the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 11 Jan 2024
Analysis of Turbine Yaw Misalignment Estimated by LIDAR Assuming Homogeneous Flow
Abstract. The 2D homogeneous flow assumption derived wind field reconstruction method is widely employed in the Doppler LIDAR. This paper aims to analyse the uncertainty in wind direction estimation and to improve the estimation accuracy. First, to quantify the uncertainty, a static model is proposed to describe the relationship between horizontal wind shear and yaw misalignment in LIDAR measurement. Subsequently, an analytic model of temporal-averaged misalignment uncertainty is built by using the Kaimal turbulence spectral model. This analytic model reveals that the standard deviation of yaw misalignment reaches approximately ±14° and ±12° for IEC turbulence class 'A' and 'B', respectively, regarding a temporal average over 60s. Obviously, these findings demonstrate that this LIDAR estimation method is insufficient to supervise the turbine yaw control system in terms of both accuracy and timeliness. Then, the temporal-averaged uncertainties obtained from the proposed analytic model are compared with simulations in various complexity, i.e. Kaimal, Mann spectral models, and Computational Fluid Dynamics, respectively. The rotor-average wind is set as the reference for measurement. Compared to an ideal sonic measurement, the LIDAR presents a worse estimation of rotor-effective wind direction estimation. These results show that increasing the fidelity of turbulence models does not alleviate the measurement uncertainty issue. Lastly, in an attempt to address the uncertainty issue, this study investigates the effects of adjusting the scanning pattern. The optimised parameters include measurement distance and horizontal half-open angle in a 2-beam LIDAR case and the additional vertical half-open angle in a 4-beam LIDAR case. However, even with the optimal scanning pattern, u and v wind components estimation could not acquire the same accuracy on the ideal sonic measurement simultaneously. Thereby, LIDAR can not guarantee wind speed and direction estimation quality simultaneously. In conclusion, this study highlights the yaw misalignment uncertainty in the LIDAR wind field reconstruction method based on 2D homogeneous flow assumption. The observed error levels remain consistent across varying fidelity turbulence models and scanning pattern adjustments. To address this challenge, future research should apply more advanced wind flow models to explore more accurate wind field reconstruction methods.
This preprint has been withdrawn.
-
Withdrawal notice
This preprint has been withdrawn.
-
Preprint
(2313 KB)
Interactive discussion
Status: closed
-
RC1: 'Comment on wes-2023-162', Jennifer Rinker, 27 Jan 2024
General comments
This work has potential to be of scientific significance to the wind energy community, but unfortunately it has far too many issues in quality to be acceptable. These issues ranged from poor writing to more egregious errors in basic derivations of the fundamental equations used in the analysis. The problems were so serious that I stopped reviewing partway through the paper, as I believe any subsequent analyses were based on faulty assumptions and thus would be meaningless. The idea is interesting, and I would very much like to see a proper treatment of the scientific content, but this manuscript is unfortunately insufficient in both scientific and presentation quality.
Specific comments
I have many – please see comments in attached PDF. Yellow highlight indicates something to be addressed.
Technical corrections
I have many – please see comments in attached PDF.
-
RC2: 'Comment on wes-2023-162', Anonymous Referee #2, 11 Feb 2024
Zhang et al (2024) analyse nacelle-based lidar measurement accuracy of inflow wind speed and direction with different scanning strategies. This is an important field of research to qualify nacelle-based lidars for control applications. The authors conduct a series of virtual measurements in synthetically generated turbulence and LES simulations which is a sound method to optimize the measurement setup. They show the wind direction estimation error for different averaging times and optimize the setup using the measurement coherence bandwidth. In the end, the manuscript fails to give error estimates of the optimized scanning pattern and a clear statement on the achievable uncertainties in comparison to the requirements. The figures and the text are not well prepared and difficult to read. At this point, I cannot recommend the manuscript for publication in WES without significant improvements. I suggest to revise structure and text completely and resubmit.
General comments:
- Abstract: It is not really clear what the actual goal is? determine wind direction and yaw misalignment angle from lidar? Estimate the uncertainty? Optimize scanning patterns?
- Introduction: I miss an explanation on the requirement for wind direction estimation for control purposes. At which frequencies / averaging times should a yaw controller work? What are acceptable uncertainties? What is the goal for the controller? Maximum power or minimum loads? What is the state of the art?Â
- From my point of view, the term "yaw misalignment" is not very clearly and sufficiently defined. In my understanding yaw misalignment is the average error between turbine yaw and wind direction in 10-minute averages. But here, it is used for the error of wind direction estimation in a turbulent flow. I find this a bit misleading.
- The language of the manuscript makes it very difficult to understand the content. Significant improvement in English grammar is necessary. Even the virtual measurement setup with the lidars is hard to understand. I had to read some paragraphs multiple times back and forth to get the basic idea. Maybe a schematic would help to understand the geometry of the scans and the distances at which data are analyzed. The resolution of the simulations in comparison to the measurement resolution is not sufficiently explained.
