Pith. sign in

REVIEW 5 major objections 5 minor 60 references

Model Predictive and Reinforcement Learning Methods for Active Flow Control of an Airfoil with Dual-point Excitation of Plasma Actuators

T0 review · 5 major / 5 minor · reviewed 2026-08-08 · deepseek-v4-flash

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

Pith's one-line read In a near-stall airfoil simulation, reinforcement learning controllers drive dual-point plasma actuators to a lift coefficient of about 1.62 within 2.5 seconds, while adaptive model predictive control stops at 1.60.

desk verdict Competent 2D CFD case study whose central RL-beats-MPC claim is not supported as stated, because MPC is asked to track an infeasible set-point while RL is asked to maximize mean lift. read the letter →

arxiv 2502.05577 v2 pith:LIGABFMU submitted 2025-02-08 physics.flu-dyn

AmirHossein Ghaemi , Abbas Ebrahimi , Majid Hajipour , Seyyed Mohammad Mahdy Shobeiry , Arash Fath Lipaei This is my paper · ORCID
classification physics.flu-dyn
keywords activeflowcontrolreinforcementlearningmodelpredictiveDBDplasmaactuatorseparationNACA4412dual-pointexcitationnear-stallaerodynamics
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 asks which closed-loop controller can better hold a NACA 4412 airfoil near stall under dual-point plasma actuation: adaptive model predictive control or reinforcement learning trained entirely online. In two-dimensional simulations at Reynolds number $4\times 10^5$ and 15 degrees angle of attack, all three RL variants—temporal-difference Q-learning, deep Q-learning, and deep Q-learning with an LSTM signal-processing layer—found an excitation frequency that raised mean lift to about $C_l = 1.619$ within roughly 2.5 seconds, choosing 100 or 200 Hz. Adaptive MPC reached its set-point of $C_l = 1.60$ at about 110 Hz but could not stabilize $C_l = 1.62$, a value near the actuator's physical limit. The paper's contribution is a direct RL-versus-MPC comparison in one online framework, showing that model-free learning adapts better than model-based tracking in a highly nonlinear, separated-flow regime.

What carries the argument

The carrying mechanism is the dual-point excitation arrangement of DBD plasma actuators combined with an online control loop. Two plasma zones are placed in the flow: one on the suction side at $x/c = 0.35$, just upstream of the baseline separation point near $0.55c$, and one at the trailing edge on the pressure side at $x/c = 0.99$; both shear layers are excited simultaneously, exploiting their interaction. The plasma region follows a linear electric-field-decay body-force model modulated at a 50% duty cycle with excitation frequencies between 0 and 400 Hz. The controllers read the mean lift coefficient over a short window and output a frequency: adaptive MPC fits a linear ARIMAX model online by recursive least squares and solves a generalized predictive control cost, while the RL agents use an $\epsilon$-greedy Q-learning update over the discrete action grid 0:50:400 Hz, with reward emphasizing improvement over the baseline $C_l = 1.44$. The mechanism that makes dual-point excitation effective is periodic forcing that locks onto the separated shear layer, forming coherent vortices and entraining high-momentum fluid, which shrinks the separation bubble and raises lift.

What would settle it

Repeat the dual-point DBD control scenario at $Re=4\times10^5$ and 15 degrees in a three-dimensional wall-resolved large-eddy simulation or a wind-tunnel experiment; the central ranking would be refuted if 100 or 200 Hz excitation does not raise mean lift to about 1.62, or if adaptive MPC can hold 1.62 stably.

Watch

Extended reading notes

Core claim

The paper's central claim is that reinforcement learning controllers identify and hold a better operating point than adaptive MPC for the same dual-point dielectric-barrier-discharge plasma actuator system on a NACA 4412 airfoil at 15 degrees and $Re = 4\times10^5$. TDRL, DQL, and DQL with signal processing each converge to a mean lift coefficient of approximately 1.619, with TDRL and the signal-processing variant settling on 200 Hz and DQL on 100 Hz; the two frequencies produce nearly equal mean lift through a 50% duty-cycle averaging effect. Adaptive MPC, a generalized predictive controller with recursive least-squares identification, reaches $C_l = 1.60$ at about 110 Hz ($F^+\approx 3$) but becomes unstable trying to hold $C_l = 1.62$, which lies at the physical limit of the dual-actuator configuration. The paper attributes the RL advantage to online, reward-driven exploration of the discrete frequency space and attributes the improved aerodynamics to shear-layer instability amplification, lock-on, and enhanced mixing that reduce the suction-side separation zone and increase the pressure differential across the airfoil.

Load-bearing premise

The load-bearing premise is that the two-dimensional RANS-SAS simulation with the simplified plasma body-force model faithfully reproduces the near-stall separated-flow dynamics the controllers exploit; the paper validates this setup only for the uncontrolled baseline, not for the controlled cases.

