Pith. sign in

REVIEW 3 major objections 4 minor 1 cited by

Reinforcement Learning Increases Wind Farm Power Production by Enabling Closed-Loop Collaborative Control

T0 review · 3 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read A reinforcement learning controller that yaws wind turbines every 10 seconds raises simulated farm power by 4.30%, nearly doubling the 2.19% gain from static optimal yaw angles.

desk verdict A credible first demonstration of RL-based dynamic yaw control in LES, but the replay test does not fully establish the closed-loop claim without an open-loop dynamic baseline. read the letter →

arxiv 2506.20554 v1 pith:CMV65DK5 submitted 2025-06-25 physics.flu-dyn cs.LGcs.SYeess.SY

Andrew Mole , Max Weissenbacher , Georgios Rigas , Sylvain Laizet This is my paper · ORCID
classification physics.flu-dyncs.LGcs.SYeess.SY
keywords windfarmcontrolwakesteeringreinforcementlearninglarge-eddysimulationclosed-loopyawBayesianoptimization
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper argues that wind farm control does not have to be static: a reinforcement learning agent trained directly against a high-fidelity large-eddy simulation of a three-turbine farm learns to yaw each turbine in response to live velocity measurements and raises mean farm power by 4.30% over greedy operation, nearly double the 2.19% gain from the best static yaw angles found by Bayesian optimization. The key claim is that the extra gain is genuinely closed-loop: replaying the controller's recorded yaw sequence on statistically independent inflows produces no gain, which shows the improvement depends on sensing and reacting to the flow rather than executing a fixed pattern. If the result is right, it shifts wind farm optimization from steady-state wake-model tuning toward dynamic, flow-responsive coordination, and the trained controller is cheap enough to evaluate in real time. The authors state this is the first successful demonstration of an efficient dynamic closed-loop reinforcement-learning controller for wind farm control.

What carries the argument

The load-bearing machinery is the coupling of a high-fidelity large-eddy simulation with actuator disk turbine models to a Soft Actor-Critic (SAC) reinforcement learning policy, an algorithm that adds entropy regularization to balance exploration and exploitation. The policy is a multi-layer perceptron that maps 231 hub-height streamwise velocity sensors plus the previous yaw angles to yaw angular velocities, updating every 10 seconds with a limit of 1 degree per second and a maximum yaw of ±40 degrees. The reward is instantaneous farm power minus a large-angle penalty. The decisive verification device is the action-replay test: executing the recorded yaw sequence in fresh environments with independent inflow removes the power gain, showing the learned control is feedback-driven rather than a fixed open-loop forcing pattern.

What would settle it

Re-run the same training and evaluation with actuator-line turbine models (which resolve blade forces) instead of actuator disks at the same grid spacing, and compare the RL controller's mean power gain to the static Bayesian optimum's gain; if the closed-loop controller no longer nearly doubles the static gain, the claim that dynamic feedback is the source of the improvement is falsified.

Watch

Extended reading notes

Core claim

The paper reports that a Soft Actor-Critic RL agent, controlling the yaw of three in-line turbines separated by five rotor diameters, learns to alternate between two near-mirror-image yaw modes, $\vec{\alpha}^+_{RL}=(23^\circ,12^\circ,-1^\circ)$ and $\vec{\alpha}^-_{RL}=(-22^\circ,-8^\circ,7^\circ)$, which are close to the static Bayesian optimum and its reflection. Cross-correlation of the yaw signals shows the controller has learned the advection delay between turbines, and joint densities with the lateral barycenter of the upstream velocity show it yaws against incoming flow asymmetries, steering high-momentum gusts toward downstream turbines. Across 224 evaluation episodes the RL controller produces a 4.30% mean power gain (95% CI [4.10, 4.49]) versus 2.19% (95% CI [1.98, 2.39]) for static Bayesian optimization. The closed-loop nature is established by an action-replay test: the stored action sequence, replayed on independent inflow conditions, gives no statistically significant gain (95% CI [−0.35, +0.96]). The paper therefore claims that dynamic flow-responsive wake steering, not just optimized steady angles, is what nearly doubles the achievable power increase.

Load-bearing premise

The load-bearing premise is that the simulated turbine model—each rotor represented as a porous disk in a large-eddy simulation with a grid spacing of one-tenth of the rotor diameter—faithfully reproduces the wake recovery and transient power response that occur in real wind farms, since every quantitative claim in the paper comes from this simulator.