- While the MCB is probably a good parameter for optimization of scan patterns, an estimation of the uncertainty in physical variables should follow in the end. I wonder if the optimization goal is chosen very clearly. Trying to beat an "ideal sonic" is maybe not a good strategy. What is the targeted uncertainty and at which scale / averaging time? What is the quantitative outcome? How does the target and the result compare? The conclusions remain very vague, qualitative and pessimistic. How far is the result from the target and what could be possible changes in the measurement setup to reach the target values?Specific comments:
p.2. l.38: what does "adaptive yaw speed" mean?
p.4, Eq.3: I think this equation is trivial and can be removed.
p.5f, Sect. 2.6 and 2.7: I think a description of the grid size and resolution for Mann model and LES would be helpful here.
p.6, ll. 144ff: This is a very strange and misleading introduction to LES. I think it would make more sense to describe which code with which resolution and numerical setup is used.
p.6, l.159: "from the of the"
p.7, l.162f: $d$ is first used in Eq. 6 but only introduced here. $\delta_H$ has already been introduced in the context of Eq. 6
p.8, l.188: "wind components is identical in a certain turbulence model". I think it is important to specify this explicitly here. You do not mean wind components as in $u$ and $v$, but longitudinal wind speed at two separate points $u_1$ and $u_2$.
p.8, Eq. 18: I find the notation with the $R$ for the magnitude a bit strange. Absolute value bars would be more intuitive.
p.9, l.194: "substituting Eq. 20". Equation 20 follows later and I do not understand how this is substituted herein.
p.9, Eq. 20: $sigma$ is not a function, but a variable name, so the $e$ should go in the index, not parantheses.
p.10, l.217: "section ??".
p.10, Fig.3: It is hard to see differences between the two plots, especially because the axis are not scaled equally. It is completely unclear what the colored dots mean.
p.12, l.253: I do not understand this sentence "lidar earns much more advantages..."
p.12, l.264: I do not understand "consistently consistent..."
p.12, l.269: It is not important for the reader if it is extracted with MATLAB, but the temporal and spatial resolution would be interesting.
p.13, l.278: Is the mean not zero because the grid of the LES is not aligned with the mean wind direction?
p.14, l.280: I find this observation very vague: "similar results". I compare Fig. 5 and Fig. 8, but the differences in absolute values in turbulence intensity in the quality of the fit are so different at first glance that I do not know how to evaluate such a statement.
p.15, l.301: "the current LIDAR scanning pattern": that means the pattern that has been regarded in Section 4?
p.15, l.305: I disagree that "MCB is usually used as a criterion to evaluate the measurement quality of LIDAR". It may have been used by co-authors of this study a lot in context of lidar-assisted control, but outside this topic, it is not a very common metric.Â
p.15, l.307: is that really the goal, to be more robust than the ideal sonic anemometer? In reality there is no ideal sonic anemometer and what you actually want is the REWS I assume, so I wonder why to even compare with the ideal sonic anemometer.
p.16, l.326f: I really doubt that this is a surprising result and that it should be the goal of the optimization.
p.18, l.356f: What are the consistent results, what did you find? What is the trend?
p.18, l.362f: What is the estimated uncertainty? Is it larger than the requirements? What are the practical limits?Citation: https://doi.org/10.5194/wes-2023-162-RC2
Interactive discussion
Status: closed
-
RC1: 'Comment on wes-2023-162', Jennifer Rinker, 27 Jan 2024
General comments
This work has potential to be of scientific significance to the wind energy community, but unfortunately it has far too many issues in quality to be acceptable. These issues ranged from poor writing to more egregious errors in basic derivations of the fundamental equations used in the analysis. The problems were so serious that I stopped reviewing partway through the paper, as I believe any subsequent analyses were based on faulty assumptions and thus would be meaningless. The idea is interesting, and I would very much like to see a proper treatment of the scientific content, but this manuscript is unfortunately insufficient in both scientific and presentation quality.
Specific comments
I have many – please see comments in attached PDF. Yellow highlight indicates something to be addressed.
Technical corrections
I have many – please see comments in attached PDF.
-
RC2: 'Comment on wes-2023-162', Anonymous Referee #2, 11 Feb 2024
Zhang et al (2024) analyse nacelle-based lidar measurement accuracy of inflow wind speed and direction with different scanning strategies. This is an important field of research to qualify nacelle-based lidars for control applications. The authors conduct a series of virtual measurements in synthetically generated turbulence and LES simulations which is a sound method to optimize the measurement setup. They show the wind direction estimation error for different averaging times and optimize the setup using the measurement coherence bandwidth. In the end, the manuscript fails to give error estimates of the optimized scanning pattern and a clear statement on the achievable uncertainties in comparison to the requirements. The figures and the text are not well prepared and difficult to read. At this point, I cannot recommend the manuscript for publication in WES without significant improvements. I suggest to revise structure and text completely and resubmit.