Editorial extensions

If this is right

  • If the comparison holds, online RL can control near-stall separation without a pre-trained dataset or an explicit plant model, reaching $C_l\approx 1.62$ within about 2.5 seconds of simulated time.
  • Adaptive MPC remains suitable for moderate set-points: it stabilizes $C_l = 1.60$ at about 110 Hz, which is close to the 100 Hz that DQL selects, so both approaches converge on $F^+\approx 3$ as an effective excitation regime.
  • The near-identical mean lift at 100 and 200 Hz implies that mean $C_l$ alone cannot distinguish these two operating points; richer feedback, such as lift fluctuation amplitude or spectral content, would be needed to break the tie.
  • More complex RL architectures make more decisions yet stabilize the lift in less simulated time, indicating that sequence-aware state representations accelerate convergence.
  • Since the dual-actuator configuration saturates near $C_l = 1.62$, further lift improvements would require changing the actuation layout or authority rather than only improving the controller.

Reading between the lines

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

  • An implication the paper leaves implicit is that the 100 Hz versus 200 Hz tie points to a plateau in the lift-versus-frequency landscape; a controller that adds a small dithering sweep could map that plateau and reduce ambiguity.
  • Because the results come from a 2D RANS-SAS model with a simplified plasma forcing model, a natural test is whether the same frequency preference survives three-dimensional wall-resolved simulations or wind-tunnel experiments, where spanwise instabilities may shift the optimum.
  • The same online state-reward loop could tune other actuation parameters, such as duty cycle, voltage amplitude, or the phase offset between the two plasma zones, since the controller only needs a scalar performance signal.
  • For MPC, the failure near $C_l=1.62$ is a set-point-tracking saturation problem; switching to RL or augmenting the identified linear model with a saturation or disturbance model could extend adaptive MPC's useful range.
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

5 major / 5 minor

Summary. The paper presents a numerical study of closed-loop active flow control for a NACA 4412 airfoil near static stall (α=15°, Re=4×10^5) using dual-point DBD plasma actuators. Four controllers are compared: adaptive MPC (a generalized predictive controller with recursive least-squares identification), temporal-difference RL, deep Q-learning, and DQL with an LSTM layer. The uncontrolled baseline is validated against external wind-tunnel data (Cp at 14° and trailing-edge wake profiles). The authors report that adaptive MPC tracks a target Cl=1.60 at ~110 Hz but cannot stabilize Cl=1.62, whereas the RL methods reach mean Cl≈1.619 in under 2.5 s at 100 or 200 Hz, and they conclude that RL outperforms MPC for this problem.

Significance. If the central comparison were established, the paper would be a useful contribution: it is among the first to compare MPC with online RL methods for active flow separation control using dual-point DBD plasma actuation, and the baseline validation against external wind-tunnel data is a strength. However, the current evidence does not support the headline claim of RL superiority. The comparison is confounded by different controller objectives, the 'optimal frequency' result is underdetermined by the near-identical means at 100 and 200 Hz, and the controlled-flow simulations rest on a single unverified 2D RANS-SAS setup. The paper is of interest to the active flow control community, but the quantitative comparisons should be regarded as preliminary rather than conclusive.

