arxiv: 2605.10737 · v1 · submitted 2026-05-11 · ⚛️ physics.flu-dyn

Recognition: 2 theorem links

· Lean Theorem

Cross-correlating blade--wake dynamics for a model wind turbine

Francisco J. G. de Oliveira, Oliver R. H. Buxton, Zahra Sharif Khoadei

Authors on Pith no claims yet

Pith reviewed 2026-05-12 04:35 UTC · model grok-4.3

classification ⚛️ physics.flu-dyn

keywords wind turbineblade wake interactioncross-correlationfluid-structure couplingtip-speed ratiowake shear layersturbulence modulation

0 comments

The pith

Experiments reveal that blade strain fluctuations precede wake velocity fluctuations, suggesting blades shape their wakes.

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

In experiments with a model wind turbine, researchers simultaneously measured blade strain using fibre-optic sensors and wake velocities at various downstream locations. They observed that the blade's strain dynamics depend strongly on the tip-speed ratio, while turbulence has a secondary effect. Zero-lag correlations between blade and wake were negligible, but lagged cross-correlations showed responses localized in shear layers at rotation frequencies. A consistent negative lag indicates blade strain changes occur before wake velocity changes, implying the blades drive aspects of the wake structure. This coupling is important for understanding fatigue and performance in dense wind farms.

Core claim

The authors demonstrate through time-synchronised measurements that blade strain fluctuations systematically precede downstream wake velocity fluctuations, as evidenced by a consistent negative-lag peak in cross-correlation analysis. This suggests a causal, blade-driven imprint on the wake. The coupling is spatially localised within the wake shear layers, organised around rotation-coherent frequencies, and peaks at intermediate downstream locations, with the operating condition (tip-speed ratio) playing the dominant role over free-stream turbulence.

What carries the argument

The negative-lag peak observed in cycle-averaged cross-correlation and cross-power spectral density between distributed blade strain and spatially resolved wake velocity.

If this is right

The amplitude, coherence, and temporal/spectral organisation of blade structural dynamics are strongly governed by tip-speed ratio.
Free-stream turbulence primarily modulates the blade responses shaped by operating condition.
Wake-induced blade response is spatially localised within the wake shear layers.
Coupling strength peaks at intermediate downstream locations and is organised around rotation-coherent frequencies.
A negative time lag shows blade strain precedes wake velocity, indicating blade-driven wake imprint.

Where Pith is reading between the lines

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

This causal blade-to-wake link may create feedback in wind farms where upstream blades modify wakes affecting downstream turbines.
Turbine control could use blade strain data to anticipate and reduce wake-induced fatigue loads.
The approach of concurrent strain and velocity measurements could be applied to other rotating fluid machines like fans or pumps to study similar couplings.

Load-bearing premise

The observed negative time lag between blade strain fluctuations and downstream wake velocity fluctuations reflects a true causal blade-to-wake influence rather than an artifact of sensor placement, flow convection, or post-processing in the cross-correlation analysis.

What would settle it

Performing the cross-correlation with velocity probes at varying distances or with a non-rotating blade to check if the negative lag persists would test if it is due to blade-driven effects.

Figures

Figures reproduced from arXiv: 2605.10737 by Francisco J. G. de Oliveira, Oliver R. H. Buxton, Zahra Sharif Khoadei.

**Figure 2.** Figure 2: FIG. 2. Convection length scale of tip-generated flow struc [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

**Figure 3.** Figure 3: FIG. 3. Autocorrelation function [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗

**Figure 4.** Figure 4: FIG. 4. First zero-crossing of [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗

**Figure 5.** Figure 5: FIG. 5. Joint-probability density function ( [PITH_FULL_IMAGE:figures/full_fig_p008_5.png] view at source ↗

**Figure 6.** Figure 6: FIG. 6. Cycle-averaged cross-correlation [PITH_FULL_IMAGE:figures/full_fig_p009_6.png] view at source ↗

**Figure 7.** Figure 7: FIG. 7. Extended cycle-averaged cross-correlation [PITH_FULL_IMAGE:figures/full_fig_p010_7.png] view at source ↗

**Figure 8.** Figure 8: FIG. 8 [PITH_FULL_IMAGE:figures/full_fig_p011_8.png] view at source ↗

**Figure 9.** Figure 9: FIG. 9. Energy integration of [PITH_FULL_IMAGE:figures/full_fig_p012_9.png] view at source ↗

**Figure 10.** Figure 10: FIG. 10. Zones of dominant frequencies in a wind turbine’s [PITH_FULL_IMAGE:figures/full_fig_p013_10.png] view at source ↗

