pith. sign in

arxiv: 2606.30484 · v1 · pith:X6SCAQVJnew · submitted 2026-06-29 · ⚛️ physics.flu-dyn

Offline accuracy is not enough: closed-loop instability and stabilisation of a wall-sensor neural estimator in opposition control

Giorgio Maria Cavallazzi , Miguel P\'erez-Cuadrado , Alfredo Pinelli This is my paper

Pith reviewed 2026-06-30 03:26 UTC · model grok-4.3

classification ⚛️ physics.flu-dyn
keywords opposition controlneural estimatorclosed-loop instabilitydrag reductionwall sensorsturbulence controldistribution shiftspectral consistency
0
0 comments X

The pith

Retraining a neural estimator on its own closed-loop data restores stable wall-only opposition control with much of the ideal drag reduction.

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

A recurrent neural network is trained to infer the detection-plane velocity needed for opposition control from the two wall shear-stress components. Offline tests on fixed flow states show excellent accuracy, with 0.99 correlation and high coherence on energetic scales. In closed loop the same estimator rapidly decorrelates because the control action drives the flow away from the training distribution and unresolved high-wavenumber errors propagate through the wall boundary condition. Filtering and averaging only postpone collapse. Imposing spectral consistency on the actuation and retraining the estimator on data generated by its own closed-loop operation recovers stable control that retains a large fraction of the drag reduction obtained when the true detection-plane velocity is available.

Core claim

The paper establishes that offline reconstruction accuracy is insufficient for live surrogate sensing in opposition control. The estimator fails in feedback because the controller itself shifts the flow state off the training attractor, allowing high-wavenumber errors to enter through the wall and produce out-of-distribution inputs. Enforcing spectral consistency on the deployed actuation together with retraining on the estimator's own closed-loop trajectories restores dynamic consistency, yielding wall-only control that preserves much of the drag reduction of ideal opposition control.

What carries the argument

Recurrent neural estimator mapping wall shear stresses to detection-plane velocity, stabilized by spectral consistency on actuation and retraining on closed-loop data

If this is right

  • Offline accuracy of a surrogate sensor does not guarantee closed-loop stability.
  • The actuation itself can move the flow into regions where the estimator receives out-of-distribution inputs.
  • Spectral consistency on the deployed wall-normal velocity prevents immediate amplification of unresolved scales.
  • Retraining on trajectories generated by the estimator itself allows it to track the new controlled attractor.
  • Wall measurements alone can support opposition control that retains a large share of the drag reduction obtained with direct detection-plane sensing.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same distribution-shift problem is likely to appear in other machine-learning flow-control applications that move from open-loop training data to closed-loop deployment.
  • Iterative collection of closed-loop data followed by retraining may be required for reliable neural sensors in any turbulent control setting.
  • Training data for control-oriented estimators should include trajectories from the controlled system rather than only the uncontrolled or ideal case.

Load-bearing premise

Distribution shift caused by the controller moving the flow off the training attractor, together with unresolved high-wavenumber errors entering through the wall, is the dominant mechanism of closed-loop instability.

What would settle it

Deploy the spectrally consistent estimator without any closed-loop retraining and measure whether long-time stability and drag reduction are nevertheless recovered.

read the original abstract

Opposition control reduces skin-friction drag by opposing the wall-normal velocity on a near-wall detection plane, but the detection-plane velocity it requires is not available from wall-mounted sensors. Wall data can reconstruct inner-flow quantities accurately when assessed offline on a fixed flow state, and we ask whether such a reconstructed field can instead serve as a live surrogate sensor inside the feedback loop. We train a recurrent estimator to infer the detection-plane velocity from the two wall-shear-stress components in opposition-controlled turbulence. Offline it performs extremely well, reaching a correlation of 0.99 and near-unity coherence across the energetic scales; yet the same estimator fails in closed loop, decorrelating from the true field within a few viscous time units as the control collapses. The failure is not one of accuracy but of distribution shift induced by the controller itself: small closed-loop errors carry the flow off the attractor represented in the training data, while unresolved high-wavenumber errors enter through the wall boundary condition and return as out-of-distribution inputs. Standard remedies such as low-pass filtering and exponential averaging only delay numerical breakdown while accelerating decorrelation. Stable wall-only control is recovered by imposing spectral consistency on the deployed actuation and retraining the estimator on its own closed-loop data, giving a controller that holds much of the drag reduction of ideal opposition control from wall quantities alone. The obstacle is not whether the near-wall flow can be reconstructed offline, but whether that reconstruction stays dynamically consistent when allowed to modify the flow it senses.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 0 minor