major comments (5)
  1. [§5.2–§5.5, Eq. (10), abstract/findings] The headline claim that RL methods outperform adaptive MPC is confounded by asymmetric objectives. The MPC cost in Eq. (10) minimizes tracking error to a reference w(t); in Section 5.2 the Cl=1.62 case is explicitly called 'near the physical limit', and the paper reports only that MPC 'could not stabilize' this set-point, with no time-averaged Cl reported for that run. The RL methods instead maximize a Cl-based reward, and their success metric is a mean Cl of 1.619 (Sections 5.3–5.5), which is below the MPC reference of 1.62. The observed outcome (a tracking controller with an infeasible reference fails, while a maximizing controller returns the maximum feasible mean) does not establish that MPC cannot reach mean Cl=1.619. To support the claim, the authors should either run MPC with the same objective (e.g., tracking 1.619 or maximizing mean Cl) or report the time-averaged Cl for the MPC 1.62 case.
  2. [§3.1–§3.4, §5.1] The controlled results depend entirely on a single 2D RANS-SAS setup with the Shyy plasma body-force model, but the validation is limited to the uncontrolled baseline: Cp at 14° (Fig. 7) and trailing-edge wake profiles (Fig. 8). No grid-convergence, time-step sensitivity, turbulence-model comparison, or three-dimensionality check is provided for the near-stall 15° separated flow used in all controlled cases (Sections 3.3–3.4; 5.2–5.5). The paper itself concedes in the Conclusions that 'future work could improve turbulence modeling'. Because the central quantitative claims (mean Cl=1.619 at 100/200 Hz, MPC's failure at 1.62) are grounded in this unverified simulation environment, the authors should add at least a grid-refinement and time-step sensitivity study, and preferably a turbulence-model comparison, before drawing conclusions about controller performance at these lift levels.
  3. [§4 (paragraph before §4.1) and Table 2] The action-space description is internally inconsistent. The text states that the discrete action space lies 'within the superharmonic frequencies of wake, specifically from F+ = 1 to F+ = 6' and that 'frequencies beyond this range do not elicit an effective fluid response' (citing [25]), yet Table 2 specifies Actions 0:50:400. With c=0.4 m and U∞=14.6 m/s, F+=f·c/U∞, so the action set includes 250–400 Hz, corresponding to F+ ≈ 6.85–10.96, which the text itself declares ineffective. The RL agents are therefore allowed to choose frequencies the authors state are ineffective. Either restrict the action set to F+ ≤ 6 (≈219 Hz) or justify the inclusion of 250–400 Hz; as written, the 'optimal frequency' search includes known-irrelevant actions.
  4. [§5.3–§5.5, Fig. 15] The claim that RL methods 'effectively optimized excitation frequencies' is underdetermined by the authors' own data. Section 5.5 states that the mean Cl at 100 Hz and 200 Hz is 'nearly identical', and Fig. 15 shows the two duty-cycle traces produce almost the same average. The converged frequency (TDRL and signal-processing DQL select 200 Hz; DQL selects 100 Hz) therefore reflects initial exploration and reward-landscape details rather than a distinct physical optimum. The authors should quantify the Cl difference between 100 and 200 Hz (with a tolerance or statistical test), or else weaken the 'optimal frequency' wording to 'a selected frequency among nearly equivalent options'.
  5. [§4.2–§4.4, §5.3–§5.5] Each RL method is evaluated with a single stochastic run; no seeds, multiple runs, or confidence intervals are reported, and the hyperparameters were chosen 'after multiple initial runs' (Section 4). For a comparison of learning algorithms, single-run outcomes are insufficient to distinguish method performance from run-to-run variance. The authors should report statistics over several independent runs, or at minimum demonstrate insensitivity to exploration seed and to the chosen hyperparameters, before claiming that one RL variant converges faster or more reliably than another.
minor comments (5)
  1. [Throughout] There are several typographical errors: 'desicion makings' (Sections 5.4 and 5.5), and 'Shy et al.' instead of 'Shyy et al.' (Sections 3.2 and 3.3). These should be corrected.
  2. [Throughout] The lift coefficient notation is inconsistent: 'Cl' is used in most sections but 'CL' appears in the Conclusions. Please standardize to a single notation.
  3. [§5.6, Fig. 16] The definition of 'stabilization time' used in Fig. 16 is not given. Please specify the criterion (e.g., time when the moving-average Cl remains within a tolerance band), since the comparison across methods depends on this metric.
  4. [§5.2] For the MPC Cl=1.60 case, the paper reports that the lift coefficient 'stabilizes' at 2.8 s, but it is unclear whether this refers to the instantaneous value or the moving average shown in Fig. 10. Please clarify what quantity is being used to define stabilization.
  5. [§3.2] The normalized frequency F+ is defined as F+ = f·c/U∞, but U∞ is not defined until Section 3.3 (14.6 m/s). Please define U∞ at first use and state the resulting F+ values for the action set.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the RL-versus-MPC comparison is an empirically measured CFD result, not a derivation from fitted inputs or self-citations.

full rationale

The paper's central claim is a numerical comparison of adaptive MPC and RL controllers on a common CFD baseline, and the reported outcomes (Cl values, convergence times, frequencies) are measured outputs of the simulations rather than quantities obtained by substituting the conclusion into the method definitions. The baseline simulation is validated against external experimental data from Mallor et al. and Tabatabaei et al., and the controlled cases use the same model, so the main comparison is not constructed to force the headline result. Self-citations to the authors' prior dual-point excitation work [36, 43] describe the actuation configuration, but the controller comparison does not reduce to those citations; the dual-point strategy is also independently motivated in the present paper by the baseline dual-shear-layer structure. The MPC set-points were selected using knowledge of the system's performance, and RL hyperparameters were tuned in initial runs; these are in-sample design choices rather than fitted parameters renamed as predictions, because the paper does not claim to predict a separate dataset. The most substantive concern is a fairness confound: MPC tracks a fixed set-point while RL maximizes mean Cl, so the comparison does not perfectly isolate algorithm capability. That is a correctness or experimental-design issue, not a circular reduction, and no equation or fitted value is shown to be equivalent to the conclusion. Therefore, no circular step is identified.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The central comparison is a CFD experiment built on standard turbulence and actuator models, with the controllers tuned on the same simulation environment. No new physical entities are introduced. The main hidden cost is the assumption that the 2D RANS-SAS model and the mean-lift reward capture the control-relevant physics; the paper's own flat reward between 100 and 200 Hz shows the frequency optimization is not strongly identified. Self-citation of the dual-point actuation concept (Ebrahimi and Hajipour, 2018) is prior work, not circular.

