the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Novel CFD approach for simulation of an ABL wind tunnel flow: validation and application to a FOWT model
Abstract. When comparing Large Eddy Simulations with wind tunnel experiments, choosing the appropriate boundary conditions is crucial to ensure an accurate representation of wind flow. This becomes particularly challenging for Atmospheric Boundary Layer (ABL) wind tunnels, which frequently incorporate calibrated obstacles to replicate the appropriate model-scale ABL flow. Although many researchers prefer to include these elements within the CFD domain, this approach leads to high computational costs and the necessity for precise replication of each obstacle. Alternatives to avoid this high computational cost typically include slip boundary conditions at the top, leading to fast decay in turbulence quantities near the ground. In this study, the authors propose a new methodology based on the precursor technique, which is commonly used for full-scale ABL simulations, combined with a stress top boundary condition. The method is validated against experimental measurements showing significant improvement in the inlet flow quality, when compared to previous methods.
Floating Offshore Wind Turbines (FOWT) are expected to experience significant growth in the coming decades. However, due to the effect of platform motions, their wake structures can be challenging to predict. As a second part of this study, the novel technique is applied together with an actuator disk to represent an oscillating wind turbine model, designed to study FOWT wakes. Simulations with varying turbulence intensities and motion frequencies are conducted. The results corroborate previous findings that the near wake is not significantly influenced by surge motion frequency, although certain frequency cases exhibit more persistent coherence structures than others, which results in a slower wake recovery. This phenomenon is found to be less significant in the context of higher turbulence intensities.
Competing interests: One of the co-authors is a member of the editorial board of the journal 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
(4994 KB) - Metadata XML
- BibTeX
- EndNote
Status: closed
-
RC1: 'Comment on wes-2025-2', Anonymous Referee #1, 10 Mar 2025
The reviewer has been trying to make sense of Figures 3 and 4 but he is still quite puzzled.
From the captions it seems both Figures 3 and 4 correspond to profiles at the inlet.
That is, the yellow line in Figure 2.
Do they include an average in the transversal direction over the inlet?
Do they involve a time average? Over which time period?
Since the domain to the left of the yellow line is not refined, and the flow at the yellow
line only depends on the flow in that region, the reviewer would expect the velocity results
in Figures 3 and 4 to match (unless there was insufficient averaging or some other problem).
However, the velocity profiles in Figures 3 and 4 are quite different. Do the authors have an explanation for this?Citation: https://doi.org/10.5194/wes-2025-2-RC1 -
AC1: 'Reply on RC1', Alejandro Daniel Otero, 12 Mar 2025
We thank the reviewer for taking the time to carefully review the manuscript. We would like to address the concerns raised as follows:
First, both Figures 3 and 4 are not plotted at the yellow line in Figure 2 but rather in the test section located 25D downstream of the second mesh refinement. This is the location where the Actuation Disk (AD) would be positioned in the subsequent analysis, as shown in Figure 2.
Figure 3 shows the velocity profile in the original domain before any mesh refinement was applied. This simulation is run for 1200 s, and the time average is computed over the last 600 s. Figure 4 corresponds to the velocity profile after mesh refinement, following 60 s of simulation after the precursor run. In this case, the time average is computed over the last 30 s.
These profiles do not include any transverse averaging.
We hope this clarification addresses the reviewer's concerns. We will improve the mentioned figures' description in the revised version of the manuscript.
Citation: https://doi.org/10.5194/wes-2025-2-AC1 -
RC2: 'Reply on AC1', Anonymous Referee #1, 12 Mar 2025
Based on the previous answer, we can attribute the difference between Figures 3 and 4 to the mesh. Taking into account that Figure 4 uses a finner mesh this would be the best solution. It is interesting to note that the black line in Figure 4, which corresponds to slip boundary conditions, is the one closer to experimental results. Please comment on this.
Citation: https://doi.org/10.5194/wes-2025-2-RC2 -
AC2: 'Reply on RC2', Alejandro Daniel Otero, 13 Mar 2025
Dear Reviewer #1,
Thank you for your comment. As you correctly point out, in Figure 4, the velocity profile corresponding to the slip boundary condition is indeed closer to the experimental results, although the difference is quite small. However, the turbulence quantities shown in the other plots of Figures 4 and 5 show much better agreement with the experimental data when the stress boundary condition is used. In particular, the behavior of the slip boundary condition shows an excess near the bottom and a steeper decrease, which deviates from the experimental results. In contrast, the stress boundary condition provides a closer match in both magnitude and profile shape, with a smaller surplus near the ground and a more gradual decrease.Citation: https://doi.org/10.5194/wes-2025-2-AC2
-
AC2: 'Reply on RC2', Alejandro Daniel Otero, 13 Mar 2025
-
RC2: 'Reply on AC1', Anonymous Referee #1, 12 Mar 2025
-
AC1: 'Reply on RC1', Alejandro Daniel Otero, 12 Mar 2025
-
RC3: 'Comment on wes-2025-2', Anonymous Referee #2, 13 Mar 2025
The paper describes a method for saving computational time for running neutral boundary layer cases for comparing with wind tunnel results. The model is then applied to a test case where the disc is moving sinusoidal in the streamwise direction.