**Figure 11.** Figure 11: FIG. 11. Distribution of Θ( [PITH_FULL_IMAGE:figures/full_fig_p014_11.png] view at source ↗

read the original abstract

Understanding how wakes interact with wind turbine blades under varying operating and inflow conditions is essential for improving fatigue prediction and performance assessment in increasingly dense wind farms. We present an experimental investigation of wake-blade coupling in a model wind turbine, focusing on the role of tip-speed ratio, $\lambda$, under varying free-stream turbulence conditions. Spatially resolved wake velocity measurements are acquired concurrently with distributed blade strain measurements using Rayleigh backscattering fibre-optic sensing, enabling direct, time-synchronised analysis of fluid-structure interaction across the blade's span. The blades' strain dynamics are strongly governed by $\lambda$, where variations of the operating condition of the turbine modify the amplitude, coherence, and the temporal/spectral organisation of the blade's structural dynamics, while free-stream turbulence primarily modulates these responses. Instantaneous joint statistics reveal negligible zero-lag dependence between wake velocity and blade strain, motivating a lagged and frequency-resolved analysis. Cycle-averaged cross-correlation and cross-power spectral density analyses demonstrate that wake-induced blade response is spatially localised within the wake shear layers and organised around rotation-coherent frequencies, with the coupling strength peaking at intermediate downstream locations. These results highlight the dominant role of operating condition in shaping wake-mediated blade loading and demonstrate the value of concurrent, spatially resolved flow-structure measurements for resolving blade-exciting flow dynamics in wind-turbine wakes. Furthermore, a consistent negative-lag peak indicates that blade strain fluctuations systematically precede downstream wake velocity fluctuations, suggesting a causal, blade-driven imprint on the wake.

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.

Desk Editor's Note private letter to a colleague

The negative-lag cross-correlation between blade strain and wake velocity is the clearest new observation, but its causal interpretation needs an explicit check against convection time and possible shared drivers.

read the letter

The paper reports a consistent negative time lag in cycle-averaged cross-correlations between blade strain and downstream wake velocity on a model turbine. They read this as the blades leaving an imprint on the wake ahead of the measurement station. That specific lagged relationship, localized to the shear layers and tied to rotation-coherent frequencies, is not something I recall from the earlier wake literature they cite, so it is worth noting for fatigue work in wind farms. The concurrent use of Rayleigh backscattering fibre-optic strain sensing across the blade span with spatially resolved wake data is a practical technical step that lets them separate tip-speed ratio effects from free-stream turbulence reasonably cleanly. The finding that operating condition dominates the structural response and that coupling peaks at intermediate downstream stations comes through from the joint statistics and spectral analysis. Those parts are grounded in direct measurements and standard processing, with no free parameters or circular fitting. The main soft spot is the causal claim. The abstract gives no downstream distance, mean velocity, or lag convention, so it is not possible to test whether the observed lag simply equals the convection time from rotor to probe or whether the sign convention in the correlation definition produces the negative peak. A common upstream fluctuation that reaches the blade first would also generate the same pattern without any blade-to-wake driving. If the full paper does not include those checks or raw time-series examples, the interpretation stays provisional. This work is for people doing experimental fluid-structure studies in wind energy or turbine wake modeling. A reader already working on blade loading or wake mitigation would get usable data points from the measurement approach and the localization results. It is coherent enough on its own terms to deserve peer review, mainly to get the convection and synchronization details on record and to see whether the lag survives basic sensitivity tests.

Referee Report

3 major / 2 minor

Summary. The manuscript presents an experimental investigation of blade-wake coupling on a model wind turbine. Concurrent Rayleigh backscattering fibre-optic strain measurements along the blade span and spatially resolved wake velocity measurements are used to examine the effects of tip-speed ratio λ and free-stream turbulence. Cycle-averaged cross-correlation and cross-power spectral density analyses show that wake-induced blade responses are localized in shear layers and organized at rotation-coherent frequencies, with negligible zero-lag correlation. A central result is a consistent negative-lag peak in the cross-correlation, interpreted as evidence that blade strain fluctuations causally precede and imprint on downstream wake velocity fluctuations.