Editorial extensions

If this is right

  • The trained controller runs in real time at essentially no compute cost, so the expensive supercomputer-based training is a one-time investment; deployment only needs upstream flow sensing and yaw actuation within about 1 degree per second.
  • The learned strategy deliberately sacrifices first-turbine power to redirect high-momentum gusts toward downstream turbines, implying that greedy per-turbine operation leaves a repeatable power gain of about 4% on the table for this three-turbine configuration.
  • The controller's dominant actuation frequency ($\mathrm{St}\approx 2\times 10^{-2}$) and reduced yaw rates (below 0.4 degrees per second after filtering) sit within ranges likely achievable by current commercial turbines, so the dynamic strategy is not obviously infeasible mechanically.
  • Because the replayed action sequence yields no gain, any practical deployment must preserve the closed loop: open-loop or scheduled yaw patterns, even with optimal static angles, forfeit most of the benefit.
  • The static Bayesian optimum found in this study—upstream turbines yawed more, downstream less—is consistent with the controller's two switching modes, suggesting the dynamic policy simply modulates between known good setpoints based on instantaneous flow asymmetry.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Editorial inference: If the switching-between-two-optima mechanism is the true source of the gain, a much simpler hand-designed controller—thresholding on the upstream velocity barycenter to pick between the two static yaw modes—might reproduce most of the 4.30% gain without any neural network; this is directly testable in the same LES environments.
  • Editorial inference: The paper's own frequency analysis implies the controller is tracking large-scale, slowly varying gusts rather than small-scale turbulence; a plausible extension is that training on a wider ensemble of wind speeds and turbulence intensities would produce a policy whose gain generalizes beyond the single inflow condition studied.
  • Editorial inference: The authors' suggestion to constrain a dynamic controller to switch among finitely many static optima could be formalized as a hybrid supervisory control problem, reducing both training cost and the risk of mechanically damaging yaw actuators.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. The manuscript presents an RL-based yaw controller for a three-turbine, in-line wind farm simulated with the Winc3D large-eddy simulation solver. The controller receives upstream velocity-sensor measurements and updates yaw angles every 10 s, and it is trained with the SAC algorithm across 32 parallel LES environments. Evaluated on 224 statistically independent episodes, the RL policy yields a mean power increase of 4.30% relative to greedy operation, compared with 2.19% for a static yaw configuration found by Bayesian optimization. The authors use a closed-loop replay test, cross-correlation analysis, spectral analysis, and flow-field visualizations to argue that the gain arises from closed-loop switching between two near-symmetric yaw modes in response to turbulent gusts.

Significance. If the quantitative claims survive scrutiny, this is a valuable demonstration that RL can be coupled directly to high-fidelity LES for dynamic wind farm control, with a careful statistical evaluation protocol and openly available code. The replay test is a thoughtful attempt to distinguish closed-loop from open-loop behavior, and the independent-inflow evaluation with confidence intervals is a strength. However, two load-bearing aspects require attention: the replay test does not exclude the possibility that a well-chosen open-loop dynamic schedule achieves a large fraction of the gain, and the reported headline improvement is susceptible to selection bias because hyperparameters were chosen by sweeps that used evaluation power as the selection criterion.

