the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 25 Jul 2025
Bidirectional wakes over complex terrain using SCADA data and wake models
Abstract. We investigate bidirectional wake effects between two identical wind turbines in a hill region of northern Japan using Supervisory Control and Data Acquisition (SCADA) data and validate the performance of 12 wake models. The two turbines are located 3.7 times a rotor diameter apart with a different elevation of half of a rotor diameter. We identify the wake effects in terms of wind speed ratio, which is defined as a ratio of wind speed at the downstream wind turbine to that at the upstream wind turbine. By comparing the conditions according to the operating state of the upstream wind turbine, the wakes are clearly detected as minimum wind speed ratios for northeasterly and southwesterly winds. The wind speed ratio decreases with inflow wind speed below the rated wind speed. Increase in turbulence intensity and decrease in power output are greater for southwesterly wind than for northeasterly wind because of the combined effects of the turbine-induced wake and the terrain-induced reduction in wind speed. Then, we compare the simulated wakes from the validate the wake models implemented in PyWake software by using simulated wind fields derived from Wind Atlas Analysis and Application Program (WAsP) Computational Fluid Dynamics (CFD). The wind speed ratios derived from the models show strong dependence on inflow wind speed, reflecting the thrust curve used in the engineering wake models. The wake models commonly overestimate the reduction in wind speed for northeasterly wind and underestimate it for southwesterly wind. Thus, this study demonstrates that additional topographic effects alter the wake effects.
- Preprint
(3104 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on wes-2025-130', Anonymous Referee #1, 04 Aug 2025
This paper addresses a practical topic of bidirectional wake effects in complex terrain using SCADA data and wake modeling. The comparative studies contributed to the selection of wake models in complex terrain. The following comments are intended to help strengthen the manuscript for potential publication.
- Abstract: The abstract lists the work and results but does not clearly articulate the motivation and contributions of the study. It is recommended that the authors restructure the abstract to begin with a broader background, narrow down to the specific focus on bidirectional wake effects, and end with a stronger conclusion that clearly states the novelty and significance of the work.
- Article Structure: The overall structure of the paper could be reorganized; for example, lines 54-63 seem more appropriate in the methodology section rather than the introduction. Additionally, some subheadings could be more informative. For instance, a title such as "2.1 Two Turbines" is too generic, and “2.2 SCADA” does not concisely describe the content of the section.
- Introduction: The background does not effectively introduce the primary object of the study. Moreover, literature reviews were unable to identify the progress and the key research gaps of the research. It is recommended to supplement the review with more recent and relevant work.
- Methods: The theoretical framework is unclear. The authors defined the “wind speed ratio” and conducted analyses based on it. It is recommended to provide a mathematical formula and a detailed explanation. Similarly, it is better to present the calculation of turbulence intensity.
- Validation and Analysis: For this wind field test, using wake models to validate the observed SCADA data seems unreasonable. Similarly, the use of a CFD approach in Section 3.2 as a supplementary analysis of bidirectional wake effects is not fully convincing, as the CFD model itself has not been sufficiently validated for this application.
- Reference: The bibliography contains a number of outdated references and lacks literature from the past five years that can reflect the current research status.
- Figures: Four separate figures are used to illustrate the relative positions of the two turbines, which appear redundant. It is recommended to consolidate these into a more informative figure to improve the conciseness and information density of the manuscript.
- Language and Readability: The manuscript would benefit from a thorough proofread to correct grammatical errors (e.g., "is critical issue" should be "is a critical issue" in the introduction). Attention should also be paid to improving sentence structure and logical coherence; in the same paragraph, shifts in voice (active vs. passive) and subject detract from readability.
Citation: https://doi.org/10.5194/wes-2025-130-RC1 -
RC2: 'Comment on wes-2025-130', Anonymous Referee #2, 18 Aug 2025
Review of “Bidirectional wakes over complex terrain using SCADA data and wake models” by N. Sasanuma and co-authors.
This paper describes observations made using SCADA data over roughly two years of operation of two wind turbines sited on complex terrain in northern Japan. The orientation of the two turbines is roughly southwest-northeast. WT2 is slightly elevated compared to WT1. WT2 is upstream for southwesterly wind (2250) and WT1 is upstream when the wind is northeasterly (450). Analysis of the wind speed ratio (wind speed at downstream turbine divided by wind speed at upstream turbine) and turbulence intensity shows that wake effects of the upstream turbine are felt on the downstream turbine only for southwesterly wind (i.e. when WT2 is upstream). Wake effects are not felt on the downstream turbine for the northeasterly wind (when WT1 is upstream). This is attributed to the effect of topography and because WT2 is at a higher elevation as compared to WT1. Twelve analytical or semi-analytical wake models available in PyWake are evaluated for their ability to predict the wake effects and show mixed results with errors ranging from under 1 % to 66 %.
Overall, the paper is well-written and easy to read. Real-field data are always scarce, and hence valuable, and this paper fills this gap. The presentation and interpretation of the results could be improved a bit. The effect of stability, which can have a non-trivial impact, is completely ignored in the paper. The section on evaluation of analytical models is quite superficial and should be improved substantially. Specific comments that the authors should address are given below.
Major Points:
1. The major takeaway from this study seems to be that if an upstream turbine is at a higher elevation than the downstream turbine, its wake affects the downstream turbine. If the upstream turbine is at a lower elevation than the downstream turbine, its wake does not affect the downstream turbine.
