the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 02 Dec 2024
Comparison of wind-farm control strategies under realistic offshore wind conditions: turbine quantities of interest
Abstract. Wind farm control is a strategy to increase the efficiency, and therefore lower the levelized cost of energy, a wind farm. This is done by using turbine settings such as the yaw angle, blade pitch angles, or generator torque to manipulate the wake behind the turbine affecting downstream turbines in the farm. Two inherently different wind farm control methods have been identified in literature: wake steering and wake mixing. This paper focuses on comparing the turbine quantities of interest between these methods for a simple two-turbine wind farm setup, while a companion article (Brown et al., 2025) focuses on the wake quantities of interest for a single wind turbine setup. Both papers use the same set of wind farm simulations based on high-fidelity large-eddy simulations (LES) coupled with OpenFAST turbine models. First, precursor simulations are executed in order to match wind conditions measured with lidars in an offshore wind farm off the US east coast. These measurements indicate general wind conditions that exhibit substantially higher vertical wind shear and veer than any of the LES studies performed with wind farm control strategies currently available in literature. The precursors are used to evaluate the effectiveness of the control methods. In the LES simulations, the wind veer leads to highly skewed wakes, which has considerable influence on the power uplift of wind farm control strategies. In addition to a baseline controller, four different control strategies, each of which uses either pitch or yaw control, are implemented on the upstream turbine of a simple two-turbine wind farm. Assuming the wind direction is known and constant over time, the simulations show that wake steering is generally the superior wind farm control strategy considering both wind farm power production and turbine damage equivalent loads (DELs) when substantial wind veer is present. This result is consistent over different wind speeds and wind directions. On the other hand, for similar wind conditions with lower veer, wake mixing was found to yield the highest power production, although at the expense of generally higher loads. This leads us to conclude that the effect of wind veer, which was so far not usually considered, can not be neglected when determining the optimal wind farm control strategy.
- Preprint
(3611 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on wes-2024-164', Anonymous Referee #1, 04 Jan 2025
The submitted paper presents a numerical comparison of several Wind Farm Flow Control (WFFC) strategies. Specifically, it investigates the performance of a tandem of wind turbines under different operating scenarios parametrized by inflow velocity and turbulence intensity. The analysis considers power production and Damage Equivalent Loads (DELs) as performance metrics. Alongside the baseline greedy controller, four additional controllers are evaluated: Wake Steering and three Active Wake Mixing (AWM) strategies (Pulse, Helix, and Side-to-Side).
The key finding is that Wake Steering generally outperforms AWM strategies in terms of power production, except under low veer conditions, where AWM yields higher power at the expense of increased loads.
Overall, the paper is well written and provides valuable insights into the impact of WFFC strategies on power production and turbine loads while also highlighting the limitations of the experimentation framework used.
However, several aspects of the paper require further clarification and adjustments, as detailed below. I recommend the publication once these comments have been addressed adequately.
Major Comments
- Sections 2.2 and 2.3.3: These sections have some degree of overlap and could be merged inside a single "Turbine control" subsection.
- Table 4: The simulations presented in Table 4 have different simulation lengths and cell sizes. Please provide a justification for these choices, especially considering that wake recovery can be sensitive to the mesh resolution.
- Line 212: Given that loads are considered in the study, more details on the specific OpenFAST settings used are required.
- Table 4: This table could be removed. It would recommend to reference the original publication and, instead, plot the power and thrust curves, along with the three wind speeds considered in this study.
- Line 236: The choice of yaw angle must be justified. The performance of yaw steering strategies (in terms of both power and loads) varies significantly with the yaw offset chosen. Clarify how the yaw angle was selected and demonstrate that it achieves the desired wake offset.
- Line 247: The phrase “a constant yaw alignment with respect to the mean wind direction” is ambiguous. In practice, defining the freestream velocity is not straightforward. Clarify whether the wind direction is assumed to be known or if it is dynamically measured by the simulated turbines.
- Precursor Setup: Given the critical role of veer in the study, the precursor simulation setup must be described in detail. Include the velocity profile generated and any adjustments made to match inflow conditions.
- Figure 7: As expected, the power uplift for AWM strategies climbs back to ~100% away from the wake. However, it is unclear why this is not the case for wake steering. Please review this figure or provide an explanation for this behavior.
- Figures 8, 10, 11, 12, 13, 14, and 15: The baseline controller’s performance should be more visible in these plots. Consider adding vertical and horizontal lines to highlight its location. Additionally, the font size for axis labels is too small. To facilitate comparison, I recommend merging some of these figures (e.g., cases 0.5D and -0.5D).
Minor Comments
- Line 30, Line 43, and subsequent occurrences: The term “set point” should be written as “setpoint”.
- Line 31: The term “Wind Farm Control” is too generic. Consider using “Wind Farm Flow Control” (WFFC) as defined by Meyers et al. (2022). "We define WFFC as the coordinated control of the turbines in the farm, with the aim of influencing the flow (wakes, turbulence) in such a way that it improves the overall figure of merit of the farm"
- Line 53: The phrase "allow or even depend" is confusing.
- Line 145: To differentiate vector notation from script, consider using the \mathbf notation.
- Table 1: The column heading "Desired Effect" is not ideal, as the desired effect is always to increase global power production. Consider a more descriptive heading.
- Line 231 and subsequent: A = 4 deg, degree need to be specified.
- Line 233: A reference is missing for the statement on the selection of yaw offset angles.