major comments (3)
  1. [§2.3] The replay test does not establish that the 4.30% gain requires closed-loop feedback. Replaying a single action trajectory from the converged policy on statistically independent inflows shows only that this particular trajectory is not portable to different inflow conditions; it does not compare against an open-loop dynamic controller optimized over the inflow distribution. Concretely, one would need to evaluate a periodic schedule that switches between the two modes α_RL^+ and α_RL^- identified in §2.4 and Figure 3, with frequency and phase swept, or train an RL policy with the same action space but no sensor observations. The authors' own Discussion item 1 suggests that switching between static optima is a plausible alternative design, which further highlights the need for this control. The statement that 'accurate sensor measurements are required' is therefore too strong, and the title's 'Closed-Loop Collaborative Control' claim is not yet fully supported.
  2. [§2.1, Appendix Table 1 and Figure 11] The headline 4.30% is likely optimistically biased because the SAC hyperparameters (discount factor, actor/critic learning rate, entropy learning rate, initial entropy multiplier, and number of velocity probes) were selected by sweeps that used evaluation power as the selection criterion. The reported 95% confidence interval [4.10%, 4.49%] treats the policy as a pre-specified estimator rather than as the best of many configurations examined. To place the RL result on equal statistical footing with the BO baseline, the authors should either report the total number of hyperparameter configurations evaluated and compute the final estimate on a holdout set, or apply a multiple-comparisons correction. Without this, the comparison '4.30% versus 2.19%' may overstate the true advantage of the RL approach.
  3. [§2.5, Figure 4(c)(d)] The bandwidth-limiting study replays low-pass filtered versions of the same yaw trajectories on inlet conditions identical to those used to generate the trajectories. This setup can demonstrate that the learned policy contains a dominant low-frequency component, but it cannot provide an unbiased estimate of the power that a bandwidth-limited closed-loop controller would achieve on independent inflows. The claim that the controller operates at an effective frequency below 3.125×10^-3 Hz and with yaw rates below 0.4°/s is presented as a practical feasibility result; it should be either evaluated on independent inflow conditions or explicitly qualified as a property of the specific replayed trajectory.
minor comments (4)
  1. [§4.3.2] The sentence 'SAC is a model-free RL algorithm without policies' appears to contain a typo; presumably the intended meaning is that SAC is an off-policy algorithm or that it does not require on-policy updates.
  2. [References] References [18] and [19] are the same paper (Font et al., 2025, Deep reinforcement learning for active flow control in a turbulent separation bubble) and should be merged or renumbered.
  3. [§2.5] The statement that the maximum yaw velocity is reduced 'from 1°/s to below 0.4°/s' is partly by construction, since 1°/s is the imposed action limit; the more informative quantity is the effect of the filter on the resulting power, which is currently only shown for the non-independent replay.
  4. [§4.3.3, Eq. (14)] The angle-penalty term is described as 'large', but the chosen values λ=0.5 and κ=26 are not varied in the reported sweeps; a sentence justifying this choice or citing a sensitivity test would be useful.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the RL power-gain result is an independently evaluated simulation outcome, not a fitted input renamed as a prediction.

full rationale

The paper's central quantitative claims are empirical simulation results rather than derived quantities. The RL policy is trained to maximize discounted wind-farm power inside the Winc3D LES environment, and the reported 4.30% mean increase is computed from 224 evaluation episodes across 16 environments with inlet and atmospheric boundary conditions independent of the training data. The static Bayesian-optimization baseline is optimized in the same simulator, which is the correct point of comparison, and the resulting 2.19% gain is not fed into the RL training or used as a target for the RL improvement. The closed-loop replay test compares a saved action sequence on statistically independent inflow realizations; this is a control experiment intended to test the role of sensing, not a prediction derived from a fitted parameter. Even if the replay test is not a sufficient control against all possible open-loop dynamic schedules, that is a limitation of the inference, not a circularity: the paper does not define the closed-loop claim in terms of the replay outcome, nor does it fit the replayed gain as an input. The LES solver and actuator disk model are supported by prior validation citations, including work by the same group, but those citations are independent evidence for the flow solver's fidelity and are not used to define the headline result. No equation in the paper defines the target quantity in terms of itself, and no fitted parameter is renamed as a prediction. The analysis therefore displays no step in which a 'prediction' reduces by construction to its inputs.

Assumptions & free parameters 9 free parameters · 6 assumptions · 0 invented entities

The central simulation result rests on the LES/actuator-disk setup and a set of hand-tuned RL hyperparameters, especially the reward penalty (lambda, kappa) and control bounds. No new physical entities are introduced. The static BO baseline is itself an optimization over the same simulator, so the comparison is self-consistent but not validated externally.

