Wake steering deliberately changes an upstream turbine’s orientation to redirect wake effects and potentially improve the electrical output of other turbines. It can help in some wind conditions and reduce output in others.
The gold standard is a fair and independently verified plant-level comparison that measures achieved yaw, wind conditions and electrical energy rather than accepting a modelled peak gain as a guaranteed result.
eduKateSG · Wind Turbine Wake Steering Validation
Find your next learning step
Trace the farm objective, calculate pair-level effects and test what independent field trials actually establish.
Full chapter index · Practice and transfer · Education root · How X Works
Full chapter index
Define the farm objective · Chapters 1–2
Calculate measured effects · Chapters 3–4
Test wind and control · Chapters 5–7
Verify causal improvement · Chapters 8–10
Follow further reading · Chapters 11–12
CHAPTER 1 OF 12 · Define the farm objective
1. Wake steering changes the target from one turbine to the plant
An upstream turbine extracts momentum from moving air and leaves a slower, more turbulent wake. When another turbine lies downstream, its output can be affected. Wake steering deliberately changes the upstream machine’s orientation in a qualified control strategy to redirect some wake effects.
This is different from correcting an unintended yaw-misalignment error. Ordinary yaw tracking seeks an appropriate alignment; wake steering may intentionally introduce an offset as part of an authorized whole-farm optimisation.
An upstream turbine can lose some power while its neighbor gains. The relevant target is therefore aggregate output or another explicitly defined farm objective, not simply the first turbine’s instantaneous meter reading.
National-laboratory FLORIS resources describe engineering wake models used to explore such control strategies. Model predictions are useful hypotheses, not guaranteed energy gains in a real plant.
The gold standard specifies which turbines are considered, control mode, wind-sector definition, measurement period, electrical power boundary and safety approvals before evaluating improvement.
A research comparison does not provide permission to command yaw changes on operating machines. Turbine-specific safety envelopes and authorized controllers are outside the scope of this educational guide.
CHAPTER 2 OF 12 · Define the farm objective
2. Wind direction controls whether turbines interact
Wake steering is sensitive to where the wind arrives from. Two turbines may be closely aligned under one sector and largely independent under another, so the same yaw offset can have different consequences as direction changes.
Real wind direction varies during a ten-minute or shorter interval. A controller tracking an estimated direction has delays and uncertainty; the achieved yaw offset may be smaller than the requested command.
A field validation study published through national-laboratory research reports that measured yaw positions did not always attain desired offsets, with some unintended offsets outside the intended region. Those deviations affected actual power gains.
Define whether wind direction comes from a mast, lidar, nacelle vane or consensus estimator. These sensors sample different flows and can suffer calibration or turbine-induced distortion.
Use signed angular differences with a documented compass convention and wrapping at 360 degrees. An apparent 358-degree mismatch between 359-degree wind and 1-degree heading is really a two-degree separation under a common circular convention.
Checking the controller’s requested angle alone is insufficient. Validation needs the achieved orientation and an independent inflow reference wherever feasible.
CHAPTER 3 OF 12 · Calculate measured effects
3. Worked example: an upstream loss with a downstream gain
Take an invented turbine pair under a specified, steady, comparable wind condition. Without steering, turbine A produces 1.80 MW and turbine B produces 0.70 MW; their pair total is 2.50 MW.
Under a hypothetical authorized wake-steering state at the same matched inflow, turbine A produces 1.70 MW and B produces 0.90 MW. The steered pair total is 2.60 MW.
The upstream reduction is 1.80 − 1.70 = 0.10 MW, while the downstream gain is 0.90 − 0.70 = 0.20 MW. The net pair gain is 0.10 MW, or 0.10/2.50 = 4% of the baseline pair output.
That four-percent conditional effect does not establish four-percent annual generation gain. It assumes comparable inflow, no displaced output from other turbines, consistent metering and genuinely observed control states.
If the changed wake also affects a third machine, that machine must be included in an appropriate farm-level accounting boundary. A profitable-looking pair result can conceal a loss beyond the pair.
Treat these values solely as invented arithmetic. Real trials demand repeated, paired comparisons across wind regimes rather than one synthetic observation.
CHAPTER 4 OF 12 · Calculate measured effects
4. Worked example: weigh control effects across wind sectors
Consider two fictional mutually exclusive wind sectors that together represent a defined period. Steering is associated with a conditional farm-power increase of 0.10 MW in a favorable sector occupying 40% of hours and a decrease of 0.02 MW in the remaining sector occupying 60%.
Under an ideal assumption of fixed conditional differences, the weighted mean change is 0.40×0.10 + 0.60×(−0.02) = 0.040 − 0.012 = +0.028 MW.
Across exactly 1,000 comparable operating hours, the idealized energy difference would be 0.028 MW×1,000 h = 28 MWh. This calculation does not claim that one thousand hours represent a real year or that all hours are available.