Summary. The manuscript claims that a recurrent neural estimator trained offline to reconstruct detection-plane wall-normal velocity from wall shear stresses achieves high accuracy (correlation 0.99, near-unity coherence) but rapidly decorrelates and destabilizes closed-loop opposition control; the failure arises from distribution shift, and stability with substantial drag reduction is recovered by enforcing spectral consistency on the actuation signal together with retraining the estimator on its own closed-loop trajectories.

Significance. If the empirical demonstration holds, the result is significant for data-driven flow control: it isolates a concrete mechanism (distribution shift from controller-induced attractor departure plus unresolved boundary-condition errors) that prevents offline-trained estimators from functioning in feedback, and supplies a practical, reproducible fix that recovers most of the drag reduction of ideal opposition control from wall quantities alone. The direct numerical simulation basis and the explicit before/after performance contrast constitute a clear, falsifiable contribution.

major comments (1)
  1. [Abstract] Abstract: the central claim that the proposed remedies restore 'much of the drag reduction of ideal opposition control' is only moderately supported without reported error budgets, run-to-run variability, or statistical tests on the achieved drag-reduction percentages; these details are needed to establish that the recovered performance is robust rather than a single-run outcome.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the positive assessment of the work and the recommendation for minor revision. We address the single major comment below.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the central claim that the proposed remedies restore 'much of the drag reduction of ideal opposition control' is only moderately supported without reported error budgets, run-to-run variability, or statistical tests on the achieved drag-reduction percentages; these details are needed to establish that the recovered performance is robust rather than a single-run outcome.

    Authors: We agree that the abstract claim requires stronger statistical grounding to demonstrate robustness. In the revised manuscript we will augment the abstract (and the corresponding results section) with quantitative error budgets, standard deviations across multiple independent closed-loop runs, and a brief statement of the number of realizations used, thereby making the performance comparison with ideal opposition control statistically explicit rather than qualitative. revision: yes

Circularity Check

0 steps flagged

No significant circularity

full rationale

The paper reports an empirical demonstration: a recurrent neural estimator is trained on DNS data, evaluated offline (high correlation reported), observed to fail in closed loop due to distribution shift, and then stabilized via spectral consistency on actuation plus retraining on the resulting closed-loop trajectories. No central quantity is defined in terms of another (no self-definitional mapping), no fitted parameter is relabeled as an independent prediction, and no load-bearing premise rests on a self-citation chain or imported uniqueness theorem. The reported outcome (drag reduction recovered) is obtained directly from the simulations and retraining procedure rather than by algebraic reduction to the training inputs; the derivation chain is therefore self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

The central claim rests on the assumption that the direct numerical simulations faithfully represent the controlled turbulence and that the recurrent network can be retrained to track the modified attractor; no new physical entities are introduced.

free parameters (1)
  • recurrent network architecture and training hyperparameters
    Chosen to achieve the reported offline correlation; their specific values are not stated in the abstract.
axioms (1)
  • domain assumption Direct numerical simulation of channel flow with the chosen resolution and boundary conditions accurately captures the near-wall dynamics relevant to opposition control.
    All reported results depend on the fidelity of the underlying flow solver.

pith-pipeline@v0.9.1-grok · 5816 in / 1386 out tokens · 80263 ms · 2026-06-30T03:26:26.098976+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

51 extracted references · 8 canonical work pages

  1. [1]

    Cambridge University Press, Cambridge (2000)

    Gad-el-Hak, M.: Flow Control: Passive, Active, and Reactive Flow Management. Cambridge University Press, Cambridge (2000)

    2000

  2. [2]

    Philosophical Transactions of the Royal Society A369(1940), 1556–1569 (2011)