Significance. If the negative-lag interpretation is robust, the work supplies direct experimental evidence of rotor-to-wake causality that can inform fatigue-load models and wake-evolution predictions in wind farms. The distributed fibre-optic sensing approach for span-resolved, time-synchronized fluid-structure data is a methodological strength that enables the reported joint statistics.

major comments (3)

[Results section on cycle-averaged cross-correlation] The negative-lag peak is load-bearing for the causal claim, yet the manuscript provides no numerical lag value, no downstream probe distance, and no mean velocity to compare against the expected convection time x/U. Without these quantities it is impossible to confirm that the observed lag differs from simple advection or to exclude a convection artifact.
[Methods and data-processing subsections] The cross-correlation definition and lag convention (whether C(τ) correlates blade(t) with wake(t+τ) or the reverse) are not stated explicitly, nor are any corrections for sensor synchronization, filtering, or cycle-averaging procedures. These details are required to verify that the sign of the peak is not an artifact of post-processing.
[Discussion of causality] Alternative explanations for the negative lag, such as a shared upstream inflow fluctuation that reaches the blade before the wake probe, are not quantitatively addressed. The claim of a 'blade-driven imprint' therefore rests on an untested assumption that no common driver or flow feature produces the observed ordering.

minor comments (2)

[Abstract and corresponding results] The abstract states that 'free-stream turbulence primarily modulates these responses' but the results do not quantify the relative effect sizes of λ versus turbulence intensity on coherence or amplitude.
[Figure captions] Figure captions for the cross-correlation plots should include the exact downstream locations and the number of cycles averaged to allow reproducibility.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for their careful and constructive review. The comments correctly identify areas where additional quantitative detail and explicit discussion will strengthen the manuscript. We have revised the text to incorporate the requested information on lags, cross-correlation definitions, processing steps, and alternative explanations, while preserving the original experimental results and interpretations.

read point-by-point responses

Referee: [Results section on cycle-averaged cross-correlation] The negative-lag peak is load-bearing for the causal claim, yet the manuscript provides no numerical lag value, no downstream probe distance, and no mean velocity to compare against the expected convection time x/U. Without these quantities it is impossible to confirm that the observed lag differs from simple advection or to exclude a convection artifact.

Authors: We agree that these quantities are necessary to substantiate the interpretation. In the revised manuscript we have added the measured negative-lag peak value, the downstream probe location, and the local mean velocity. We now explicitly compute the advection time x/U and demonstrate that the observed negative lag is inconsistent with pure convection, thereby supporting that blade strain fluctuations precede the wake velocity fluctuations. revision: yes
Referee: [Methods and data-processing subsections] The cross-correlation definition and lag convention (whether C(τ) correlates blade(t) with wake(t+τ) or the reverse) are not stated explicitly, nor are any corrections for sensor synchronization, filtering, or cycle-averaging procedures. These details are required to verify that the sign of the peak is not an artifact of post-processing.

Authors: We accept that the lag convention and processing steps must be stated unambiguously. The revised Methods section now provides the exact definition of the cycle-averaged cross-correlation (with the sign convention that negative τ corresponds to blade leading wake), the synchronization protocol between the fibre-optic and PIV systems, the applied filtering, and the phase-averaging procedure over multiple rotations. These additions confirm that the negative peak is not produced by post-processing choices. revision: yes
Referee: [Discussion of causality] Alternative explanations for the negative lag, such as a shared upstream inflow fluctuation that reaches the blade before the wake probe, are not quantitatively addressed. The claim of a 'blade-driven imprint' therefore rests on an untested assumption that no common driver or flow feature produces the observed ordering.

Authors: We have expanded the Discussion to address this point directly. We now include a quantitative estimate of the convection time from an upstream location to both the blade and the wake probe, showing that a common inflow fluctuation would produce a positive rather than negative lag. In addition, the observed spatial localization of the correlation to the shear layers and its organization at rotation-coherent frequencies are inconsistent with a spatially uniform upstream driver. These arguments, drawn from the existing data, support the blade-driven interpretation while acknowledging that simultaneous upstream measurements would provide further confirmation. revision: yes

Circularity Check

0 steps flagged

No circularity: purely experimental measurements and standard statistics

full rationale

The paper reports an experimental study using direct, time-synchronised measurements of blade strain (via fibre-optic sensing) and wake velocity. All reported quantities—cycle-averaged cross-correlations, cross-power spectral densities, and observed negative-lag peaks—are computed from raw data via standard statistical operations with no fitted parameters, derivations, or model equations that could feed back into the inputs. No self-citations, ansatzes, or uniqueness theorems are invoked to support the central claim; the negative-lag interpretation is presented as an empirical observation rather than a derived result. The analysis chain is therefore self-contained against external benchmarks and contains no load-bearing steps that reduce to their own inputs by construction.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The work rests on standard experimental fluid-dynamics assumptions (accurate synchronization of strain and velocity signals, negligible probe interference, and that cycle-averaging removes random turbulence while preserving rotation-coherent signals). No new entities are postulated and no free parameters are fitted to produce the reported lag result.