free parameters (4)
  • RL exploration and learning hyperparameters = epsilon=0.9, kappa=0.95, alpha=0.4 (TDRL) / 0.01 (DQL), gamma=0.9 (TDRL) / 0.25 (DQL), FoM=0.25
    Hand-tuned after multiple initial runs on the same simulation environment (Section 4); no sensitivity analysis is reported.
  • MPC target lift set-points = Cl=1.60 and Cl=1.62
    Chosen based on prior open-loop studies and 'the present analysis of the system's performance characteristics' (Section 5.2), so the MPC success/failure contrast is partly set by the authors.
  • DQL and LSTM network architecture = 4 hidden layers x 4 neurons; LSTM 64 units, 251 input features; SGDM lr=0.01, 100 epochs
    Selected after exploratory runs; more complex architectures reportedly performed worse (Section 4).
  • Discrete excitation frequency action set = 0:50:400 Hz (9 actions)
    Discretization at 50 Hz steps, chosen to represent F+=1 to 6 but actually includes F+ up to about 11 (Section 4, Table 2).
assumptions (5)
  • domain assumption RANS-SAS with the stated 2D grid adequately captures the separated shear layers and the controlled-flow lift at Re=4e5 near stall.
    Invoked in Sections 3.1 and 3.4 for all controlled-flow results; no validation of the controlled case or sensitivity study is provided.
  • domain assumption The Shyy phenomenological body-force model represents DBD plasma actuation well enough for quantitative lift comparison.
    Used in Section 3.2; the model is a simplified linear-field approximation from Shyy et al. [52], not validated here for the dual-point configuration.
  • domain assumption The mean lift coefficient over the FoM window is a sufficient statistic for state and reward in this control problem.
    Used in Sections 4.2-4.4 and Table 2; the paper later shows 100 Hz and 200 Hz produce nearly equal mean Cl, so the state/reward does not distinguish control-relevant frequencies.
  • domain assumption The baseline validation at 14 degrees and wake profiles transfers to the 15-degree near-stall configuration.
    Validation in Section 5.1 is at alpha=14 degrees and for wake velocity; the controlled study runs at alpha=15 degrees.
  • ad hoc to paper Frequencies with F+ above 6 are ineffective (from [25]) while the action space includes 250-400 Hz (F+ up to about 11).
    Section 4 says effective frequencies lie in F+=1 to 6, but Table 2 and the action space '0:50:400' include values up to 400 Hz; the inconsistency is not explained.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Model Predictive and Reinforcement Learning Methods for Active Flow Control of an Airfoil with Dual-point Excitation of Plasma Actuators." pith.science (2026). https://pith.science/paper/LIGABFMU

@misc{pith2026250205577,
  author       = {Pith},
  title        = {Pith review of: Model Predictive and Reinforcement Learning Methods for Active Flow Control of an Airfoil with Dual-point Excitation of Plasma Actuators},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/LIGABFMU}},
  note         = {Machine review of arXiv:2502.05577}
}
read the original abstract

This study investigates the effectiveness of Model Predictive Control (MPC) and Reinforcement Learning (RL) for active flow control over a NACA 4412 airfoil near static stall at Reynolds number 4*10^5. By systematically evaluating these strategies, the research addresses a critical gap in optimizing excitation frequency and improving response time in flow control. The work contributes to understanding RL adaptability and performance versus MPC in aerodynamic flow separation control. Numerical simulations of the Reynolds Averaged Navier-Stokes equations with the Scale-Adaptive Simulation turbulence model are used. Dielectric Barrier Discharge plasma actuators in dual-point excitation mode control flow separation. The study evaluates adaptive MPC, temporal difference RL (TDRL), and deep Q-learning (DQL) for optimizing excitation frequency and expediting stabilization. An integrated signal processing DQL approach is also examined. Adaptive MPC achieved Cl = 1.60 at 110 Hz but struggled near physical limits. RL optimized excitation frequencies, reaching Cl = 1.62 in under 2.5 s at 100 or 200 Hz. The study presents a novel RL - MPC comparison for active flow control with DBD actuators, contrasting with prior work focusing on MPC or RL alone. Using an online learning framework, RL methods dynamically adapt to real-time conditions. Evaluating adaptive MPC and RL together in this setup yields new insights into comparative performance in dynamic environments.

Figures

Figures reproduced from arXiv: 2502.05577 by the authors.