- Figures 3 and 4: Consider merging these figures to facilitate comparison. Ensure the font size is consistent and readable across all figures.
- Figure 5: Add a vertical line at 8.5D to indicate the location from which the data is extracted.
- Line 385: The statement about narrower wakes resulting in higher power production for the downstream turbine is valid. However, the narrowest wake would be achieved by shutting down the upstream turbine entirely. Please clarify that the aim is to achieve a narrower wake while maintaining a similar power output for the upstream turbine.
Citation: https://doi.org/10.5194/wes-2024-164-RC1 -
RC2: 'Comment on wes-2024-164', Anonymous Referee #2, 05 Jan 2025
General Comments
The paper presents interesting and highly relevant research on the effectiveness of different wind farm control methods, with special attention paid to the effect of veer.
Overall, the paper is of high scientific quality and the methodology is sound, clear and explicit.
The results are presented in a clear manner and the conclusions drawn are well supported by the presented results.Major Comments
I believe the studied LES simulation requires a more detailed description. The equations used and common quantities defining the inflow, such as roughness length, inversion height, coriolis parameter, etc. are missing. Also, the link provided in the reference list doesn't work. Similarly, the used turbine model should be described in more detail, such as what kind of solver is used (elastodyn, beamdyn) in openFAST and which degrees of freedom are activated.
Furthermore, I think the methodology of chapter 3.2.2 can be improved by considering a different method to calculate the power of a virtual turbine, for example by using the rotor equivalent wind speed (Wagner, R., Courtney, M., Gottschall, J., Lindelöw-Marsden, P., 2011. Accounting for the speed shear in wind turbine power performance measurement. Wind Energy 14, 993–1004. [https://doi.org/10.1002/we.509](https://doi.org/10.1002/we.509)) instead of current equation 3. Furthermore, I believe there are some typos in equations 3 and 4. I also find the notation used in section 2.1 not entirely clear. I suggest to clarify the difference between indices and name-particles by using italicized and non-italicized subscripts, respectively. (I especially stumbled over $u_{h,j,k}$, is $h$ supposed to be the index of the point in x at hub height?)
I find there are some limitations in the conclusions that could be addressed by the authors. Specifically, using a wind farm of two turbines and applying the control mechanism only at the first turbine limits the applicability of the findings. I believe this is especially relevant in the case of wake steering, as here the power losses at the actuated turbine are significant. Presumably, wind farm control would be applied to the first few rows in a large wind farm ( see, for example Howland, M.F., Lele, S.K., Dabiri, J.O., 2019. Wind farm power optimization through wake steering. PNAS 116, 14495–14500. [https://doi.org/10.1073/pnas.1903680116](https://doi.org/10.1073/pnas.1903680116)). Hence, reductions in power would also be observed at further downstream turbines. The authors should, at least, discuss this limitation, or, even better, provide additional results to quantify the difference.
Furthermore, the authors discuss the importance of uncertainty in wind direction on wake steering. I suggest to refer to work by Taschner et al. (Taschner, E., Becker, M., Verzijlbergh, R., Van Wingerden, J., 2024. Comparison of helix and wake steering control for varying turbine spacing and wind direction. J. Phys.: Conf. Ser. 2767, 032023. [https://doi.org/10.1088/1742-6596/2767/3/032023](https://doi.org/10.1088/1742-6596/2767/3/032023)).
In section 3.3.1 the authors write that the AWM achieve a narrowing of the wake. I would argue that this narrowing is a combined effect of the time-averaging and enhanced mixing. As this is not the focus of the paper I don't suggest to go into detail but rather rephrase in a more precise manner.
Finally, in the abstract and introduction the authors refer to the companion paper, however, in the rest of the paper no references are made. It would be nice if the authors linked their observations to the companion paper throughout the article. For example, in line 370 the author point out that the tower loads are most significantly affected by the side-to-side strategy. It would be nice to offer an explanation, or link to the companion paper if an explanation can be found there.Minor comments
l. 1: ... and therefore lower the levelized cost of energy, [of] a wind ...
l. 163: In my opinion, $C^\mathrm{opt}_\mathrm{P}$ should not depend on $\lambda$
l. 231: missing ° for the amplitude
l. 232: missing reference (rendered only as a ?)
Figure 1: I believe labels for side-to-side and pulse are mixed up
l. 258: $m$ is described as the slope of the Wöhler curve, but in table 6 its called the Wöhler exponent. I suggest to use the same description in both instances.
Figure 3: the x-axis label is X, but I believe it should be Y and the y-axis label is Y but should be z, similarly in the caption the hub height is given as y=... but it should be z.
Figure 4: same as figure 3, also the upper plot is referred to as a velocity profile but I would call it a velocity field or slice and not a profile.
Figure 5 (b): the slope marking 1:1 veer is green, but was black in 5a
Figure 6: in the upper figure, only round markers mark the rotor average velocity planes, while a dashed line with round markers is used in the lower plot. I suggest to use the same style in both plots to emphasize to the reader that they are based on the same calculation.
Figure 9: this figure is very large. Maybe the figure size can be reduced by showing less of the region upstream of the first turbine.
Table 8: Maybe a visual representation of this overview, such as a heatmap, is easier to digest.
Similarly table 9.Citation: https://doi.org/10.5194/wes-2024-164-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 148 | 39 | 3 | 190 | 2 | 1 |
- HTML: 148
- PDF: 39
- XML: 3
- Total: 190
- BibTeX: 2
- EndNote: 1
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1