The small negative effect outside the intended region matters. Field-validation research reports that unintended yaw offsets can reduce power outside the control region, offsetting some targeted gains.
Reporting only the best-performing direction bins can exaggerate results. A credible annual estimate integrates the relevant joint distributions of wind direction, speed, stability and availability.
Wind conditions and turbine control can be correlated with the sector itself. The measured difference should not simply be assumed causal without matched or controlled comparisons.
CHAPTER 5 OF 12 · Test wind and control
5. A wake model is a tool for predicting flow, not an independent meter
Engineering models such as FLORIS describe wind-farm interactions through simplified wake deficit, deflection, turbulence and superposition representations. They can guide candidate control regions, expected power changes and field-test planning.
Different wake models make different assumptions about terrain, atmospheric stability, wind shear, turbine thrust, directional variability and wake recovery. A fit to one configuration may not remain accurate for another.
A model may predict a desired yaw offset that cannot be achieved because the controller acts slowly or reacts to a distorted wind-vane signal. Compare commanded, achieved and independently measured offsets separately.
Published field experiments compare measured turbine-pair gains with FLORIS predictions that incorporate wind-direction and yaw uncertainty. Their results demonstrate that modeling the variability can improve explanation of the trial, not that every farm must see the same gain.
Record FLORIS model version, turbine layout, rotor diameters, power and thrust curves, ambient turbulence, inflow distribution and optimization objective. Model provenance is part of scientific reproducibility.
Use the model as a predeclared prediction to test, rather than choosing the most favorable assumptions after observing a small measured gain.
CHAPTER 6 OF 12 · Test wind and control
6. Lidar and SCADA offer complementary but imperfect evidence
SCADA records can include electrical output, yaw position, vane-derived wind direction, pitch, rotor speed and turbine status. These operational signals are valuable but may share measurement errors with the controller under evaluation.
A nacelle-mounted Doppler lidar can provide an upstream line-of-sight wind observation. Reconstructing a wind vector may still depend on scan geometry, homogeneity and the relationship between the measured volume and turbine inflow.
Field-validation studies report differences between yaw offsets inferred from nacelle vanes and those inferred using independent lidar or sodar references. The disagreement helps explain why observed steering performance may fall short of ideal predictions.
Synchronize observations at appropriate time resolution and retain sensor uncertainty, availability and invalid-data flags. An attractive paired scatterplot is not enough if wind conditions and operating modes differ.
Nacelle-mounted instruments, nearby met masts and neighboring turbine estimators sample different flows. Where possible, use independent measurements to challenge a controller’s internal estimate rather than confirming it with the same biased signal.
Explicitly mark missing lidar scans, icing, maintenance, curtailment and communication gaps. Filtering out difficult data without disclosure can create misleading validation.
CHAPTER 7 OF 12 · Test wind and control
7. Wind speed and atmospheric stability change the opportunity
The benefit of wake steering varies with wind speed because turbine power response, rotor thrust and downstream losses change across the operating curve. Atmospheric turbulence and stability affect how quickly wakes mix back into ambient flow.
Research on a commercial wind-plant experiment found that performance benefits varied across speed bins, including limited improvements in some intermediate ranges and stronger effects in others. Such findings cannot be transplanted numerically to another plant.
Near cut-in speed, small changes in downstream flow can alter whether a turbine generates at all. Near rating, the relation between wind speed and electrical output becomes less sensitive because turbine control can limit power.
A controller that improves net power at one speed may generate no benefit or a penalty at another. Optimisation must consider joint wind-speed and directional distributions and maintain appropriate safety constraints.
Stratify measured results by direction, speed and where possible stability or turbulence. Report sample counts and confidence intervals in each group.
Separating wake-steering performance from the site wind climate is essential: annual energy improvement is not simply a mathematical property of a yaw offset.
CHAPTER 8 OF 12 · Verify causal improvement
8. Field validation needs a fair control or reference period
A rigorous trial must compare results under appropriate counterfactual conditions: what would the farm have produced if steering had not been active during the same inflow? This cannot be observed directly for every minute, so a design must estimate it.
Possible approaches include alternating approved control modes, comparable reference turbines, matched meteorological bins and predefined analytic corrections. Each has assumptions about weather trends, interference and operational changes.
Researchers should decide exclusion criteria before looking at which intervals yielded more power. Faulted turbines, curtailment and atypical test modes must be handled consistently for both control and treatment.
Consecutive observations within a weather system are not independent. Statistical confidence should respect temporal dependence, and turbine-to-turbine interactions can also make nearby observations correlated.
Keep a time-stamped record of commanded yaw, achieved yaw, wind estimates, power meters, status flags, control transitions and predeclared analysis code. This record permits a later auditor to understand how the comparison was made.