    Spalart, P.R., McLean, J.D.: Drag reduction: enticing turbulence, and then an industry. Philosophical Transactions of the Royal Society A369(1940), 1556–1569 (2011)

    1940

  3. [3]

    Journal of Fluid Mechanics627, 161–178 (2009)

    Quadrio, M., Ricco, P., Viotti, C.: Streamwise-travelling waves of spanwise wall velocity for turbulent drag reduction. Journal of Fluid Mechanics627, 161–178 (2009)

    2009

  4. [4]

    Journal of Fluid Mechanics 693, 150–200 (2012)

    Touber, E., Leschziner, M.A.: Near-wall streak modification by spanwise oscil- latory wall motion and drag-reduction mechanisms. Journal of Fluid Mechanics 693, 150–200 (2012)

    2012

  5. [5]

    Journal of Fluid Mechanics802, 553–582 (2016)

    Gatti, D., Quadrio, M.: Reynolds-number dependence of turbulent skin-friction drag reduction induced by spanwise forcing. Journal of Fluid Mechanics802, 553–582 (2016)

    2016

  6. [6]

    Journal of Fluid Mechanics681, 154–172 (2011)

    Kametani, Y., Fukagata, K.: Direct numerical simulation of spatially developing turbulent boundary layers with uniform blowing or suction. Journal of Fluid Mechanics681, 154–172 (2011)

    2011

  7. [7]

    International Journal of Heat and Fluid Flow55, 132–142 (2015)

    Kametani, Y., Fukagata, K., ¨Orl¨ u, R., Schlatter, P.: Effect of uniform blow- ing/suction in a turbulent boundary layer at moderate Reynolds number. International Journal of Heat and Fluid Flow55, 132–142 (2015)

    2015

  8. [8]

    Journal of Fluid Mechanics683, 57–93 (2011) 29

    Hasegawa, Y., Kasagi, N.: Dissimilar control of momentum and heat transfer in a fully developed turbulent channel flow. Journal of Fluid Mechanics683, 57–93 (2011) 29

    2011

  9. [9]

    Applied Mechanics Reviews67(5), 050801 (2015)

    Brunton, S.L., Noack, B.R.: Closed-loop turbulence control: progress and chal- lenges. Applied Mechanics Reviews67(5), 050801 (2015)

    2015

  10. [10]

    Physics of Fluids23(2), 025102 (2011)

    Chung, Y.M., Talha, T.: Effectiveness of active flow control for turbulent skin friction drag reduction. Physics of Fluids23(2), 025102 (2011)

    2011

  11. [11]

    Nature Communications12, 5805 (2021)

    Marusic, I., Chandran, D., Rouhi, A., Fu, M.K., Wine, D., Holloway, B., Chung, D., Smits, A.J.: An energy-efficient pathway to turbulent drag reduction. Nature Communications12, 5805 (2021)

    2021

  12. [12]

    Physics of Fluids14(11), 73–76 (2002)

    Fukagata, K., Iwamoto, K., Kasagi, N.: Contribution of Reynolds stress distribu- tion to the skin friction in wall-bounded flows. Physics of Fluids14(11), 73–76 (2002)

    2002

  13. [13]

    Journal of Fluid Mechanics262, 75–110 (1994)

    Choi, H., Moin, P., Kim, J.: Active turbulence control for drag reduction in wall- bounded flows. Journal of Fluid Mechanics262, 75–110 (1994)

    1994

  14. [14]

    Physics of Fluids10(9), 2421–2423 (1998)

    Hammond, E.P., Bewley, T.R., Moin, P.: Observed mechanisms for turbu- lence attenuation and enhancement in opposition-controlled wall-bounded flows. Physics of Fluids10(9), 2421–2423 (1998)

    1998

  15. [15]

    Physics of Fluids14(11), 4069–4080 (2002)

    Chang, Y., Collis, S.S., Ramakrishnan, S.: Viscous effects in control of near-wall turbulence. Physics of Fluids14(11), 4069–4080 (2002)

    2002

  16. [16]

    Physics of Fluids 27(7), 075101 (2015)

    Stroh, A., Frohnapfel, B., Schlatter, P., Hasegawa, Y.: A comparison of opposition control in turbulent boundary layer and turbulent channel flow. Physics of Fluids 27(7), 075101 (2015)