free parameters (9)
  • Reward angle-penalty coefficient lambda = 0.5
    Hand-chosen weight in the reward (Eq. 14); the policy maximizes a proxy, not raw mean power, so lambda shapes the learned control objective.
  • Reward angle-penalty exponent kappa = 26
    Hand-chosen exponent in Eq. (14); makes the penalty negligible at small angles and sharp near alpha_max.
  • Maximum yaw velocity v_max = 1.0 deg/s
    Action bound on yaw rate (Sec. 4.3.3), imposed for numerical stability and practical feasibility.
  • Maximum yaw angle alpha_max = 40 deg
    Bound on yaw angle in both BO search and RL actions; restricts the reachable policy set.
  • RL control interval dt_RL = 10 s
    Time between yaw updates; selected in hyperparameter sweeps, determines the closed-loop bandwidth.
  • Velocity sensors per turbine = 77 (231 total)
    Observation count from sweeps; more sensors were found to improve performance, directly shaping the state vector.
  • SAC discount factor gamma = 0.99
    With gamma<1 and lambda>0 the optimized return is only a truncated/penalized approximation of mean farm power (Sec. 4.3.3).
  • Initial entropy multiplier alpha_init = 10
    SAC exploration temperature tuned in sweeps to avoid premature convergence.
  • BO UCB exploration parameter beta = 4 (annealed to lower values)
    Exploration-exploitation knob for the static baseline optimizer (Sec. 4.4).
assumptions (6)
  • domain assumption Actuator disk model with uniform thrust and 1D momentum theory represents turbine power and far-wake dynamics.
    Eqs. (6)-(7); does not resolve blades or near wake, but paper cites prior validation (Bempedelis et al. 2023; Revaz and Porte-Agel, 2021). All power numbers depend on this model.
  • domain assumption Smagorinsky subgrid model with Mason-Thomson damping adequately captures sub-filter stresses in the ABL.
    Eqs. (4)-(5); standard LES closure, not validated for this specific control case.
  • domain assumption Grid spacing D/10 is sufficient to resolve the wake and power response relevant to yaw control.
    Section 4.1; consistent with prior studies but no grid-convergence study is provided for the dynamic control setup.
  • domain assumption The mean inflow is statistically symmetric in the lateral direction, so yaw configurations alpha and -alpha yield equal mean power.
    Section 2.4 ("Since the mean flow through the turbines is statistically symmetric..."). Used to interpret the RL switching behavior; lateral inhomogeneity or veer would break this.
  • standard math The MDP/SAC formulation is a valid and sufficient model for optimizing closed-loop yaw control.
    Section 4.3; standard RL framework, no new theory required.
  • domain assumption The modified precursor recycling (domains 224D long and lateral shift by one-eighth width) removes phase-locked periodic forcing from the inflow.
    Section 4.2.2; authors state this was necessary to prevent the RL controller from phase-locking to periodic inlet artifacts; the independent-inflow evaluation relies on this.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Reinforcement Learning Increases Wind Farm Power Production by Enabling Closed-Loop Collaborative Control." pith.science (2026). https://pith.science/paper/CMV65DK5

@misc{pith2026250620554,
  author       = {Pith},
  title        = {Pith review of: Reinforcement Learning Increases Wind Farm Power Production by Enabling Closed-Loop Collaborative Control},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/CMV65DK5}},
  note         = {Machine review of arXiv:2506.20554}
}
read the original abstract

Traditional wind farm control operates each turbine independently to maximize individual power output. However, coordinated wake steering across the entire farm can substantially increase the combined wind farm energy production. Although dynamic closed-loop control has proven effective in flow control applications, wind farm optimization has relied primarily on static, low-fidelity simulators that ignore critical turbulent flow dynamics. In this work, we present the first reinforcement learning (RL) controller integrated directly with high-fidelity large-eddy simulation (LES), enabling real-time response to atmospheric turbulence through collaborative, dynamic control strategies. Our RL controller achieves a 4.30% increase in wind farm power output compared to baseline operation, nearly doubling the 2.19% gain from static optimal yaw control obtained through Bayesian optimization. These results establish dynamic flow-responsive control as a transformative approach to wind farm optimization, with direct implications for accelerating renewable energy deployment to net-zero targets.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Three-dimensional Navier-Stokes-Biot coupling via a moving reticular plate interface: existence of weak solutions

    math.AP 2025-08 unverdicted novelty 6.0 of 10

    A regularized three-dimensional Navier-Stokes-Biot fluid-structure problem with a moving permeable plate interface is shown to admit finite-energy weak solutions.

Reference graph

Works this paper leans on

75 extracted references · 70 canonical work pages · cited by 1 Pith paper

  1. [1]

    URL https://arxiv.org/abs/1910.06403, 1910.06403

    Balandat M, Karrer B, Jiang DR, et al (2020) Botorch: A framework for efficient monte-carlo bayesian optimization. URL https://arxiv.org/abs/1910.06403, 1910.06403

    arXiv 2020

  2. [2]

    Wind Energy 13(6):573–586

    Barthelmie RJ, Jensen LE (2010) Evaluation of wind farm efficiency and wind turbine wakes at the Nysted offshore wind farm. Wind Energy 13(6):573–586

    work page 2010

  3. [3]

