Smart derating adds years to wind farm life for under one percent energy loss

Wind farms are usually operated with a simple rule: every turbine runs at full power whenever the wind allows it. But turbines inside a farm do not age equally. Those sitting in the wake of their neighbors endure stronger turbulence and higher fatigue loads, quietly burning through their design fatigue budget years before the rest of the farm. A new study published in Wind Energy Science shows that this hidden imbalance can be exploited rather than endured. By letting individual turbines run at reduced power under carefully chosen wind conditions, operators can add two to five years of operational life to a wind farm while sacrificing less than one percent of its energy production — and, in many cases, actually increasing the farm’s net present value.

The research, carried out by Niklas Requate and Lukas Vollmer of Fraunhofer IWES in Bremerhaven together with René Hofmann of TU Wien, goes beyond the standard remedy known as wind-sector management. Under current practice, if a turbine’s fatigue budget is projected to be exceeded during nominal operation, the farm may be required to shut that turbine down entirely in certain wind directions, at the cost of significant energy loss. The new approach replaces this blunt instrument with a fine-grained optimization: instead of switching turbines off, it derates them — reducing available power by a specific percentage — only under those wind conditions where the trade-off between saved fatigue life and lost energy is most favorable.

Technically, the framework formulates wind farm operational planning as a large-scale nonlinear program. The continuous range of ambient conditions — wind direction, wind speed, and turbulence intensity — is discretized into bins, and for each turbine and each bin the optimizer selects a derating set point. An engineering wake model maps the ambient inflow and the chosen set points to the local conditions each turbine actually experiences, accounting for velocity deficits and wake-added turbulence from upstream machines. Polynomial surrogate models, fitted through damage-equivalent loads obtained from aeroelastic simulations, then translate those local conditions into per-hour power output and fatigue damage rates for selected failure modes. Lifetime energy and lifetime damage are computed by weighting the per-bin values with the expected frequency of each wind condition, derived from a 30-year ERA5 reanalysis time series in the German North Sea.

The scale of the problem is considerable. With wind direction binned at 10 degrees and wind speed at 1 meter per second across a 5 to 25 meter-per-second range, a nine-turbine farm already yields nearly 6,500 optimization variables, and a 100-turbine offshore farm would produce around 72,000. Because the problem is nonconvex — wake interactions and damage surrogates are nonlinear — the team selected the gradient-based interior point solver IPOPT, which scales far better than global deterministic methods or metaheuristics. To keep the problem tractable, rare wind conditions whose maximum influence on fatigue damage across all turbines and failure modes is below three percent are excluded and automatically set to full power.

The method was demonstrated on an example farm of nine 7.5 MW turbines with a rotor diameter of 164 meters, arranged in a regular 3-by-3 grid, under both onshore and offshore turbulence conditions. The team compared two solution strategies: a coordinated optimization of the entire farm in a single run, and an iterative approach in which each turbine is optimized separately while the wake field is held fixed and updated between iterations. The iterative strategy is easily scalable to large farms because each per-turbine problem has only as many variables as there are wind bins, and the problems can be solved in parallel.

The results are striking in their asymmetry. Onshore, where ambient turbulence is higher and wake-induced differences between turbines are smaller, extending the tower lifetime by two to five years costs only one to four percent of annual energy production and raises the net present value by up to ten percent. For the most heavily loaded turbine — the one in the center of the farm — twenty percent more lifetime was achieved at an energy loss of just five percent. Offshore, the picture is even more favorable. Because the lower ambient turbulence makes wake-added turbulence the dominant load driver, derating upstream turbines more effectively reduces the turbulence experienced by downstream machines. Extending the tower lifetime of the most highly loaded turbines by up to five years required derating only a few turbines, at energy losses below 0.2 percent, and increased the net present value by more than fifty percent.

The coordinated full-farm optimization proved particularly valuable offshore. When the target lifetime was balanced at 45 years, upstream turbines derated to their maximum level in certain wind directions, sparing downstream turbines with only four rotor diameters of separation from having to derate under all conditions. The per-turbine approach achieved similar results in most cases — the difference in annual energy production was below one percent almost everywhere — but it could not fulfill the lifetime constraint for three of the nine turbines at the 45-year target, because it cannot capture the benefit of upstream derating on wake turbulence. The full-farm approach yielded a slightly higher net present value in that demanding scenario.

The study also examined what happens when multiple failure modes are considered simultaneously. Turbulence-driven modes such as tower bottom bending and flapwise blade root bending respond well to wake-aware derating. But the edgewise blade bending moment is dominated by gravity and rotational speed rather than turbulence, making it far harder to manage. Extending the lifetime of all turbines and failure modes to 28 years required an average energy loss of about six percent, because the rotational speed must be reduced substantially to cut edgewise loads. Under these conditions the net present value fell by sixty percent compared to nominal operation, showing that the economic benefit of derating depends critically on which failure mode governs the farm’s end of life.

The authors are careful to note the limitations of their first-order analysis. The evaluation relies on steady-state wake models, which cannot represent dynamic effects, and on design-style damage estimation using damage-equivalent loads and linear damage accumulation, which does not capture sequence effects or material nonlinearities relevant to composites. The assumption that farm lifetime ends deterministically when the first turbine reaches its fatigue budget is strong and should not be applied directly to real projects. The economic evaluation, based on constant electricity prices, capital and operating expenditure, and a fixed weighted average cost of capital, is sensitive to price trajectories and discount rates. The authors recommend periodic re-planning with updated site data, measured fatigue consumption, and revised economic assumptions — effectively treating the framework as a very slow model-predictive control scheme applied throughout the farm’s lifetime.

Nevertheless, the practical output is remarkably simple: the optimizer delivers look-up tables of derating set points per turbine and per bin of wind direction and speed, compatible with most commercial turbines since derating is already standard for grid power tracking. In operation, wind direction and speed estimates from SCADA or lidar can feed the tables, with safeguards ensuring certification compliance and addressing set-point-specific risks such as negative damping offshore. The framework, implemented in the Python tool VIOLA together with the wake modeling software FOXES, is modular and scalable, and the authors see natural extensions in integrating electricity prices into the planning process, exploring participation in balancing markets, and combining derating with additional control channels such as yaw misalignment. For an industry facing aging fleets and rising repowering costs, the message is clear: the fatigue budget of a wind farm is not a fixed ceiling but a resource that can be redistributed — turbine by turbine, wind condition by wind condition — to squeeze years of additional revenue from assets already built and paid for.

Subject of Research: Beyond wind-sector management: optimal wind farm operational planning for balancing fatigue damage and extending lifetime

Article Title: Beyond wind-sector management: optimal wind farm operational planning for balancing fatigue damage and extending lifetime

Article References: Beyond wind-sector management: optimal wind farm operational planning for balancing fatigue damage and extending lifetime. (n.d.). https://doi.org/10.5194/wes-11-3863-2026

Image Credits: AI Generated

DOI: 10.5194/wes-11-3863-2026

Keywords: Beyond, wind-sector, management, optimal, wind, farm, operational, planning, balancing, fatigue, damage, extending