Evaluate the target farm or turbine-pair output rather than cherry-picking whichever machine showed a gain.
CHAPTER 9 OF 12 · Verify causal improvement
9. Worked example: distinguish average uplift from uncertainty
Suppose an imaginary, prespecified trial estimates mean farm uplift at +1.5% compared with a matched baseline, with a 95% uncertainty interval extending from −0.4% to +3.4%. This is an invented illustrative statistical summary.
The point estimate is positive, but the interval includes zero and negative values. It would therefore be misleading to report the study as conclusive proof of increased generation at that confidence level.
Another hypothetical experiment might estimate +1.0% with an interval of +0.5% to +1.5%, supporting a positive average effect under its model and sample. It could still fail to generalise to other farms or different control settings.
The validity of either interval depends on data quality, dependence-aware uncertainty estimation and a credible counterfactual. Precision of displayed percentages cannot substitute for independent evidence.
Report both conditional sector-level effects and the integrated result, together with the definition of energy delivered and the limitations of the evaluation period.
An energy gain is not necessarily a financial gain after control, monitoring, maintenance and availability consequences are included.
CHAPTER 10 OF 12 · Verify causal improvement
10. Wakes, structural loading and authorized operation
Intentional yaw can change not only average electrical output but also aerodynamic loading and the trajectory of wake turbulence. A controller optimized solely for instantaneous farm power may not satisfy all fatigue, maintenance or safety objectives.
Research models can represent load-related constraints, but the measured structural response and turbine manufacturer’s operating provisions remain essential for real deployment.
Distinguish validation of a scientific hypothesis from approval to implement a controller. Real yaw-offset schedules, actuator settings, safety interlocks and maintenance procedures are outside the remit of this educational guide.
An honest benefit assessment considers which turbines experience changed loads, whether changes persist in variable wind, and how downtime or reliability may affect useful energy.
The most convincing experimental record combines conditional electrical measurements, achieved control states, independent wind evidence and an explicit statement of what loading or reliability questions remain unresolved.
Practice A: Two turbines produce 1.80+0.70 = 2.50 MW without steering and 1.70+0.90 = 2.60 MW with it. Verify a conditional 4% pair improvement and describe what other turbines might need to be counted.
Practice B: A steering strategy has +0.10 MW for 40% of comparable hours and −0.02 MW for 60%. Calculate +0.028 MW weighted mean and +28 MWh over 1,000 such hours.
Practice C: A trial claims +1.5% average uplift with a 95% interval −0.4% to +3.4%. Explain why it does not establish an unambiguous positive effect.
FAQ — Is intentional wake steering the same as correcting a faulty yaw vane? No. One is an authorized plant-level control strategy; the other is a sensor or tracking calibration issue.
FAQ — Does a modelled wake deflection prove energy gain? No. Output must be compared against a valid baseline under relevant conditions.
FAQ — Is the biggest sector gain the annual energy gain? No. Annual effects require all wind sectors and hours, including off-target penalties.
Continue through the connected wind-energy, lidar and statistical roots for a complete explanation of how a claimed performance improvement becomes defensible.
Connected, verified eduKateSG roots for the different physical and measurement questions:
- The Gold Standard of Wind Farm Wake Modelling — the physics and assumptions behind predicted deflection
- The Gold Standard of Wind Turbine Yaw Misalignment Measurement — observed versus commanded orientation
- The Gold Standard of Wind Turbine Power Curves — conditional electrical output comparison
- The Gold Standard of Wind Lidar — upstream wind observations
- The Gold Standard of Wind Energy — larger energy and reliability objectives
- The Gold Standard of Atmospheric Science — turbulence and flow variability
- The Gold Standard of Statistics — counterfactual comparisons and uncertainty
National laboratory field trials and FLORIS documentation demonstrate why achieved yaw offsets, direction variability and conditional wind speed matter. They provide specific research cases, not universally guaranteed gains.
All turbine powers, sector frequencies, trial results and confidence intervals above are fictional teaching data. The article is not a controller deployment plan or a claim about a named wind farm.
A next-step validation worksheet should include turbine layout, independent inflow reference, controller status, predeclared baseline, achieved yaw angles, power meters, exclusion rules, correlated uncertainty, sector results and overall energy boundaries.
Primary research and national-laboratory references:
- NLR/NREL — Field Validation of Wake Steering — achieved offsets, variability and measured power
- National Laboratory — FLORIS Wind Farm Optimization — engineering wake simulation context
- National Laboratory — Wake-Steering Field Validation — paired-power trends and yaw uncertainty
- National Laboratory — Commercial Wind Plant Experiment — wind-speed-dependent field outcomes
- Wind Energy Science — Field Experiment Article — field validation and conditional gain
The immutable Clementi Secondary 1 Mathematics reference illustrates the educational sequence, not wind-turbine control or safety authorization.