Figure 1
Figure 1. (a) Schematic of a DBD plasma actuator and (b) Schematic of dual-point [PITH_FULL_IMAGE:figures/full_fig_p007_1.png] view at source ↗
Figure 2
Figure 2. Triangular zone associated with plasma formation [36]. [PITH_FULL_IMAGE:figures/full_fig_p011_2.png] view at source ↗
Figure 3
Figure 3. Side view of the computational domain. Source: Authors’ own work [PITH_FULL_IMAGE:figures/full_fig_p012_3.png] view at source ↗
Figures from the paper (16 more)
Figure 4
Figure 4. Figure 4: Detailed view of computational grid around airfoil and the plasma zones. [PITH_FULL_IMAGE:figures/full_fig_p012_4.png]
Figure 5
Figure 5. Figure 5: DQL Network Architecture. W and B indicate weight and Bias of layers, [PITH_FULL_IMAGE:figures/full_fig_p021_5.png]
Figure 6
Figure 6. Figure 6: Signal processing integrated with DQL Network Architecture. [PITH_FULL_IMAGE:figures/full_fig_p022_6.png]
Figure 7
Figure 7. Figure 7: Comparison of Cp distribution along the x/c between experimental and CFD data. Source: Authors’ own work [PITH_FULL_IMAGE:figures/full_fig_p023_7.png]
Figure 8
Figure 8. Figure 8: Comparison of wake velocity profiles at the trailing edge. [PITH_FULL_IMAGE:figures/full_fig_p024_8.png]
Figure 9
Figure 9. Figure 9: Contours of baseline flow: (a) Velocity magnitude [m/s], (b) Turbulent kinetic [PITH_FULL_IMAGE:figures/full_fig_p025_9.png]
Figure 10
Figure 10. Figure 10: Desired Cl of 1.6 using adaptive MPC and its corresponding excitation fre￾quencies. Source: Authors’ own work. To push the boundaries of the control system, aiming to achieve higher lift coefficients, a second target Cl of 1.62 was selected. However, reaching such a h…
Figure 11
Figure 11. Figure 11: Desired Cl of 1.62 using Adaptive MPC and its corresponding excitation fre￾quencies. Source: Authors’ own work. 5.3. TDRL In contrast to the previous adaptive MPC approach, TDRL seeks to iden￾tify the optimal excitation frequency of plasma actuators that maximizes the…
Figure 12
Figure 12. Figure 12: Exploring the optimal excitation frequency that maximizes [PITH_FULL_IMAGE:figures/full_fig_p028_12.png]
Figure 13
Figure 13. Figure 13: Selection of the optimal excitation frequency for maximizing [PITH_FULL_IMAGE:figures/full_fig_p029_13.png]
Figure 14
Figure 14. Figure 14: Selection of the optimal excitation frequency for maximizing [PITH_FULL_IMAGE:figures/full_fig_p029_14.png]
Figure 15
Figure 15. Figure 15: Comparison of results from different reinforcement learning methods for deter [PITH_FULL_IMAGE:figures/full_fig_p030_15.png]
Figure 16
Figure 16. Figure 16: Stabilization time and converged frequencies of different control methods. [PITH_FULL_IMAGE:figures/full_fig_p031_16.png]
Figure 17
Figure 17. Figure 17: Velocity profiles within the plasma actuator zones on the suction and pressure sides, shown with and without plasma actuation. Red and blue shading represent plasma￾on and plasma-off conditions, respectively; darker shades indicate downstream streamwise locations. Sou…
Figure 18
Figure 18. Figure 18: Contours of controlled flow using signal processing integrated with DQL [PITH_FULL_IMAGE:figures/full_fig_p033_18.png]
Figure 19
Figure 19. Figure 19: Comparison of Cp distribution along x/c between baseline flow and signal processing integrated with DQL. Source: Authors’ own work. environment. The main findings are summarized as follows: • Adaptive MPC Performance: – The adaptive MPC approach effectively achieved t…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

60 extracted references · 58 canonical work pages

  1. [25]

    Shimomura, S

    S. Shimomura, S. Sekimoto, A. Oyama, K. Fujii, H. Nishida, Closed- loop flow separation control using the deep q network over airfoil, AIAA Journal 58 (10) (2020) 4260–4270

    work page 2020

  2. [1]

    Z. Liu, L. Zhou, H. Tang, Z. Wang, F. Zhao, X. Ji, H. Zhang, Primary instability, sensitivity and active control of flow past two tandem circular cylinders, Ocean Engineering 294 (2024) 116863

    work page 2024

  3. [2]

    H. Ding, Z. Cheng, M. Liu, L. Xiao, S. Zhu, Effects of synthetic jet control parameters on characteristics of flow around a square cylinder at subcritical reynolds number, Ocean Engineering 309 (2024) 118577

    work page 2024

  4. [3]

