the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Field comparison of load-based wind turbine wake tracking with a scanning lidar reference
Abstract. Wind farm control concepts require awareness and observation methods of the inner-farm flow field. The relative location of the wake, to which a downstream turbine is exposed, is of high interest. It can be used as feedback to support closed-loop wake-steering control, ultimately leading to higher power extraction and fatigue load reduction. With increasing fidelity, not only time-averaged wakes but also instantaneous wake conditions, subject to meandering and wind direction changes, are considered within a controller. This paper presents a quantitative field comparison of two independently applied wake centre estimation methods: a scanning lidar and an Extended Kalman Filter (EKF) based on the rotor loads of the waked turbine. No ground truth is available in the field environment, therefore the methodology accounts for the fact that two uncertain estimates are compared. The lidar estimates, with a derived uncertainty in the order of 0.05 rotor diameters D, can be used as a suitably precise reference to draw conclusions regarding the load-based EKF. The EKF uses Coleman-transformed blade root bending moments, linked to the wake centre position via an analytical model with a low number of tuning parameters. The model can easily be trained with aeroelastic simulations including the Dynamic Wake Meandering model. The formulation adds robustness to the tracking and allows to determine the confidence in the wake position estimate, which can be used for wake impingement detection or for a wake-steering controller to judge whether a yaw manoeuvre is adequate. The results indicate agreement of the methods with root-mean-square errors of 0.2 D for low and moderate turbulence intensity, and 0.3 D for turbulence intensities above 12 %. The paper focuses on wake position estimation but also outlines a methodology, how wind farm models or wind field reconstruction techniques can be validated with complementary lidar data.
- Preprint
(6632 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on wes-2024-188', Anonymous Referee #1, 06 Feb 2025
Dear authors,
I enjoyed reviewing your well-written and relevant publication on the wake location estimator. One aspect I really value is the consideration of uncertainty and the resulting bounds. My criticism is mainly related to the methodology presentation. Some sections would benefit from some clarification and additional information.
You can find my comments in the attached pdf
With kind regards
-
RC2: 'Comment on wes-2024-188', Anonymous Referee #2, 14 Jul 2025
Field comparison of load-based wind turbine wake tracking with a scanning lidar reference
This paper presents a method for tracking the wake center for downstream wind turbines. The method is used for wake-steering control, and two methods for implementation are considered: a scanning lidar and an Extended Kalman Filter (EKF) based on rotor loads. Results show 0.2D-0.3D RMS agreements for moderate to high turbulence levels.
VERDICT: It is very much appreciated that the authors implement their EKF method and compare it to scanning LIDAR in real-world wind turbines! Please state your contributions of this work very clearly in a pointwise list. I also think that the structure of the paper can be improved: sometimes you introduce things that are not defined yet, see comments below for further details. This especially holds for the methodology and results sections (2+3). I suggest a major revision before publication.
COMMENTS:
- While kind of discussed, it would help to state your contributions to this work explicitly in a pointwise list at the end of the introduction.
- Sect 2.1: You directly start describing the wind farm, while I would expect it would be more interesting to say something first about the scientific contribution you are bringing with your work. Consider changing the order.
- Fig XX: All figures need a more elaborate caption. Now the figures are not interpretable apart from the main text.
- Eq 3 to 7: very standard theory, really needed to include in this paper? Or make it more specific to your case. Also, explain why you assume 0 noise acting on the state and output.
- Eqs 8a-8d: Please elaborate more on this model. It seems very simple for the dynamics you want to capture. Is it linear? If yes, why do you need an EKF, and not a normal KF? Also, elaborate more about how a (linear?) combination of the chosen state vector elements leads to the 3 nonrotating blade moments. An elaborate explanation and justification of the dynamic model and chosen measurements are largely missing. ---> Ah, you explain this in the next subsection. Would it make sense to swap the order 2.2.3 and 2.2.2? So first fully define f() and h(), and then incorporate them into the state estimator.
- 2.2: Kind of a literature survey. Can it be largely moved to the introduction of the paper?
- 2.2.3: You use the Coleman transformation to obtain the nonrotating blade moments (tilt/yaw). It is well-known that for larger, more flexible rotors, you need some sort of decoupling strategy -- possibly in the Coleman transformation by an azimuth offset -- to obtain decoupled axes. You do seem to consider this aspect with the variable "d". Because it is a crucial aspect for larger flexible rotors, I highly recommend that you incorporate it into your research and elaborate more; there have been publications on this topic in the past.
- Sect 3.: I got lost in the structure of this section. Please announce what you will be discussing in the first part of the section (directly under 3.), and come up with a clearer structure, so that the storyline makes more sense.
- Sect 4.: Also, what is the purpose of this section? What will you discuss? Announce that at the start of the section.
MINOR:
- Often, a "?" appears when citing, check
Citation: https://doi.org/10.5194/wes-2024-188-RC2
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
246 | 74 | 21 | 341 | 20 | 32 |
- HTML: 246
- PDF: 74
- XML: 21
- Total: 341
- BibTeX: 20
- EndNote: 32
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1