the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
From the center of wind pressure to loads on the wind turbine: A stochastic approach for the reconstruction of load signals
Abstract. In the context of the wind industry, there is an increasing need for a more comprehensive understanding of the atmospheric wind, particularly with respect to wind structures, which have not been thoroughly investigated in the current standard guidelines. This necessity arises in light of the current trends toward larger, higher, and more flexible wind turbine designs. Of particular importance are the correlations between the yet-to-be-characterized atmospheric turbulent structures and the specific responses of the turbines. These correlations may be crucial in assessing load events relevant to new designs that were negligible for the earlier, smaller, and stiffer turbines. The Center of Wind Pressure (CoWP) (Schubert et al., 2025) was recently introduced as a feature of a wind field that characterizes large-scale wind structures and, at the same time, correlates with the large-scale or low-frequency content of the bending moments at the main shaft of the wind turbines. In this paper, we comprehensively compare the CoWP and the bending moments in terms of their statistical properties and fatigue estimates, quantified by Damage Equivalent Loads (DEL). Furthermore, a stochastic method for the reconstruction of synthetic CoWP signals is proposed. The strong correlation with the bending moments allows the proposed stochastic CoWP model to be used as a surrogate and relatively simple estimator of the large-scale dynamics of these loads, which is based solely on the properties of the inflow wind field. A notable advantage of the approach is the capability to reconstruct very long time series, which are critical for assessing loads over the operational lifetime of the turbine. As an alternative, the proposed stochastic model of the CoWP from atmospheric data can be integrated as an extension of existing wind turbulence models, thereby accurately reproducing the dynamics of large-scale wind structures inherent to the CoWP.
Competing interests: An author is a member of the editorial board of Wind Energy Science.
Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this preprint. The responsibility to include appropriate place names lies with the authors.- Preprint
(5972 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 18 Jun 2025)
-
RC1: 'Comment on wes-2025-78', Anonymous Referee #1, 12 Jun 2025
reply
This study presents a novel stochastic approach to estimate and reconstruct load signals on wind turbines based on atmospheric wind field data. The focus lies on the Center of Wind Pressure (CoWP), a recently introduced descriptor that characterizes large-scale turbulent structures within a wind field and exhibits good correlations with the low-frequency content of bending moments at the turbine’s main shaft. Using both synthetic (Kaimal model) and real (GROWIAN) wind data, the authors aim to demonstrate that the CoWP can capture wind dynamics often overlooked in standard models and enable a reduced-complexity, yet accurate, representation of turbine loads. The Langevin stochastic model is used to reconstruct synthetic CoWP signals, with the aim of preserving the statistical features and Damage Equivalent Load (DEL) metrics of the original turbine response data. The method should allow the generation of long-duration load time series without requiring computationally intensive simulations. The comparisons reported in the paper show good statistical agreement between reconstructed and original signals, validating the model’s effectiveness.
The paper is well written, and the topic is of interest to the scientific community. However, several aspects should be improved:
The method proposed by the authors is intended to enable a fast estimation of the Damage Equivalent Load (DEL) over very long time histories (years). The comparison between DEL, DEL_CoWP, and DEL_CoWP_R is satisfactory, but not excellent: the slope of the interpolation line is clearly below one, and there is a notable spread of data around the fitted line. The authors do not comment on how this discrepancy might affect the DEL estimation over long time periods. Should we expect differences in the order of 1%, 5%, 10%, 30%, or even 50%?
The method can predict DELs associated with the low-frequency component of the loads. However, the overall DEL also includes high-frequency loads, which are characterized by many cycles. The authors do not comment on the difference between DELs from low-frequency load components and total DELs.
The simulations using Kaimal data are performed at a single mean wind speed, under partial load conditions (region II). What is the impact of turbine control on the accuracy of the proposed method? It would be useful to compare the results with wind speeds above rated.
There is no discussion on how the proposed method could be integrated into current engineering practice, which involves estimating total DELs (including both low- and high-frequency components) from 10-minute simulations. In other words, how can the DELs associated with low-frequency atmospheric variations (predictable using the CoWP or its surrogate) be integrated with those due to high-frequency atmospheric fluctuations?
The analysis focuses on DELs. However, wind turbines are also designed to withstand ultimate loads. Could the proposed approach be useful in this context as well? Is there any correlation between CoWP and ultimate hub loads?
The section on data availability is missing. If possible, the authors should share both the data and the scripts used.
Specific comments:
-
Section 2.2 (Damage Equivalent Load): This section summarizes well-known information about DEL calculation, already available in IEC guidelines. It should be omitted.
-
Figure 8: What exactly do the whiskers represent? The caption mentions they indicate the most extreme data points, but does not specify how these are defined (e.g., 95% or 99% confidence interval?).
-
Lines 221–222: The cut-off frequency used for filtering CoWP and loads should be close to the 3P frequency. However, the authors use 0.1 Hz, which is significantly lower than the 3P frequency of the NREL 5MW turbine operating at rated power (approximately 0.6 Hz). The reason for this choice is unclear. The type and order of the filter should also be specified.
-
Figure 8: Why is the correlation between DEL and DEL_CoWP better for the T_yaw loads (left-right CoWP displacement) than for T_tilt loads (up-down CoWP displacement)?
-
Figures 10, 16, D2: The CoWP is defined with respect to the center of the rotor disk. Therefore, the CoWP_z should oscillate around 90 m (Kaimal data) or 125 m (GROWIAN data). This is not evident in the figures.
-
Figures 8a, 8b, 19a, 19b: It would be helpful to include the RMSE values.
Citation: https://doi.org/10.5194/wes-2025-78-RC1 -
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
133 | 14 | 4 | 151 | 7 | 7 |
- HTML: 133
- PDF: 14
- XML: 4
- Total: 151
- BibTeX: 7
- EndNote: 7
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1