axioms (2)

domain assumption Time synchronization between fibre-optic strain and wake velocity measurements is accurate to within the sampling interval
Invoked implicitly when computing zero-lag and lagged cross-correlations; any systematic offset would invert the sign of the reported negative lag.
domain assumption Cycle-averaging isolates rotation-coherent dynamics without introducing spurious phase relationships
Used to compute the cross-power spectral density organized around rotation frequencies.

pith-pipeline@v0.9.0 · 5578 in / 1446 out tokens · 35959 ms · 2026-05-12T04:35:44.134556+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear
Cycle-averaged cross-correlation and cross-power spectral density analyses demonstrate that wake-induced blade response is spatially localised within the wake shear layers and organised around rotation-coherent frequencies
IndisputableMonolith/Foundation/ArrowOfTime.lean arrow_from_z unclear
a consistent negative-lag peak indicates that blade strain fluctuations systematically precede downstream wake velocity fluctuations

Reference graph

Works this paper leans on

42 extracted references · 42 canonical work pages

[1]

World Wind Energy Association, WWEA Half-year Re- port 2024 (2024). 15

work page 2024
[2]

Thomsen and P

K. Thomsen and P. Sørensen, Journal of Wind Engineer- ing and Industrial Aerodynamics80, 25 (1999)

work page 1999
[3]

Biswas and O

N. Biswas and O. R. H. Buxton, Journal of Fluid Me- chanics979, A34 (2024)

work page 2024
[4]

Maldonado, L

V. Maldonado, L. Castillo, A. Thormann, and C. Mene- veau, Journal of Wind Engineering and Industrial Aero- dynamics142, 246 (2015)

work page 2015
[5]

Thompson, H

C. Thompson, H. Biler, S. Symon, and B. Ganapathisub- ramani, Journal of Fluid Mechanics974, A9 (2023)

work page 2023
[6]

L. P. Chamorro and F. Port´ e-Agel, Boundary-Layer Me- teorology132, 129 (2009)

work page 2009
[7]

L. P. Chamorro, R. E. A. Arndt, and F. Sotiropoulos, Wind Energy15, 733 (2012)

work page 2012
[8]

L. J. Vermeer, J. N. Sørensen, and A. Crespo, Progress in Aerospace Sciences39, 467 (2003)

work page 2003
[9]

M. O. L. Hansen,Aerodynamics of Wind Turbines, 3rd ed. (Routledge, Abingdon, Oxon, and New York, 2015)

work page 2015
[10]

Gambuzza and B

S. Gambuzza and B. Ganapathisubramani, Journal of Fluid Mechanics963, 10.1017/jfm.2023.302 (2023)

work page doi:10.1017/jfm.2023.302 2023
[11]

Bourhis, T

M. Bourhis, T. Messmer, M. H¨ olling, and O. R. H. Bux- ton, Journal of Fluid Mechanics1023, A3 (2025)

work page 2025
[12]

E. L. Hodgson, N. Troldborg, and S. J. Andersen, Re- newable Energy238, 121804 (2025)

work page 2025
[13]

Port´ e-Agel, M

F. Port´ e-Agel, M. Bastankhah, and S. Shamsoddin, Boundary-Layer Meteorology174, 1 (2020)

work page 2020
[14]

Biswas and O

N. Biswas and O. R. H. Buxton, arXiv2506.23728, 10.48550/arXiv.2506.23728 (2025)

work page doi:10.48550/arxiv.2506.23728 2025
[15]

B. Wen, Z. Li, Z. Jiang, X. Tian, X. Dong, and Z. Peng, Ocean Engineering199, 107061 (2020)

work page 2020
[16]

Trigaux, P

F. Trigaux, P. Chatelain, and G. Winckelmans, Wind Energy Science9, 1765 (2024)

work page 2024
[17]

Bernardi, S

C. Bernardi, S. Cherubini, F. Manganelli, G. D. Posta, S. Leonardi, and P. D. Palma, Large eddy simulation of the iea 15-mw wind turbine using a two-way coupled fluid-structure interaction model (2025)

work page 2025
[18]

F. J. G. de Oliveira, M. Bourhis, Z. S. Khodaei, and O. R. H. Buxton, Wind Energy Science Discussions2025, 1 (2025)

work page 2025
[19]