The technique presented is trivial in its formulation and overly simplifies ABL flows. The computational time savings are not reported and many basic details are neither. The introduction does not clearly emphasize the need for this method nor is its benefit sufficiently shown in the remainder of the paper. The model is compared to a single wind tunnel case and deemed "validated". How the method improves wake studies of for- and aft moving disks remains unclear and thus also why that part is included in the paper at all. It is a far stretch to point out that the technique improves wake studies of floating turbines.
The authors need to rewrite the paper and include: 1) Why the method is needed 2) Clearly detail all numerical techniques 3) Motivate their validation choices 4) Actually validate by using different tunnels and ABL profiles 5) Report time savings 6) Make the FOAT study a side note, as it is just a test case or show how the "new" method actually changes wake behavior.
More detailed comments were unfortunately lost due to a technical glitch, apologies.Citation: https://doi.org/10.5194/wes-2025-2-RC3
Status: closed
-
RC1: 'Comment on wes-2025-2', Anonymous Referee #1, 10 Mar 2025
The reviewer has been trying to make sense of Figures 3 and 4 but he is still quite puzzled.
From the captions it seems both Figures 3 and 4 correspond to profiles at the inlet.
That is, the yellow line in Figure 2.
Do they include an average in the transversal direction over the inlet?
Do they involve a time average? Over which time period?
Since the domain to the left of the yellow line is not refined, and the flow at the yellow
line only depends on the flow in that region, the reviewer would expect the velocity results
in Figures 3 and 4 to match (unless there was insufficient averaging or some other problem).
However, the velocity profiles in Figures 3 and 4 are quite different. Do the authors have an explanation for this?Citation: https://doi.org/10.5194/wes-2025-2-RC1 -
AC1: 'Reply on RC1', Alejandro Daniel Otero, 12 Mar 2025
We thank the reviewer for taking the time to carefully review the manuscript. We would like to address the concerns raised as follows:
First, both Figures 3 and 4 are not plotted at the yellow line in Figure 2 but rather in the test section located 25D downstream of the second mesh refinement. This is the location where the Actuation Disk (AD) would be positioned in the subsequent analysis, as shown in Figure 2.
Figure 3 shows the velocity profile in the original domain before any mesh refinement was applied. This simulation is run for 1200 s, and the time average is computed over the last 600 s. Figure 4 corresponds to the velocity profile after mesh refinement, following 60 s of simulation after the precursor run. In this case, the time average is computed over the last 30 s.
These profiles do not include any transverse averaging.
We hope this clarification addresses the reviewer's concerns. We will improve the mentioned figures' description in the revised version of the manuscript.
Citation: https://doi.org/10.5194/wes-2025-2-AC1 -
RC2: 'Reply on AC1', Anonymous Referee #1, 12 Mar 2025
Based on the previous answer, we can attribute the difference between Figures 3 and 4 to the mesh. Taking into account that Figure 4 uses a finner mesh this would be the best solution. It is interesting to note that the black line in Figure 4, which corresponds to slip boundary conditions, is the one closer to experimental results. Please comment on this.
Citation: https://doi.org/10.5194/wes-2025-2-RC2 -
AC2: 'Reply on RC2', Alejandro Daniel Otero, 13 Mar 2025
Dear Reviewer #1,
Thank you for your comment. As you correctly point out, in Figure 4, the velocity profile corresponding to the slip boundary condition is indeed closer to the experimental results, although the difference is quite small. However, the turbulence quantities shown in the other plots of Figures 4 and 5 show much better agreement with the experimental data when the stress boundary condition is used. In particular, the behavior of the slip boundary condition shows an excess near the bottom and a steeper decrease, which deviates from the experimental results. In contrast, the stress boundary condition provides a closer match in both magnitude and profile shape, with a smaller surplus near the ground and a more gradual decrease.Citation: https://doi.org/10.5194/wes-2025-2-AC2
-
AC2: 'Reply on RC2', Alejandro Daniel Otero, 13 Mar 2025
-
RC2: 'Reply on AC1', Anonymous Referee #1, 12 Mar 2025
-
AC1: 'Reply on RC1', Alejandro Daniel Otero, 12 Mar 2025
-
RC3: 'Comment on wes-2025-2', Anonymous Referee #2, 13 Mar 2025
The paper describes a method for saving computational time for running neutral boundary layer cases for comparing with wind tunnel results. The model is then applied to a test case where the disc is moving sinusoidal in the streamwise direction.
The technique presented is trivial in its formulation and overly simplifies ABL flows. The computational time savings are not reported and many basic details are neither. The introduction does not clearly emphasize the need for this method nor is its benefit sufficiently shown in the remainder of the paper. The model is compared to a single wind tunnel case and deemed "validated". How the method improves wake studies of for- and aft moving disks remains unclear and thus also why that part is included in the paper at all. It is a far stretch to point out that the technique improves wake studies of floating turbines.
The authors need to rewrite the paper and include: 1) Why the method is needed 2) Clearly detail all numerical techniques 3) Motivate their validation choices 4) Actually validate by using different tunnels and ABL profiles 5) Report time savings 6) Make the FOAT study a side note, as it is just a test case or show how the "new" method actually changes wake behavior.
More detailed comments were unfortunately lost due to a technical glitch, apologies.Citation: https://doi.org/10.5194/wes-2025-2-RC3
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
160 | 40 | 9 | 209 | 6 | 6 |
- HTML: 160
- PDF: 40
- XML: 9
- Total: 209
- BibTeX: 6
- EndNote: 6
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1