Investigating wake reproduction of a model-scale wind turbine: experimental measurements versus Large Eddy Simulation with actuator line

Gillyns, Emmanuel; Buckingham, Sophia; van Beeck, Jeroen; Winckelmans, Grégoire

doi:10.5194/wes-2025-242

Preprints

https://doi.org/10.5194/wes-2025-242

Preprints

02 Dec 2025

| 02 Dec 2025

Status: a revised version of this preprint is currently under review for the journal WES.

Investigating wake reproduction of a model-scale wind turbine: experimental measurements versus Large Eddy Simulation with actuator line

Emmanuel Gillyns, Sophia Buckingham, Jeroen van Beeck, and Grégoire Winckelmans

Abstract. Accurate modeling of wind turbine wakes is essential for understanding turbine performance, wake interactions, and structural loading in wind energy applications. This work presents both numerical and experimental investigations aimed at evaluating the strengths and limitations of the Actuator Line Method (ALM) for wind turbine simulations in a controlled environment. Wind tunnel experiments were conducted using a purpose-built small-scale turbine TWIST (Turbine for Wind-tunnel Investigation & Scaled Testing) to provide high-resolution measurements of near wake velocity and blade deflection. In parallel, a numerical study was carried out using an ALM model developed with a force distribution designed to avoid averaging along the blade length in order to better capture variations in the near tip and hub regions. The inflow conditions in the simulations were carefully matched to the experimental profiles of mean velocity and turbulence intensity, ensuring that discrepancies would be mostly attributed to the modeling rather than to differences in inflow representation. Comparative analyzes between simulations and experiments were performed to assess the model’s ability to capture key near-wake features. Additionally, the influence of blade pitch angle on wake development was explored numerically. The blade deflection was analyzed both numerically and experimentally to evaluate the coupling between aerodynamic loading and structural response. The results show that the ALM, combined with distributing the hub drag over a Gaussian ellipsoid, is able to reproduce the main characteristics of the wake at 1.4 rotor diameter downstream, including the velocity deficit and the sharp transition between the wake and the free stream. Wake results at 4.35 diameter are also presented and are in line with the experiment, except very near the ground. Overall, the combined experimental and numerical approach provides valuable insight into the predictive capability of the ALM and its applicability for wind turbine modeling in controlled conditions.

Received: 05 Nov 2025 – Discussion started: 02 Dec 2025

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 paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.

Download & links

Emmanuel Gillyns, Sophia Buckingham, Jeroen van Beeck, and Grégoire Winckelmans

Status: final response (author comments only)

RC1:
'Comment on wes-2025-242', Anonymous Referee #1, 02 Jan 2026
Investigating wake reproduction of a model-scale wind turbine: Experimental measurements versus Large Eddy Simulation with actuator line.
General comments:
The authors present a LES framework (Nek5000) for a model-scale wind turbine in an atmospheric boundary-layer wind tunnel, with two main methodological components: (i) a recycling–inflow strategy augmented by a controller-based volume forcing, and (ii) an extended actuator line method (ALM) that represents blades, tower, and nacelle. The manuscript is well written, and the documentation of the experimental setup, LES validation, and uncertainty assessment is valuable for reproducibility. The authors also provide quantitative validation of wake profiles at two downstream locations. Overall, the work is technically competent and scientifically sound. However, several central claims require stronger supporting evidence before the work can be recommended for publication.

Specific comments
The manuscript attributes improved near-wake gradients and enhanced tip-vortex representation to specific implementation choices, including local force evaluation at every grid point, avoiding radial smoothing, and disabling the optional tip-spreading correction. These decisions are physically defensible, but the current presentation does not provide a baseline comparison that isolates their incremental contribution. To substantiate the novelty and quantify the benefit, the paper should include a controlled ablation study (e.g., standard ALM versus the proposed extended ALM), reporting the impact on clearly defined wake metrics (velocity deficit, shear gradients, turbulence intensity, and spectra). Without such comparisons, the improvements remain qualitative and cannot be unambiguously attributed to the proposed modifications.

The nacelle drag coefficient is explicitly adjusted empirically until the wake appears to match observations. This compromises the independence of validation, especially for near-wake structure where nacelle effects are non-negligible. Please discuss a sensitivity study showing impact of nacelle Cd on wake metrics.

The controller-based forcing reduces TI MAPE below 0.75 m and argues that imposed perturbations have dissipated by the recycling plane. Matching TI does not guarantee matching spectra, integral length scales, anisotropy, or coherence, all of which affect wake recovery/meandering. Since downstream errors increase, we cannot determine whether residual discrepancies are dominated by inflow-structure mismatch or turbine forcing/model assumptions. Please address this point thoroughly, with clear justification and supporting evidence.

The manuscript compares +1° and −3° pitch cases and notes different agreements near tip vs root. If the nacelle Cd is tuned and the root region is known to be problematic for ALM, then pitch-induced differences may be difficult to isolate. Please quantify the pitch sensitivity to demonstrate the extent to which the reported error is dominated by a region where the model assumptions are known to be invalid.

Blade deformation mismatch should be connected to wake interpretation more directly. Since deformation affects angle of attack and thus loads, the wake validation may be partially right for the wrong reasons. Please elaborate further on this point.

Technical corrections
Replace “looses its coherence” with “loses its coherence.”

Figure annotations contain spelling errors: “Settling chanmber” → “Settling chamber”; “Font turntable” → “Front turntable”; “Substract” → “Subtract.”

Correct “balde mean deformation” → “blade mean deformation.”