If this above understanding is correct, the authors should try to find evidence as to whether this is supported by other observations/experiments/simulations. One possible explanation is likely in recent wind-tunnel experiments on complex terrain (Chen et al., Applied Energy, 2025. 10.1016/j.apenergy.2024.125044). This study shows that the wake of a turbine sited on a hilltop follows the terrain and bends down along with the surface. However, the wake of a turbine sited upstream of the hill does not bend upwards and follow the terrain in a similar manner.
The authors should consider whether this aligns with their observations and include a discussion on this. Other similar observations/experiments/simulation results would also strengthen the main argument of this paper.
2. Are atmospheric stability effects important at this site? Hypothetically, it is possible that the southwesterly wind is always accompanied by stable atmospheric conditions, under which the wake is known to persist longer, and northeasterly wind is always accompanied by unstable conditions, under which the wake recovers faster. Thus, the significant wake effects observed (not observed) for southwesterly (northeasterly) wind could hypothetically be simply due to thermal stratification. Can this explanation be ruled out using some data, or is this a valid explanation for the observations?
3. The section on wake model evaluation lacks many details. For example,
(a) what is the inflow provided to the models? Is it a uniform in the vertical, or a sheared profile?
(b) Line 91 mentions ‘bush trees’ on the terrain. Is the effect of these trees incorporated into the wake models through, e.g., a canopy displacement height, or through an aerodynamic roughness length?
(c) Each of the 12 wake models have at least one (probably more) tunable parameters which can drastically change their predictions. The authors have not mentioned what values were assigned to these parameters. If some ‘default’ values were used, those should be mentioned. Also, it is crucial to mention sensitivity of the predictions to these parameters.
(d) Is there any particular reason that the ‘Bastankhah’ and ‘TurboGaussian’ models are more accurate than the others?
(e) The two models seem to be performing better when the wind speed is set to a certain value. However, wake models are generally agnostic to wind speed since they predict normalized velocity deficits. The only way wind speed enters into a wake model is through the thrust coefficient that gets modulated with wind speed. Is this the reason for the two models to be performing better in certain wind regimes than others?
In view of the above points, perhaps it would be better to focus on a smaller number of models more thoroughly than to superficially show results of a dozen models.
Minor Points:
- Is Fig 1a really needed? It should be sufficient to mention latitude and longitude of the site of study. Is there anything specific that is being conveyed by map at this scale?
- Similarly, I am not sure if the photograph in Fig. 2a is really necessary.
- The difference in the elevations between the two turbines seems to be 0.44D ((169-132)/83.3) rather than 0.5D. This should be mentioned precisely rather than rounding off to 0.5D.
- Appendix A used to correct wind direction and wind speed due to turbine rotation is not clear. Lines 385-395 should be expanded and the method should be explained in more detail. Are there any previous references that can be cited for this?
- Since the ‘Bastankhah’ and ‘TurboGaussian’ models seem to be performing the best, they should be explained at least briefly, maybe in an appendix. Also, please provide a reference on line 164.
- The paragraph around line 150 is inconsistent. North-easterly wind implies WT1 is upstream and WT2 is downstream, and wind speed ratio is defined as WS2/WS1. This means the wind speed ratio is defined as the ratio of wind at downstream turbine to that at upstream turbine. The first sentence of this paragraph states the opposite. Please clarify this.
- Line 155: It is slightly awkward to say that computational models are used to validate the observations. It should be the other way around. Perhaps ‘consistency check’ or something similar would be a better phrase here.
- Line 165: Is the power curve of the actual turbine model not available, and can’t it be implemented into the PyWake code? How do the power curves of the J82-2.0 and Vestas V80 turbines compare with each other?
- 5 can be augmented with CP of the V80 turbine for completeness.
- 7(a) and 7(c) deal with no-wake conditions. Here, there is an average ratio of 1.2 in 7(a) and of a little below 1 in 7(c). This speedup/slowdown is entirely because of terrain effects. Can this be explained using, say, the WAsP simulation described later?
- Labels (a), (b) etc can be reduced in size in almost all figures.
- 7: How are ‘no-wake’ and ‘wake’ conditions identified? Why is there no wake at certain time instants? Is it because the upstream turbine is not operating, and measurements at such time instants are called ‘no-wake’? Or is the ‘no-wake’ condition due to other effects such as lateral deflection of the wake due to yaw misalignment? Where does the wake ‘go’ under ‘no-wake’ conditions?
- Line 213: Minor typo: “To exam the maximum wake effects…”
- Are the lines corresponding to turbulence levels A and A+ (Fig. 9) described by an analytical expression? If so, can these expressions be provided?
- 9(a) and 10(a) are both for northeasterly wind, when WT1 is upstream and WT2 is downstream. The wind speed at WT2 (i.e. x axis) in Fig. 9(a) goes till 18 m/s while in Fig. 10(a) it goes till only 6 m/s. Are these plots consistent with each other?
- 10: what is the power normalized with?
- In Appendix B or Table 1, references for the 12 models should be provided.
- Why is the WaSP simulation not performed along the same line joining the turbines?
- Line 358: “Additional effects of topography to the wake effects cause opposite changes in the simulated wakes.” It is unclear what this line means. Where has this been shown in the paper?
Citation: https://doi.org/10.5194/wes-2025-130-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 239 | 26 | 16 | 281 | 15 | 17 |
- HTML: 239
- PDF: 26
- XML: 16
- Total: 281
- BibTeX: 15
- EndNote: 17
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1