    Wind Energy 12(5):431–444 17

    Barthelmie RJ, Hansen K, Frandsen ST, et al (2009) Modelling and measuring flow and wind turbine wakes in large wind farms offshore. Wind Energy 12(5):431–444 17

    work page 2009

  4. [4]

    SoftwareX 12:100550

    Bartholomew P, Deskos G, Frantz R, et al (2020) Xcompact3d: An open-source framework for solving turbulence problems on a cartesian mesh. SoftwareX 12:100550

    work page 2020

  5. [5]

    Renewable Energy 70:116–123

    Bastankhah M, Port´ e-Agel F (2014) A new analytical model for wind-turbine wakes. Renewable Energy 70:116–123

    work page 2014

  6. [6]

    URL https: //doi.org/10.5281/zenodo.14507040

    Beckett G, Beech-Brandt J, Leach K, et al (2024) Archer2 service description. URL https: //doi.org/10.5281/zenodo.14507040

    doi:10.5281/zenodo.14507040 2024

  7. [7]

    Journal of Fluid Mechanics 955:A12

    Bempedelis N, Laizet S, Deskos G (2023) Turbulent entrainment in finite-length wind farms. Journal of Fluid Mechanics 955:A12

    work page 2023

  8. [8]

    Wind Energy Science 9(4):869–882

    Bempedelis N, Gori F, Wynn A, et al (2024) Data-driven optimisation of wind farm layout and wake steering with large-eddy simulations. Wind Energy Science 9(4):869–882

    work page 2024

Show all 75 references
  1. [9]

    2306.00577

    Bou A, Bettini M, Dittert S, et al (2023) Torchrl: A data-driven decision-making library for pytorch. 2306.00577

    2023 arXiv

  2. [10]

    Bui VH, Nguyen TT, Kim HM (2020) Distributed Operation of Wind Farm for Maximizing Output Power: A Multi-Agent Deep Reinforcement Learning Approach 8

    2020

  3. [11]

    Physics of Fluids 22(1):015110

    Calaf M, Meneveau C, Meyers J (2010) Large eddy simulation study of fully developed wind- turbine array boundary layers. Physics of Fluids 22(1):015110

    2010

  4. [12]

    In: Kuerten H, Geurts B, Armenio V, et al (eds) Direct and Large-Eddy Simulation VIII

    Calaf M, Meneveau C, Parlange M (2011) Large eddy simulation study of a fully developed thermal wind-turbine array boundary layer. In: Kuerten H, Geurts B, Armenio V, et al (eds) Direct and Large-Eddy Simulation VIII. Springer Netherlands, Dordrecht, pp 239–244

    2011

  5. [13]

    Physical Review Fluids 9(4):043902

    Chatzimanolakis M, Weber P, Koumoutsakos P (2024) Learning in two dimensions and control- ling in three: Generalizable drag reduction strategies for flows past circular cylinders through deep reinforcement learning. Physical Review Fluids 9(4):043902

    2024

  6. [14]

    Nature 602(7897):414–419

    Degrave J, Felici F, Buchli J, et al (2022) Magnetic control of tokamak plasmas through deep reinforcement learning. Nature 602(7897):414–419

    2022

  7. [15]

    Wind Energy

    Deskos G, Laizet S, Palacios R (2020) WInc3D: A novel framework for turbulence-resolving simulations of wind farm wake interactions. Wind Energy

    2020

  8. [16]

    Applied Energy 292:116928

    Dong H, Zhang J, Zhao X (2021) Intelligent wind farm control via deep reinforcement learning and high-fidelity simulations. Applied Energy 292:116928

    2021

  9. [17]

    Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences 476(2238):20200097

    Erichson NB, Mathelin L, Yao Z, et al (2020) Shallow neural networks for fluid flow recon- struction with limited sensors. Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences 476(2238):20200097

    2020

  10. [19]

    Nature Communications 16(1):1422

    Font B, Alc´ antara-´Avila F, Rabault J, et al (2025) Deep reinforcement learning for active flow control in a turbulent separation bubble. Nature Communications 16(1):1422

    2025

  11. [20]

    Wind Energy 23(8):1739–1751

    Frederik JA, Doekemeijer BM, Mulders SP, et al (2020) The helix approach: Using dynamic individual pitch control to enhance wake mixing in wind farms. Wind Energy 23(8):1739–1751

    2020

  12. [21]

    In: International conference on machine learning, PMLR, pp 1587–1596 18

    Fujimoto S, Hoof H, Meger D (2018) Addressing function approximation error in actor-critic methods. In: International conference on machine learning, PMLR, pp 1587–1596 18

