixed-price remote diagnosis checks it. From the farm’s production series, prices, forecasts and the connection terms I model the farm and the battery together. The fixed-price remote diagnosis tells you what a joint dispatch would have earned compared with your current operation, and whether an optimisation is worth putting into the controller.

Background: Economic predictive control

New turbines in old layouts

What happens. When old turbines are replaced by larger ones (repowering), the new machines often stand where the old ones did, and the layout was drawn for smaller rotors. Wakes reach farther, and neighbouring turbines sit in them more often.

What it costs. Wake losses cut into the extra output that the larger machine was bought for, and loads rise on turbines in the wake. The farm yields less than the data sheets promised.

Why the usual fix fails. Moving a turbine is limited by permits and land. Running every new turbine at its own optimum, as before, does not use what is known about the layout.

How a fixed-price remote diagnosis checks it. From the planned layout, the wind data of the site and, where available, the operating data of the old farm I model the wakes and test control strategies. The fixed-price remote diagnosis tells you what a layout and control optimisation would add, and whether it should enter the planning before the permit is final.

Background: Economic predictive control

Turbines beyond twenty years need proof of structural reserves

What happens. Turbines that leave the fixed feed-in support after about twenty years need an assessment of component fatigue and structural stability before they continue (TÜV NORD). At the same time, low market prices make many hours uneconomic.

What it costs. Output is given up when the turbine is stopped at low prices, and the remaining fatigue reserve is unknown. Without a convincing assessment the operator risks a stop, even if the structure could have run longer.

Why the usual fix fails. Fixed derating throws away yield on turbines that did not need it. A one-off assessment on paper says little about the loads each turbine saw in its own years of operation.

How a fixed-price remote diagnosis checks it. From the operating data of the turbine over its years I estimate the loads it has seen and the reserve that is left. The fixed-price remote diagnosis tells you whether load-reducing control and curtailment tied to prices can make continued operation pay, and what the data must show. I have worked on wind-turbine control at IAV (an automotive and energy engineering firm).

Background: Control that adapts to drift and wear

Reactive power becomes a market

What happens. Grid operators have to buy voltage support (reactive power) on a market, and the regions are being brought in step by step (50Hertz). Wind farms can earn for reactive power beyond what the connection rules ask.

What it costs. The revenue exists only if the farm delivers on request, within the limits of the converters and the grid code. A farm that treats reactive power as an afterthought leaves money on the table, or gets penalised for missing a promise.

Why the usual fix fails. Each turbine’s converter has its own reactive power set point, and the farm controller usually just divides the request. This ignores losses in the cables, the limits per converter and the interaction with active power.

How a fixed-price remote diagnosis checks it. From the farm’s electrical data (voltages, currents, converter limits, cable layout) I model what the farm can deliver at the connection point. The fixed-price remote diagnosis tells you whether a farm-level reactive power control and dispatch is worth putting in, and what the market conditions would require.

Background: Economic predictive control