General comments:
- Abstract: It is not really clear what the actual goal is? determine wind direction and yaw misalignment angle from lidar? Estimate the uncertainty? Optimize scanning patterns?
- Introduction: I miss an explanation on the requirement for wind direction estimation for control purposes. At which frequencies / averaging times should a yaw controller work? What are acceptable uncertainties? What is the goal for the controller? Maximum power or minimum loads? What is the state of the art?Â
- From my point of view, the term "yaw misalignment" is not very clearly and sufficiently defined. In my understanding yaw misalignment is the average error between turbine yaw and wind direction in 10-minute averages. But here, it is used for the error of wind direction estimation in a turbulent flow. I find this a bit misleading.
- The language of the manuscript makes it very difficult to understand the content. Significant improvement in English grammar is necessary. Even the virtual measurement setup with the lidars is hard to understand. I had to read some paragraphs multiple times back and forth to get the basic idea. Maybe a schematic would help to understand the geometry of the scans and the distances at which data are analyzed. The resolution of the simulations in comparison to the measurement resolution is not sufficiently explained.
- While the MCB is probably a good parameter for optimization of scan patterns, an estimation of the uncertainty in physical variables should follow in the end. I wonder if the optimization goal is chosen very clearly. Trying to beat an "ideal sonic" is maybe not a good strategy. What is the targeted uncertainty and at which scale / averaging time? What is the quantitative outcome? How does the target and the result compare? The conclusions remain very vague, qualitative and pessimistic. How far is the result from the target and what could be possible changes in the measurement setup to reach the target values?Specific comments:
p.2. l.38: what does "adaptive yaw speed" mean?
p.4, Eq.3: I think this equation is trivial and can be removed.
p.5f, Sect. 2.6 and 2.7: I think a description of the grid size and resolution for Mann model and LES would be helpful here.
p.6, ll. 144ff: This is a very strange and misleading introduction to LES. I think it would make more sense to describe which code with which resolution and numerical setup is used.
p.6, l.159: "from the of the"
p.7, l.162f: $d$ is first used in Eq. 6 but only introduced here. $\delta_H$ has already been introduced in the context of Eq. 6
p.8, l.188: "wind components is identical in a certain turbulence model". I think it is important to specify this explicitly here. You do not mean wind components as in $u$ and $v$, but longitudinal wind speed at two separate points $u_1$ and $u_2$.
p.8, Eq. 18: I find the notation with the $R$ for the magnitude a bit strange. Absolute value bars would be more intuitive.
p.9, l.194: "substituting Eq. 20". Equation 20 follows later and I do not understand how this is substituted herein.
p.9, Eq. 20: $sigma$ is not a function, but a variable name, so the $e$ should go in the index, not parantheses.
p.10, l.217: "section ??".
p.10, Fig.3: It is hard to see differences between the two plots, especially because the axis are not scaled equally. It is completely unclear what the colored dots mean.
p.12, l.253: I do not understand this sentence "lidar earns much more advantages..."
p.12, l.264: I do not understand "consistently consistent..."
p.12, l.269: It is not important for the reader if it is extracted with MATLAB, but the temporal and spatial resolution would be interesting.
p.13, l.278: Is the mean not zero because the grid of the LES is not aligned with the mean wind direction?
p.14, l.280: I find this observation very vague: "similar results". I compare Fig. 5 and Fig. 8, but the differences in absolute values in turbulence intensity in the quality of the fit are so different at first glance that I do not know how to evaluate such a statement.
p.15, l.301: "the current LIDAR scanning pattern": that means the pattern that has been regarded in Section 4?
p.15, l.305: I disagree that "MCB is usually used as a criterion to evaluate the measurement quality of LIDAR". It may have been used by co-authors of this study a lot in context of lidar-assisted control, but outside this topic, it is not a very common metric.Â
p.15, l.307: is that really the goal, to be more robust than the ideal sonic anemometer? In reality there is no ideal sonic anemometer and what you actually want is the REWS I assume, so I wonder why to even compare with the ideal sonic anemometer.
p.16, l.326f: I really doubt that this is a surprising result and that it should be the goal of the optimization.
p.18, l.356f: What are the consistent results, what did you find? What is the trend?
p.18, l.362f: What is the estimated uncertainty? Is it larger than the requirements? What are the practical limits?Citation: https://doi.org/10.5194/wes-2023-162-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 368 | 101 | 25 | 494 | 20 | 20 |
- HTML: 368
- PDF: 101
- XML: 25
- Total: 494
- BibTeX: 20
- EndNote: 20
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1