    Seifert, A

    A. Seifert, A. Darabi, I. Wyganski, Delay of airfoil stall by periodic excitation, Journal of aircraft 33 (4) (1996) 691–698

    work page 1996

  5. [4]

    Eulalie, E

    Y. Eulalie, E. Fournier, P. Gilotte, D. Holst, S. Johnson, C. N. Nayeri, T. Sch¨ utz, D. Wieser, Active flow control analysis at the rear of an suv, International Journal of Numerical Methods for Heat & Fluid Flow 28 (5) (2018) 1169–1186

    work page 2018

  6. [5]

    Rezaeiha, H

    A. Rezaeiha, H. Montazeri, B. Blocken, Active flow control for power enhancement of vertical axis wind turbines: Leading-edge slot suction, Energy 189 (2019) 116131

    work page 2019

  7. [6]

    C. J. Bay, J. Annoni, L. A. Mart ´ ınez-Tossas, L. Y. Pao, K. E. Johnson, Flow control leveraging downwind rotors for improved wind power plant operation, in: 2019 American control conference (ACC), Ieee, 2019, pp. 2843–2848

    work page 2019

  8. [8]

    Macquart, A

    T. Macquart, A. Maheri, K. Busawon, A decoupling control strategy for wind turbine blades equipped with active flow controllers, Wind energy 20 (4) (2017) 569–584

    work page 2017

Show all 60 references
  1. [9]

    S. Das, U. Srinivasan, J. Arakeri, Unsteady separation and vortex shed- ding from a laminar separation bubble over a bluff body, Journal of Fluids and Structures 40 (2013) 233–245

    2013

  2. [11]

    R. L. Simpson, Turbulent boundary-layer separation, Annual Review of Fluid Mechanics 21 (1) (1989) 205–232

    1989

  3. [12]

    F. Eich, C. J. K¨ ahler, Large-scale coherent motions in turbulent bound- ary layers under an adverse pressure gradient up to flow separation, International Journal of Heat and Fluid Flow 85 (2020) 108645

    2020

  4. [13]

    R. B. Kotapati, R. Mittal, O. Marxen, F. Ham, D. You, L. N. CATTAFESTA III, Nonlinear dynamics and synthetic-jet-based control of a canonical separated flow, Journal of Fluid Mechanics 654 (2010) 65–97

    2010

  5. [14]

    E. A. Deem, L. N. Cattafesta III, M. S. Hemati, H. Zhang, C. Rowley, R. Mittal, Adaptive separation control of a laminar boundary layer using online dynamic mode decomposition, Journal of Fluid Mechanics 903 (2020) A21

    2020

  6. [15]

    Ransing, Guest editorial: Data-driven methods for heat transfer and fluid flow, International Journal of Numerical Methods for Heat & Fluid Flow 34 (8) (2024) 2833–2835

    R. Ransing, Guest editorial: Data-driven methods for heat transfer and fluid flow, International Journal of Numerical Methods for Heat & Fluid Flow 34 (8) (2024) 2833–2835

    2024

  7. [16]

    Ghalambaz, M

    M. Ghalambaz, M. A. Sheremet, M. A. Khan, Z. Raizah, J. Shafi, Physics-informed neural networks (p inns): application categories, trends and impact, International Journal of Numerical Methods for Heat & Fluid Flow 34 (8) (2024) 3131–3165

    2024

  8. [18]

    J. L. Proctor, S. L. Brunton, J. N. Kutz, Dynamic mode decomposition with control, SIAM Journal on Applied Dynamical Systems 15 (1) (2016) 142–161

    2016

  9. [19]

    Dovetta, P

    N. Dovetta, P. J. Schmid, D. Sipp, Uncertainty propagation in model extraction by system identification and its implication for control design, Journal of Fluid Mechanics 791 (2016) 214–236

    2016

  10. [20]

    Huang, J

    S.-C. Huang, J. Kim, Control and system identification of a separated flow, Physics of Fluids 20 (10) (2008)

    2008

  11. [21]

    Obeid, G

    S. Obeid, G. Ahmadi, R. Jha, Narmax identification based closed-loop control of flow separation over naca 0015 airfoil, Fluids 5 (3) (2020) 100

    2020

  12. [22]

    Ishize, H

    T. Ishize, H. Omichi, K. Fukagata, Flow control by a hybrid use of machine learning and control theory, International Journal of Numerical Methods for Heat & Fluid Flow 34 (8) (2024) 3253–3277

    2024

  13. [23]

    Hachem, A

    E. Hachem, A. Vishwasrao, M. Renault, J. Viquerat, P. M´ eliga, Re- inforcement learning for cooling rate control during quenching, Inter- national Journal of Numerical Methods for Heat & Fluid Flow 34 (8) (2024) 3223–3252