Standardize unit formatting throughout (e.g., “ms−1” vs “m s−1”) and ensure consistent usage of D (rotor diameter) vs “ø” notation in figure labels/captions.
Citation: https://doi.org/10.5194/wes-2025-242-RC1
RC2:
'Comment on wes-2025-242', Anonymous Referee #2, 02 Feb 2026
The manuscripts presents a combined experimental and numerical approach to the description of a wind turbine wake. In particular, experiments performed with ta turbulent boundary layer are used to validate an actuator line model of the same turbine. The results are promising in terms of overall agreement between the two approaches. The use of a high-order spectral method to solve the N-S equations coupled with an actuator line model is an interesting prospect. However, the manuscript lacks crucial details that significantly hamper it’s long term impact. Some comments below:
The main concern is the scarce amount of details provided regarding the numerical setup and the corresponding lack of any independence study. This is a crucial aspect as a CFD code that uses high-order spectral decomposition rather than “traditional” finite difference is used. “standard” best practices for actuator line models require 80-120 spanwise elements per blade. The kernel size is typically refined based on how many elements are used, with finer grids allowing for smaller and more realistic projection kernels. This ammount of discretizaion may still not be enough to correctly resolve the tip vortices (see: https://doi.org/10.1016/j.compfluid.2024.106477). Based on the information provided in section 3.1 a lot less elements were used to discretize the rotor area. While not directly comparable, authors performing ALM LES on scaled models using finite difference used orders of magnitude more cells (https://iopscience.iop.org/article/10.1088/1742-6596/2767/5/052050, https://doi.org/10.5194/wes-10-1707-2025). Especially because this is, in my understanding, the first time this ALM setup is presented, the quality of the results is hard to judge without grid independence and a comparison to the established literature, even if a different method to solve the N-S equations is used.

A second major issue is the lack of a comparison between experiment and simulations in total thrust or power. Similar wake characteristics can arise because of different levels of power extraction.

The scope of the manuscript does not appear to be crystal clear. If the focus is on validation why does the pitch angle in simulations not match the one in experiments? On the other hand if the scope is to study the influence of pitch angle, the gap that the manuscript is trying to fill should be better motivated. Moreover, experiments with a full rotor model and ABL are not easy to come across. This could be a major selling point for the study if better explained.

No details regarding how the airfoil coefficients used in the numerical model are derived are provided beyond Figure 17. Crucial details are missing such as the Reynolds number and the Ncrit that was used in XFoil. The NCrit value influences where transition occurs, which in turn can influence the airfoils performance significantly at low Reynolds numbers. I would expect the Ncrit value to be tuned according to the TI level in the wind tunnel

Where any steps taken to control the surface finish of the blade? This can influence BL transition and 3D printing usually generates quite rough surfaces.

No details are provided regarding the turbulence model

How long do the simulations run? What is the timestep? Statistical convergence must be accounted for in scale-resolving simulations

The ALM description also lacks crucial information. For instance, what strategy was used to sample the velocity in the flowfield? How many actuator points are used? What is the actual width of the kernel compared to the grid (eq. 10).

Are all experiments performed with 0° of blade pitch? Why compare to +1° and -3° in numerical simulations? Moreover, there seems to be higher thrust force extraction at the rotor edges, which may suggest inaccuracies in the tip-loss modelling.

Figure 23: Are the forces time-averaged? Why is a certain instant considered over another? Moreover, profiles are much smoother at the tip, what is this caused by? We are also not seeing the classical decline in axial forces close to the tip. This could be due to inaccuracies in tip effect modelling, or due to the large chord near the blade tip in the model, which does not feature taper as typical blades would.
Citation: https://doi.org/10.5194/wes-2025-242-RC2
AC1: 'Comment on wes-2025-242: final response', Emmanuel Gillyns, 27 May 2026

We thank both anonymous reviewers for their valuable comments and constructive suggestions, which helped us improve and clarify the manuscript. We believe these revisions have strengthened the quality of our work. Please find attached our responses to the reviewers’ comments.
Emmanuel Gillyns, on behalf of all co-authors.

Citation: https://doi.org/10.5194/wes-2025-242-AC1

Emmanuel Gillyns, Sophia Buckingham, Jeroen van Beeck, and Grégoire Winckelmans

Viewed

Total article views: 759 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
415	312	32	759	36	33

HTML: 415
PDF: 312
XML: 32
Total: 759
BibTeX: 36
EndNote: 33

Views and downloads (calculated since 02 Dec 2025)

Month	HTML	PDF	XML	Total
Dec 2025	153	83	11	247
Jan 2026	105	49	6	160
Feb 2026	50	25	6	81
Mar 2026	29	49	3	81
Apr 2026	53	71	3	127
May 2026	25	35	3	63

Cumulative views and downloads (calculated since 02 Dec 2025)

Month	HTML	PDF	XML	Total
Dec 2025	153	83	11	247
Jan 2026	105	49	6	160
Feb 2026	50	25	6	81
Mar 2026	29	49	3	81
Apr 2026	53	71	3	127
May 2026	25	35	3	63

Viewed (geographical distribution)

Total article views: 720 (including HTML, PDF, and XML) Thereof 720 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 30 May 2026

Short summary

This study evaluates the Actuator Line Method (ALM) for simulating wind turbine wakes by comparing it with wind tunnel tests on a small-scale turbine (TWIST). ALM accurately captured key wake features like velocity deficit in the wake and sharp transition with the undisturbed flow. Additionally, the deformation of the blades is also evaluated. The results confirming ALM as a reliable tool for aerodynamic and structural analysis in wind energy.


Total:	0
HTML:	0
PDF:	0
XML:	0