    2018

  13. [22]

    Nature Communications 14(1):6480

    Fukami K, Taira K (2023) Grasping extreme aerodynamics on a low-dimensional manifold. Nature Communications 14(1):6480

    2023

  14. [23]

    Journal of Fluid Mechanics 768:5–50

    Goit JP, Meyers J (2015) Optimal control of energy extraction in wind-farm boundary layers. Journal of Fluid Mechanics 768:5–50

    2015

  15. [24]

    In: International conference on machine learning, PMLR, pp 1861–1870

    Haarnoja T, Zhou A, Abbeel P, et al (2018) Soft actor-critic: Off-policy maximum entropy deep reinforcement learning with a stochastic actor. In: International conference on machine learning, PMLR, pp 1861–1870

    2018

  16. [25]

    Frontiers in Neurorobotics 17:1127642

    Hafez MB, Immisch T, Weber T, et al (2023) Map-based experience replay: a memory- efficient solution to catastrophic forgetting in reinforcement learning. Frontiers in Neurorobotics 17:1127642

    2023

  17. [26]

    Nature Energy 9(6):735–749

    Harrison-Atlas D, Glaws A, King RN, et al (2024) Artificial intelligence-aided wind plant opti- mization for nationwide evaluation of land use and economic benefits of wake steering. Nature Energy 9(6):735–749

    2024

  18. [27]

    International Journal of Electrical Power & Energy Systems 143:108406

    He B, Zhao H, Liang G, et al (2022) Ensemble-based Deep Reinforcement Learning for robust cooperative wind farm control. International Journal of Electrical Power & Energy Systems 143:108406

    2022

  19. [28]

    Wind Energy 25(2):195–220

    Houck DR (2022) Review of wake management techniques for wind turbines. Wind Energy 25(2):195–220

    2022

  20. [29]

    MIT Press 2:39–47

    Howard RA (1960) Dynamic programming and markov processes. MIT Press 2:39–47

    1960

  21. [30]

    Wind Energy Science 5(4):1315– 1338

    Howland MF, Ghate AS, Lele SK, et al (2020) Optimal closed-loop wake steering – Part 1: Conventionally neutral atmospheric boundary layer conditions. Wind Energy Science 5(4):1315– 1338

    2020

  22. [31]

    Nature Energy 7(9):818–827

    Howland MF, Quesada JB, Mart ´ ınez JJP, et al (2022) Collective wind farm operation based on a predictive model increases utility-scale energy production. Nature Energy 7(9):818–827

    2022

  23. [32]

    URL https://www.iea.org/reports/net-zero-by-2050

    International Energy Agency (2021) Net zero by 2050: A roadmap for the global energy sector. URL https://www.iea.org/reports/net-zero-by-2050

    2021

  24. [33]

    Wind Energy p e2946

    Jan´ e-Ippel C, Bempedelis N, Palacios R, et al (2024) Bayesian Optimisation of a Two-Turbine Configuration Around a 2D Hill Using Large Eddy Simulations. Wind Energy p e2946

    2024

  25. [34]

    Jensen N (1983) A note on wind generator interaction. No. 2411 in Risø-M, Risø National Laboratory

    1983

  26. [35]

    Wind Energy 13(6):559–572

    Jim´ enez´A, Crespo A, Migoya E (2010) Application of a LES technique to characterize the wake deflection of a wind turbine in yaw. Wind Energy 13(6):559–572

    2010

  27. [36]

    Journal of Global optimization 13:455–492

    Jones DR, Schonlau M, Welch WJ (1998) Efficient Global Optimization of Expensive Black-Box Functions. Journal of Global optimization 13:455–492

    1998

  28. [37]

    Wind Energy Science 2(1):115–131

    King RN, Dykes K, Graf P, et al (2017) Optimization of wind plant layouts using an adjoint approach. Wind Energy Science 2(1):115–131

    2017

  29. [38]

    The International Journal of Robotics Research 32(11):1238–1274 19

    Kober J, Bagnell JA, Peters J (2013) Reinforcement learning in robotics: A survey. The International Journal of Robotics Research 32(11):1238–1274 19

    2013

  30. [39]

    Journal of Physics: Conference Series 1934(1):012022

    Korb H, Asmuth H, Stender M, et al (2021) Exploring the application of reinforcement learning to wind farm control. Journal of Physics: Conference Series 1934(1):012022