    2024

  14. [24]

    Vignon, J

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

    2023

  15. [26]

    Paris, S

    R. Paris, S. Beneddine, J. Dandois, Reinforcement-learning-based actu- ator selection method for active flow control, Journal of Fluid Mechanics 955 (2023) A8

    2023

  16. [28]

    Varela, P

    P. Varela, P. Su´ arez, F. Alc´ antara-´Avila, A. Mir´ o, J. Rabault, B. Font, L. M. Garc ´ ıa-Cuevas, O. Lehmkuhl, R. Vinuesa, Deep reinforcement learning for flow control exploits different physics for increasing reynolds number regimes, in: Actuators, Vol. 11, MDPI, 2022, p. 359

    2022

  17. [29]

    F. Ren, J. Rabault, H. Tang, Applying deep reinforcement learning to active flow control in weakly turbulent conditions, Physics of Fluids 33 (3) (2021)

    2021

  18. [30]

    W. Jia, H. Xu, Robust and adaptive deep reinforcement learning for enhancing flow control around a square cylinder with varying reynolds numbers, Physics of Fluids 36 (5) (2024)

    2024

  19. [31]

    B. Font, F. Alc´ antara-´Avila, J. Rabault, R. Vinuesa, O. Lehmkuhl, Ac- tive flow control of a turbulent separation bubble through deep reinforce- ment learning, in: Journal of Physics: Conference Series, Vol. 2753, IOP Publishing, 2024, p. 012022

    2024

  20. [32]

    B. Font, F. Alc´ antara-´Avila, J. Rabault, R. Vinuesa, O. Lehmkuhl, Deep reinforcement learning for active flow control in a turbulent separation bubble (2024)

    2024

  21. [33]

    Javadi, M

    K. Javadi, M. Hajipour, Separation control using quasi-radial wall jets, Aerospace Science and Technology 68 (2017) 240–251

    2017

  22. [34]

    Javadi, M

    K. Javadi, M. Hajipour, Quasi-radial wall jets as a new concept in boundary layer flow control, Journal of Turbulence 19 (1) (2018) 25– 48

    2018

  23. [35]

    Samimy, N

    M. Samimy, N. Webb, M. Crawley, Excitation of free shear-layer instabil- ities for high-speed flow control, AIAA journal 56 (5) (2018) 1770–1791

    2018

  24. [36]

    Ebrahimi, M

    A. Ebrahimi, M. Hajipour, Flow separation control over an airfoil using dual excitation of dbd plasma actuators, Aerospace Science and Tech- nology 79 (2018) 658–668

    2018

  25. [38]

    Oveisi, M

    S. Oveisi, M. Mani, B. Mojarrad, M. Kazemi, Experimental investiga- tion into the flow structure of plasma induced jet in a 2-d cross-flow, European Journal of Mechanics-B/Fluids 98 (2023) 102–119

    2023

  26. [39]

    Louste, G

    C. Louste, G. Artana, E. Moreau, G. Touchard, Sliding discharge in air at atmospheric pressure: electrical properties, Journal of Electrostatics 63 (6-10) (2005) 615–620

    2005

  27. [40]

    Moreau, Airflow control by non-thermal plasma actuators, Journal of physics D: applied physics 40 (3) (2007) 605

    E. Moreau, Airflow control by non-thermal plasma actuators, Journal of physics D: applied physics 40 (3) (2007) 605

    2007

  28. [41]

    T. C. Corke, M. L. Post, D. M. Orlov, Single dielectric barrier discharge plasma enhanced aerodynamics: physics, modeling and applications, Experiments in Fluids 46 (2009) 1–26

    2009

  29. [42]

    Hajipour, A

    M. Hajipour, A. Ebrahimi, X. Amandolese, Active flow control of a wing section in stall flutter by dielectric barrier discharge plasma actuators, Physics of Fluids 34 (7) (2022)

    2022

  30. [43]

    Ebrahimi, M

    A. Ebrahimi, M. Hajipour, K. Ghamkhar, Dual-position excitation tech- nique in flow control over an airfoil at low speeds, International Journal of Numerical Methods for Heat & Fluid Flow 30 (9) (2020) 4141–4154

    2020

  31. [44]

    B. Li, X. Meng, S. Yin, W. Hui, H. Li, Flow separation control over an airfoil using plasma co-flow jet, AIAA Journal 60 (4) (2022) 2195–2206

    2022

  32. [45]

    Samimy, N

    M. Samimy, N. Webb, A. Esfahani, Reinventing the wheel: excitation of flow instabilities for active flow control using plasma actuators, Journal of Physics D: Applied Physics 52 (35) (2019) 354002

    2019

  33. [46]

