the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 21 Jan 2026
Aeroelastic Validation of ExaWind for the Pazy Wing Wind Tunnel Experiment
Abstract. This paper presents the results of a validation campaign of the fluid-structure interaction capability of the ExaWind software suite using the Pazy wing case, an aeroelastic benchmark featuring large nonlinear deformations of a very flexible wing under low-speed conditions. The simulations used the incompressible Reynolds-averaged Navier–Stokes equations equipped with a Shear-Stress Transport turbulence model for the fluid dynamics, together with a geometrically exact nonlinear beam model for the structural dynamics. The simulations yielded predictions of pre-flutter static deflections and flutter onset speeds that demonstrated strong agreement with both wind-tunnel data and the computational results contributed to the Large Deflection Working Group of the 3rd Aeroelastic Prediction Workshop.
- Preprint
(7914 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on wes-2025-263', Anonymous Referee #1, 06 Feb 2026
-
RC2: 'Comment on wes-2025-263', Anonymous Referee #2, 18 Mar 2026
The paper "Aeroelastic validation of ExaWind for the Pazy Wing wind tunnel experiment" describes in detail how the simulations have been carried out, which is more than sufficient for others to do the same thing including all the assumptions and simplifications that have been made. This description takes 12 pages, but the final results (including the conclusions) need one page of written text and two plots. Here the paper lacks of discussion and interpretation of the results, while the conclusions read more like a summary than a conclusion part. I will try to explain what I think is missing.
Figure 5 shows the results for the ExaWind (two versions) simulations, the experimental results and the results from a workshop. One 1/3 page the authors describe what is seen in that figure without trying to explain what can bee seen. Here the experimental results are shown as lines but in the text they mention that the differences between the two ExaWind simulations are small compared to the uncertainties of the measurements. If the uncertainties are so significant, it would be good to add them to the plot so the reader can also see them. If these uncertainties are so big, what is the goal of comparing the results to the mean values? Where are the uncertainties coming from. What determines a "good" result?
In the text the authors also mention, that the experimental results depend on the direction of the sweep -- for which sweep direction are the experimental results, which are shown here? Â Why did the chose this direction of sweep for the comparison? Â What is the difference in terms of aerodynamics between the sweep in the experiments compared to the fixed angle fixed wind speed in the simulations? When changing the AoA dynamically, like in the AoA sweep, results depend on the rate of change of the AoA. This is not specified or even mentioned. The same might also be true for the velocity -- the experimental conditions should be mentioned in more detail and also be considered in the discussion of the results.
The results from the workshop are only shown as a grey area -- how many different simulations contributed to these results? Are there others that perform maybe even better than the ExaWind ? If yes, what could be the reason for it. What are the known limitations that others should take into account before using this method?  Why not only showing the best results from the workshop and see how they compare to the ExaWind? What is the novelty with ExaWind that people should use that even though others might perform better?  Right now, it looks to me  that there is just another method that can be used and I don't think that this is the take home message.Â
In figure 5 the x-axis is in dynamic pressure instead of  wind speed in m/s -- why?Â
The situation for figure 6 is somehow similar. Here the range is much wider compared to the workshop and the stable condition for AoA of 3° ist missing. Was ist not possible to run another simulation after these?
In figure 6 the abbreviations for the methods are different than before which makes it harder to understand whats going on. Why is the flutter region also added when only the onset of flutter is important? What about uncertainties in these results?
Â
I would highly recommend to extend the results part and add discussion.
Â
Â
Â
Â
Â
Â
Â
Â
Citation: https://doi.org/10.5194/wes-2025-263-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 276 | 117 | 16 | 409 | 13 | 17 |
- HTML: 276
- PDF: 117
- XML: 16
- Total: 409
- BibTeX: 13
- EndNote: 17
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Review of Â
Aeroelastic Validation of ExaWind for the Pazy Wing Wind Tunnel Experiment, by Neil Matula et al.Â
The paper is well written and addresses the relevant scientific questions of the accuracy of FSI style modeling of slender wings. The methods applied are not necessarily new, but the validation of the open source ExaWind software stack is relevant for the wind community and should have a broad international interest.Â
The paper is well structured, and the objective of documenting the capabilities of predicting stability of slender wings is clear. The scientific approach is generally well described, and only minor details are unclear.Â
In the present work, some approximation of the actual wind tunnel geometry and the wing setup is done. The present reviewer considers the approximations acceptable but is worried that neglecting the tip bar attached to the wing in the CFD setup might have some relevant influence on aerodynamics. It would be good with some discussion on this. One could make a very simple mesh around the tip bar and investigate the aero during a forced motion to evaluate the approximation.Â
As both eigenfrequencies and twist angles are available in the experiment, the present reviewer finds it relevant to compare these, especially as modern wind turbines have considerable torsional degree of freedom.Â
The comparisons with measurements are well discussed but should include the above discussion of twist and frequencies. I don’t expect that including the twist information will change the clear conclusion that the FSI can reproduce the exp.Â
The title is well chosen, and the abstract is to the point; the document is well structured and well written. The figures and tables are easily readable.Â
The main shortcoming of the paper is the references, where essentially only references withing the ExaWind community are given. It would be appropriate to position the work in context to other publications of blade resolved aeroelasticity in wind energy, and other codes used in the wind energy community, as this is the intended application.Â
Besides this shortcoming, I consider it an excellent and valuable paper.Â
Â
Mandatory changesÂ
It is mentioned in the text that the structural model is sub-stepping within each CFD time-step. It would be appropriate to mention how the loads are estimated at the intermediate time steps on the structural side.Â
Include comparison of eigenfrequencies (maybe a table) and twist angles.Â
The work should be positioned in an international context; there are several international groups among others at Univ. Stuttgart, DLR and DTU Wind and Energy Systems who have worked intensively on FSI of blade resolved flows.Â
Minor details and typosÂ
Â
Fig 4, Page 11: In my pdf the time step is shown as 10^5 not 10^-5 as intended etc.Â