    2015

  17. [17]

    Physics of Fluids19(10), 108102 (2007)

    Pami` es, M., Garnier, E., Merlen, A., Sagaut, P.: Response of a spatially devel- oping turbulent boundary layer to active control strategies in the framework of opposition control. Physics of Fluids19(10), 108102 (2007)

    2007

  18. [18]

    AIAA Journal27(3), 308–314 (1989)

    Gad-el-Hak, M., Blackwelder, R.F.: Selective suction for controlling bursting events in a boundary layer. AIAA Journal27(3), 308–314 (1989)

    1989

  19. [19]

    Journal of Fluid Mechanics358, 245–258 (1998)

    Lee, C., Kim, J., Choi, H.: Suboptimal control of turbulent channel flow for drag reduction. Journal of Fluid Mechanics358, 245–258 (1998)

    1998

  20. [20]

    International Journal of Heat and Fluid Flow 25(3), 341–350 (2004)

    Fukagata, K., Kasagi, N.: Suboptimal control for drag reduction via suppression of near-wall Reynolds shear stress. International Journal of Heat and Fluid Flow 25(3), 341–350 (2004)

    2004

  21. [21]

    Journal of Fluid Mechanics447, 179–225 (2001)

    Bewley, T.R., Moin, P., Temam, R.: DNS-based predictive control of turbulence: an optimal benchmark for feedback algorithms. Journal of Fluid Mechanics447, 179–225 (2001)

    2001

  22. [22]

    Physica D238(13), 1082–1086 (2009) 30

    Fukagata, K., Sugiyama, K., Kasagi, N.: On the lower bound of net driving power in controlled duct flows. Physica D238(13), 1082–1086 (2009) 30

    2009

  23. [23]

    Physics of Fluids18(3) (2006)

    Rebbeck, H., Choi, K.-S.: A wind-tunnel experiment on real-time opposition control of turbulence. Physics of Fluids18(3) (2006)

    2006

  24. [24]

    Physical Review Fluids 9(6), 064602 (2024)

    Dacome, G., M¨ orsch, R., Kotsonis, M., Baars, W.J.: Opposition flow control for reducing skin-friction drag of a turbulent boundary layer. Physical Review Fluids 9(6), 064602 (2024)

    2024

  25. [25]

    Progress in Aerospace Sciences38(6-7), 515–570 (2002) https://doi.org/10.1016/ S0376-0421(02)00031-3

    Naughton, J.W., Sheplak, M.: Modern developments in shear-stress measurement. Progress in Aerospace Sciences38(6-7), 515–570 (2002) https://doi.org/10.1016/ S0376-0421(02)00031-3

    2002

  26. [26]

    Journal of Fluid Mechanics830, 760–796 (2017) https://doi.org/10.1017/jfm.2017.580

    Suzuki, T., Hasegawa, Y.: Estimation of turbulent channel flow atRe τ = 100 based on the wall measurement using a simple sequential approach. Journal of Fluid Mechanics830, 760–796 (2017) https://doi.org/10.1017/jfm.2017.580

    work page doi:10.1017/jfm.2017.580 2017

  27. [27]

    Physical Review Fluids4(11), 114603 (2019) https://doi.org/10.1103/ PhysRevFluids.4.114603

    Encinar, M.P., Jim´ enez, J.: Logarithmic-layer turbulence: a view from the wall. Physical Review Fluids4(11), 114603 (2019) https://doi.org/10.1103/ PhysRevFluids.4.114603

    2019

  28. [28]

    Physics of Fluids33(7), 075121 (2021) https://doi.org/10.1063/5.0058346

    G¨ uemes, A., Discetti, S., Ianiro, A., Sirmacek, B., Azizpour, H., Vinuesa, R.: From coarse wall measurements to turbulent velocity fields through deep learning. Physics of Fluids33(7), 075121 (2021) https://doi.org/10.1063/5.0058346

    work page doi:10.1063/5.0058346 2021

  29. [29]

    Theoretical and Computational Fluid Dynamics39(1), 13 (2025)