    2021

  31. [40]

    Journal of Computational Physics 228:5989–6015

    Laizet S, Lamballais E (2009) High-order compact schemes for incompressible flows: A simple and efficient method with quasi-spectral accuracy. Journal of Computational Physics 228:5989–6015

    2009

  32. [41]

    International Journal for Numerical Methods in Fluids 67:1735–57

    Laizet S, Li N (2011) Incompact3d: A powerful tool to tackle turbulence problems with up to O(105) computational cores. International Journal for Numerical Methods in Fluids 67:1735–57

    2011

  33. [42]

    arXiv preprint arXiv:150902971

    Lillicrap T (2015) Continuous control with deep reinforcement learning. arXiv preprint arXiv:150902971

    2015

  34. [43]

    Journal of Fluid Mechanics 242:51–78

    Mason PJ, Thomson DJ (1992) Stochastic backscatter in large-eddy simulations of boundary layers. Journal of Fluid Mechanics 242:51–78

    1992

  35. [44]

    Wind Energy 15(2):305–317

    Meyers J, Meneveau C (2012) Optimal turbine spacing in fully developed wind farm boundary layers. Wind Energy 15(2):305–317

    2012

  36. [45]

    Wind Energy Science 7(6):2271–2306

    Meyers J, Bottasso C, Dykes K, et al (2022) Wind farm flow control: Prospects and challenges. Wind Energy Science 7(6):2271–2306

    2022

  37. [46]

    PhD thesis, Technical University of Denmark

    Mikkelsen R (2004) Actuator Disc Methods Applied to Wind Turbines. PhD thesis, Technical University of Denmark

    2004

  38. [47]

    arXiv preprint arXiv:13125602

    Mnih V (2013) Playing atari with deep reinforcement learning. arXiv preprint arXiv:13125602

    2013

  39. [48]

    nature 518(7540):529–533

    Mnih V, Kavukcuoglu K, Silver D, et al (2015) Human-level control through deep reinforcement learning. nature 518(7540):529–533

    2015

  40. [49]

    In: International conference on machine learning, PmLR, pp 1928–1937

    Mnih V, Badia AP, Mirza M, et al (2016) Asynchronous methods for deep reinforcement learning. In: International conference on machine learning, PmLR, pp 1928–1937

    2016

  41. [50]

    Flow, Turbulence and Combustion

    Mole A, Laizet S (2024) Multi-fidelity Bayesian Optimisation of Wind Farm Wake Steering using Wake Models and Large Eddy Simulations. Flow, Turbulence and Combustion

    2024

  42. [51]

    2501.13592

    Monroc CB, Buˇ si´ c A, Dubuc D, et al (2025) WFCRL: A Multi-Agent Reinforcement Learning Benchmark for Wind Farm Control. 2501.13592

    2025 arXiv

  43. [52]

    Energies 11(1):177

    Munters W, Meyers J (2018) Dynamic Strategies for Yaw and Induction Control of Wind Farms Based on Large-Eddy Simulation and Optimization. Energies 11(1):177

    2018

  44. [53]

    Physics of Fluids 28(2):025112

    Munters W, Meneveau C, Meyers J (2016) Shifted periodic boundary conditions for simulations of wall-bounded turbulent flows. Physics of Fluids 28(2):025112

    2016

  45. [54]

    Renewable Energy 181:445–456

    Padullaparthi VR, Nagarathinam S, Vasan A, et al (2022) F ALCON- F Arm Level CONtrol for wind turbines using multi-agent deep reinforcement learning. Renewable Energy 181:445–456

    2022

  46. [55]

    Journal of Computational Science 62:101707

    Partee S, Ellis M, Rigazzi A, et al (2022) Using machine learning at scale in numerical simulations with smartsim: An application to ocean climate modeling. Journal of Computational Science 62:101707

    2022

  47. [56]

    Advances in neural information processing systems 32

    Paszke A, Gross S, Massa F, et al (2019) Pytorch: An imperative style, high-performance deep learning library. Advances in neural information processing systems 32

    2019

  48. [57]

    Adaptive Computation and Machine Learning, MIT Press, Cambridge, Mass

    Rasmussen CE, Williams CKI (2008) Gaussian Processes for Machine Learning, 3rd edn. Adaptive Computation and Machine Learning, MIT Press, Cambridge, Mass. 20

    2008

  49. [58]