Moens and P

M. Moens and P. Chatelain, Frontiers in Energy Research 10, 10.3389/fenrg.2022.881532 (2022)

work page doi:10.3389/fenrg.2022.881532 2022
[20]

Moens, M

M. Moens, M. Duponcheel, and P. Chatelain, Wind En- ergy25, 1860 (2022)

work page 2022
[21]

L. Zhou, X. Shen, L. Ma, H. Ouyang, and Z. Du, Energy 336, 10.1016/j.energy.2025.138488 (2025)

work page doi:10.1016/j.energy.2025.138488 2025
[22]

Messmer, J

T. Messmer, J. Peinke, A. Croce, and M. H¨ olling, Jour- nal of Fluid Mechanics1018, 10.1017/jfm.2025.10509 (2025)

work page doi:10.1017/jfm.2025.10509 2025
[23]

van den Berg, D

D. van den Berg, D. van der Hoek, D. D. Tavernier, J. Gutknecht, and J. W. van Wingerden, Wind Energy Science11, 679 (2026)

work page 2026
[24]

Y. Zhou, R. M. C. So, W. Jin, H. G. Xu, and P. K. C. Chan, Experiments in Fluids27, 359 (1999)

work page 1999
[25]

F. J. G. de Oliveira, Z. S. Khodaei, and O. R. H. Buxton, Journal of Fluid Mechanics1011, A47 (2025)

work page 2025
[26]

Pacheco, F

J. Pacheco, F. Pimenta, S. Guimar˜ aes,et al., Engineering Structures300, 117140 (2024)

work page 2024
[27]

Shroff andet

S. Shroff andet. al.,Manufactured and laboratory tested scaled blades and parts of the blade, Deliverable D2.24 (INNWIND.EU, 2017)

work page 2017
[28]

C. Li, J. Tang, C. Cheng, L. Cai, and M. Yang, Photonic Sensors11, 91 (2021)

work page 2021
[29]

N. Yang, J. Zhang, L. Yu, P. Zheng, B. Zhao, H. Sun, and Q. Liu, Optik267, 169764 (2022)

work page 2022
[30]

Xu and Z

C. Xu and Z. S. Khodaei, Sensors20, 4040 (2020)

work page 2020
[31]

F. J. G. de Oliveira, Z. S. Khodaei, and O. R. H. Buxton, Experiments in Fluids65, 134 (2024)

work page 2024
[32]

Li and Z

Y. Li and Z. Sharif-Khodaei, International Journal of Me- chanical Sciences286, 109859 (2025)

work page 2025
[33]

Coscetta, A

A. Coscetta, A. Minardo, L. Olivares, M. Mirabile, M. Longo, M. Damiano, and L. Zeni, Journal of Sensors 2017, 9175342 (2017), open access

work page 2017
[34]

Z. Tang, H. Liu, C. Wang, Y. Song, H. Tian, X. Duan, W. Zhang, B. Zhang, and W. Chen, High Voltage , hve2.70057 (2025)

work page 2025
[35]

Campagnolo,Wind Tunnel Testing of Scaled Wind Turbine Models: Aerodynamics and Beyond, Ph.D

F. Campagnolo,Wind Tunnel Testing of Scaled Wind Turbine Models: Aerodynamics and Beyond, Ph.D. the- sis, Politecnico di Milano (2013)

work page 2013
[36]

Yadala, S

S. Yadala, S. Dehareng, I. Neunaber,et al., Journal of Fluid Mechanics1019, A38 (2025)

work page 2025
[37]

Crespo, J

A. Crespo, J. Hern´ andez, and S. Frandsen, Wind Energ. 2, 1 (1999)

work page 1999
[38]

Sherry, A

M. Sherry, A. Nemes, D. Lo Jacono, H. M. Blackburn, and J. Sheridan, Physics of Fluids25, 117102 (2013)

work page 2013
[39]

Chatfield,The Analysis of Time Series: An Intro- duction, 4th ed

C. Chatfield,The Analysis of Time Series: An Intro- duction, 4th ed. (Chapman and Hall, London, 1989) pp. 94–95

work page 1989
[40]

M. J. van den Broek, M. Debnath, M. P. van der Laan, S. J. Andersen, and J. N. Sørensen, Wind Energy Science 8, 1909 (2023)

work page 1909
[41]

Bossuyt, R

J. Bossuyt, R. Scott, N. Ali, and R. B. Cal, Journal of Fluid Mechanics917, A3 (2021)

work page 2021
[42]

Bastankhah and F

M. Bastankhah and F. Port´ e-Agel, Journal of Fluid Me- chanics806, 506 (2016)

work page 2016