    Guastoni, L., Balasubramanian, A.G., Foroozan, F., G¨ uemes, A., Ianiro, A., Dis- cetti, S., Schlatter, P., Azizpour, H., Vinuesa, R.: Fully convolutional networks for velocity-field predictions based on the wall heat flux in turbulent boundary layers. Theoretical and Computational Fluid Dynamics39(1), 13 (2025)

    2025

  30. [30]

    In: Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP), pp

    Cho, K., Merri¨ enboer, B., Gulcehre, C., Bahdanau, D., Bougares, F., Schwenk, H., Bengio, Y.: Learning phrase representations using RNN encoder–decoder for sta- tistical machine translation. In: Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP), pp. 1724–1734 (2014)

    2014

  31. [31]

    International Journal of Heat and Fluid Flow (2026) https: //doi.org/10.2139/ssrn.6877226

    Cavallazzi, G.M., P´ erez-Cuadrado, M., Pinelli, A.: Deep reinforcement learning with spatial and temporal awareness for active boundary control of buoyancy- driven convection. International Journal of Heat and Fluid Flow (2026) https: //doi.org/10.2139/ssrn.6877226 . Preprint, SSRN:6877226

    work page doi:10.2139/ssrn.6877226 2026

  32. [32]

    The European Physical Journal E46(4), 27 (2023)

    Guastoni, L., Rabault, J., Schlatter, P., Azizpour, H., Vinuesa, R.: Deep rein- forcement learning for turbulent drag reduction in channel flows. The European Physical Journal E46(4), 27 (2023)

    2023

  33. [33]

    Journal of Fluid Mechanics960, 30 (2023) https://doi.org/10.1017/jfm

    Sonoda, T., Liu, Z., Itoh, T., Hasegawa, Y.: Reinforcement learning of control strategies for reducing skin friction drag in a fully developed turbulent channel flow. Journal of Fluid Mechanics960, 30 (2023) https://doi.org/10.1017/jfm. 2023.147 31

    work page doi:10.1017/jfm 2023

  34. [34]

    Physics of Fluids32(9), 095108 (2020)

    Han, B.-Z., Huang, W.-X.: Active control for drag reduction of turbulent channel flow based on convolutional neural networks. Physics of Fluids32(9), 095108 (2020)

    2020

  35. [35]

    Physical Review Fluids8(2), 024604 (2023)

    Lee, T., Kim, J., Lee, C.: Turbulence control for drag reduction through deep reinforcement learning. Physical Review Fluids8(2), 024604 (2023)

    2023

  36. [36]

    Actuators11(12) (2022) https://doi.org/10.3390/act11120359

    Varela, P., Su´ arez, P., Alc´ antara-´Avila, F., Mir´ o, A., Rabault, J., Font, B., Garc´ ıa-Cuevas, L.M., Lehmkuhl, O., Vinuesa, R.: Deep reinforcement learning for flow control exploits different physics for increasing Reynolds number regimes. Actuators11(12) (2022) https://doi.org/10.3390/act11120359

    work page doi:10.3390/act11120359 2022

  37. [37]

    Journal of Fluid Mechanics1006, 12 (2025) https://doi.org/10.1017/jfm.2025.27

    Zhou, Z., Zhang, M., Zhu, X.: Reinforcement-learning-based control of turbulent channel flows at high Reynolds numbers. Journal of Fluid Mechanics1006, 12 (2025) https://doi.org/10.1017/jfm.2025.27

    work page doi:10.1017/jfm.2025.27 2025

  38. [38]

    Flow, Turbulence and Combustion115, 1291–1317 (2025) https: //doi.org/10.1007/s10494-024-00609-4

    Cavallazzi, G.M., Guastoni, L., Vinuesa, R., Pinelli, A.: Deep reinforcement learning for the management of the wall regeneration cycle in wall-bounded tur- bulent flows. Flow, Turbulence and Combustion115, 1291–1317 (2025) https: //doi.org/10.1007/s10494-024-00609-4

    work page doi:10.1007/s10494-024-00609-4 2025

  39. [39]

    Annual Review of Fluid Mechanics52, 477–508 (2020)

    Brunton, S.L., Noack, B.R., Koumoutsakos, P.: Machine learning for fluid mechanics. Annual Review of Fluid Mechanics52, 477–508 (2020)