    Energies 14(13):3745

    Revaz T, Port´ e-Agel F (2021) Large-Eddy Simulation of Wind Turbine Flows: A New Evaluation of Actuator Disk Models. Energies 14(13):3745

    2021

  50. [59]

    arXiv preprint arXiv:170706347

    Schulman J, Wolski F, Dhariwal P, et al (2017) Proximal policy optimization algorithms. arXiv preprint arXiv:170706347

    2017

  51. [60]

    Proceedings of the IEEE 104(1):148–175

    Shahriari B, Swersky K, Wang Z, et al (2016) Taking the Human Out of the Loop: A Review of Bayesian Optimization. Proceedings of the IEEE 104(1):148–175

    2016

  52. [61]

    In: Proceedings of European Wind Energy Conference 2007

    Shen WZ, Nørk J (2007) Actuator Surface Model for Wind Turbine Flow Computations. In: Proceedings of European Wind Energy Conference 2007

    2007

  53. [62]

    Monthly Weather Review 91(3):99–164

    Smagorinsky J (1963) General ciculation experiments with the primitive equa- tions. Monthly Weather Review 91(3):99–164

    1963

  54. [63]

    In: Pereira F, Burges C, Bottou L, et al (eds) Advances in Neural Information Processing Systems, vol 25

    Snoek J, Larochelle H, Adams RP (2012) Practical bayesian optimization of machine learning algorithms. In: Pereira F, Burges C, Bottou L, et al (eds) Advances in Neural Information Processing Systems, vol 25. Curran Associates, Inc

    2012

  55. [64]

    Journal of Fluids Engineering 124(2):393–399

    Sørensen JN, Shen WZ (2002) Numerical Modeling of Wind Turbine Wakes. Journal of Fluids Engineering 124(2):393–399

    2002

  56. [65]

    arXiv preprint arXiv:09123995

    Srinivas N, Krause A, Kakade S, et al (2009) Gaussian Process Optimization in the Bandit Setting: No Regret and Experimental Design. arXiv preprint arXiv:09123995

    2009

  57. [66]

    Machine learning 3:9–44

    Sutton RS (1988) Learning to predict by the methods of temporal differences. Machine learning 3:9–44

    1988

  58. [67]

    MIT press

    Sutton RS, Barto AG (2018) Reinforcement learning: An introduction. MIT press

    2018

  59. [68]

    In: Proceedings of the AAAI conference on artificial intelligence

    Van Hasselt H, Guez A, Silver D (2016) Deep reinforcement learning with double q-learning. In: Proceedings of the AAAI conference on artificial intelligence

    2016

  60. [69]

    Wind Energy Science 8(7):1071–1131

    Veers P, Bottasso CL, Manuel L, et al (2023) Grand challenges in the design, manufacture, and operation of future wind turbine systems. Wind Energy Science 8(7):1071–1131

    2023

  61. [70]

    Physics of Fluids 35(3)

    Vignon C, Rabault J, Vinuesa R (2023) Recent advances in applying deep reinforcement learning for flow control: Perspectives and future directions. Physics of Fluids 35(3)

    2023

  62. [71]

    Energy Conversion and Management 337:119888

    Wang H, He S, Yan J, et al (2025) Deep reinforcement learning-driven wind farm flow control considering dynamic wind. Energy Conversion and Management 337:119888

    2025

  63. [72]

    Machine learning 8:279–292

    Watkins CJ, Dayan P (1992) Q-learning. Machine learning 8:279–292

    1992

  64. [73]

    Watkins CJCH (1989) Learning from delayed rewards

    1989

  65. [74]

    Renewable Energy 75:945–955

    Wu YT, Port´ e-Agel F (2015) Modeling turbine wakes and power losses within a wind farm using LES: An application to the Horns Rev offshore wind farm. Renewable Energy 75:945–955

    2015

  66. [75]

    Journal of Fluid Mechanics 981:A17

    Xia C, Zhang J, Kerrigan EC, et al (2024) Active flow control for bluff body drag reduction using reinforcement learning with partial measurements. Journal of Fluid Mechanics 981:A17

    2024

  67. [76]

    Real-time digital optimisation and decision making for energy and transport systems

    Zhang T, Wang X, Liang B, et al (2022) Catastrophic interference in reinforcement learning: A solution based on context division and knowledge distillation. IEEE Transactions on Neural Networks and Learning Systems 34(12):9925–9939 21 Acknowledgments The authors gratefully ack...

    2022

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.