    Gross, H

    A. Gross, H. Fasel, Active flow control for naca 6-series airfoil at re= 64,200, AIAA journal 48 (9) (2010) 1889–1902

    2010

  34. [47]

    Darabi, I

    A. Darabi, I. Wygnanski, Active management of naturally separated flow over a solid surface. part 1. the forced reattachment process, Journal of Fluid Mechanics 510 (2004) 105–129

    2004

  35. [49]

    Ashcraft, K

    T. Ashcraft, K. Decker, J. C. Little, Control of boundary layer separa- tion and the wake of an airfoil using ns-dbd plasma actuators, in: 54th AIAA Aerospace Sciences Meeting, 2016, p. 0839

    2016

  36. [50]

    M. Sato, H. Aono, A. Yakeno, T. Nonomura, K. Fujii, K. Okada, K. Asada, Multifactorial effects of operating conditions of dielectric- barrier-discharge plasma actuator on laminar-separated-flow control, AIAA journal 53 (9) (2015) 2544–2559

    2015

  37. [51]

    Menter, Y

    F. Menter, Y. Egorov, A scale adaptive simulation model using two- equation models, in: 43rd AIAA aerospace sciences meeting and exhibit, 2005, p. 1095

    2005

  38. [52]

    W. Shyy, B. Jayaraman, A. Andersson, Modeling of glow discharge- induced fluid dynamics, Journal of applied physics 92 (11) (2002) 6434– 6443

    2002

  39. [53]

    Maden, R

    I. Maden, R. Maduta, J. Kriegseis, S. Jakirli´ c, C. Schwarz, S. Grund- mann, C. Tropea, Experimental and computational study of the flow induced by a plasma actuator, International Journal of Numerical Meth- ods for Heat & Fluid Flow 41 (2013) 80–89

    2013

  40. [54]

    Z. Li, B. Hu, S. Lan, J. Zhang, J. Huang, Control of turbulent channel flow using a plasma-based body force, Computers & fluids 119 (2015) 26–36

    2015

  41. [55]

    C. M. Rhie, W.-L. Chow, Numerical study of the turbulent flow past an airfoil with trailing edge separation, AIAA journal 21 (11) (1983) 1525–1532

    1983

  42. [56]

    S. J. Qin, T. A. Badgwell, A survey of industrial model predictive control technology, Control engineering practice 11 (7) (2003) 733–764

    2003

  43. [57]

    Morari, J

    M. Morari, J. H. Lee, Model predictive control: past, present and future, Computers & chemical engineering 23 (4-5) (1999) 667–682

    1999

  44. [58]

    J. M. Maciejowski, M. Huzmezan, Predictive control, in: Robust Flight Control: A Design Challenge, Springer, 2007, pp. 125–134

    2007

  45. [59]

    Fernandez-Camacho, C

    E. Fernandez-Camacho, C. Bordons-Alba, Model predictive control in the process industry, Springer, 1995. 42 Author Accepted Manuscript (AAM) DOI: 10.1108/HFF-02-2025-0118

    1995 doi

  46. [60]

    E. F. Camacho, C. Bordons, E. F. Camacho, C. Bordons, Constrained model predictive control, Springer, 2007

    2007

  47. [61]

    D. W. Clarke, C. Mohtadi, P. S. Tuffs, Generalized predictive con- trol—part i. the basic algorithm, Automatica 23 (2) (1987) 137–148

    1987

  48. [62]

    R. E. Kalman, A new approach to linear filtering and prediction prob- lems (1960)

    1960

  49. [63]

    V. Mnih, K. Kavukcuoglu, D. Silver, A. A. Rusu, J. Veness, M. G. Bellemare, A. Graves, M. Riedmiller, A. K. Fidjeland, G. Ostrovski, et al., Human-level control through deep reinforcement learning, nature 518 (7540) (2015) 529–533

    2015

  50. [64]

    Schmidhuber, S

    J. Schmidhuber, S. Hochreiter, et al., Long short-term memory, Neural Comput 9 (8) (1997) 1735–1780

    1997

  51. [65]

    Mallor, C

    F. Mallor, C. S. Vila, M. Hajipour, R. Vinuesa, P. Schlatter, R. ¨Orl¨ u, Experimental characterization of turbulent boundary layers around a naca 4412 wing profile, Experimental Thermal and Fluid Science (2024) 111327

    2024

  52. [66]

    Tabatabaei, M

    N. Tabatabaei, M. Hajipour, F. Mallor, R. ¨Orl¨ u, R. Vinuesa, P. Schlat- ter, Rans modelling of a naca4412 wake using wind tunnel measure- ments, Fluids 7 (5) (2022) 153. 43

    2022

Pith tools

Reviewed August 8, 2026 · model on record in the stance chip above.