    2020

  40. [40]

    Computers & Fluids 225, 104973 (2021) https://doi.org/10.1016/j.compfluid.2021.104973

    Garnier, P., Viquerat, J., Rabault, J., Larcher, A., Kuhnle, A., Hachem, E.: A review on deep reinforcement learning for fluid mechanics. Computers & Fluids 225, 104973 (2021) https://doi.org/10.1016/j.compfluid.2021.104973

    work page doi:10.1016/j.compfluid.2021.104973 2021

  41. [41]

    Physics of Fluids35(3), 031301 (2023)

    Vignon, C., Rabault, J., Vinuesa, R.: Recent advances in applying deep rein- forcement learning for flow control: perspectives and future directions. Physics of Fluids35(3), 031301 (2023)

    2023

  42. [42]

    In: Proceedings of the 14th International Conference on Artificial Intelligence and Statistics (AISTATS)

    Ross, S., Gordon, G.J., Bagnell, J.A.: A reduction of imitation learning and structured prediction to no-regret online learning. In: Proceedings of the 14th International Conference on Artificial Intelligence and Statistics (AISTATS). Proceedings of Machine Learning Research, vol. 15, pp. 627–635 (2011)

    2011

  43. [43]

    In: Proceedings of the 39th International Conference on Machine Learning (ICML)

    Brandstetter, J., Welling, M., Worrall, D.E.: Lie point symmetry data augmenta- tion for neural PDE solvers. In: Proceedings of the 39th International Conference on Machine Learning (ICML). Proceedings of Machine Learning Research, vol. 162, pp. 2241–2256 (2022)

    2022

  44. [44]

    In: Advances in Neural Information Processing Systems 33 (NeurIPS) (2020)

    Um, K., Brand, R., Fei, Y., Holl, P., Thuerey, N.: Solver-in-the-loop: learning from differentiable physics to interact with iterative PDE solvers. In: Advances in Neural Information Processing Systems 33 (NeurIPS) (2020)

    2020

  45. [45]

    Computers & Mathematics with Appli- cations76(8), 1853–1862 (2018)

    Costa, P.: A FFT-based finite-difference solver for massively-parallel direct 32 numerical simulations of turbulent flows. Computers & Mathematics with Appli- cations76(8), 1853–1862 (2018)

    2018

  46. [46]

    Comput- ers & Mathematics with Applications81, 502–511 (2021)

    Costa, P., Phillips, E., Brandt, L., Fatica, M.: GPU acceleration of CaNS for massively-parallel direct numerical simulations of canonical fluid flows. Comput- ers & Mathematics with Applications81, 502–511 (2021)

    2021

  47. [47]

    In: Proceedings of the Platform for Advanced Scientific Computing Conference (PASC ’22) (2022)

    Romero, J., Costa, P., Fatica, M.: Distributed-memory simulations of turbulent flows on modern GPU systems using an adaptive pencil decomposition library. In: Proceedings of the Platform for Advanced Scientific Computing Conference (PASC ’22) (2022)

    2022

  48. [48]

    John Wiley & Sons, Chichester (2005)

    Skogestad, S., Postlethwaite, I.: Multivariable Feedback Control: Analysis and Design, 2nd edn. John Wiley & Sons, Chichester (2005)

    2005

  49. [49]

    Journal of Fluid Mechanics54(1), 39–48 (1972)

    Wallace, J.M., Eckelmann, H., Brodkey, R.S.: The wall region in turbulent shear flow. Journal of Fluid Mechanics54(1), 39–48 (1972)

    1972

  50. [50]

    Journal of Fluid Mechanics60(3), 481–511 (1973)

    Lu, S.S., Willmarth, W.W.: Measurements of the structure of the Reynolds stress in a turbulent boundary layer. Journal of Fluid Mechanics60(3), 481–511 (1973)

    1973

  51. [51]

    Journal of Fluid Mechanics694, 100– 130 (2012) 33

    Lozano-Dur´ an, A., Flores, O., Jim´ enez, J.: The three-dimensional structure of momentum transfer in turbulent channels. Journal of Fluid Mechanics694, 100– 130 (2012) 33

    2012