Review: CFD-based curved tip shape design for wind turbine blades – Madsen, HA et al.

Opening Comments:

This manuscript presents an optimization framework for tip designs in wind turbines using CFD. The authors are attempting to come up with a load neutral tip shape for wind turbines using a direct high-fidelity CFD based approach. Only aerodynamic effects are considered. The authors begin with an extensive literature review that places the current work in context with archival literature. They divide recently published literature into parametric studies and optimization studies. They point to the need for higher model fidelity as cited by several previous studies, as well as the need for optimizations with unifying design optimization problems. These are the drivers for this study.

The optimization method involves a straightforward process of determining design variables, calculating a deformed surface mesh, subsequently calculating a flowfield, functions of interest and constraints using EllipSys3D. This cycle is repeated until an optimized shape is obtained. The IEA 10MW is used as the baseline, and the design optimization problem is performed on this turbine. The results are presented in terms of optimal step size, and finally the optimized tip shape compared to the baseline. The authors conclude by pointing out that mesh deformation and setting up the finite difference method carefully are critical in a CFD based approach, but ease of use may mean surrogate based approaches may be more viable currently.

There are a few specific concerns with methods and results that have been outlined below:

Specific concerns:

1. The complex step method is used to compute reference gradients. However, it is also mentioned that there is a lack of robustness in the authors’ implementation of the complex step method. This statement will need to be clarified to build confidence on the accuracy of reference gradient and the results presented here.
2. The variables in the design optimization process are not clearly explained. I assumed θ_1,2,3,4 and c_1,2,3,4 are twist and chord for four points along the span in the tip region (outer 10%), but this is not stated clearly and locations not given.
3. It is not clear how c is scaled. Is it with respect to chord at 0.9r/R?
4. Recommend using the term merit function when discussing design optimization problem for consistency, rather than just in the results.
5. An iso-view of the blade in comparison to Zahle (2018) may be more informative than current tile 3 of figure 15.

Technical concerns:

While this manuscript has been written well for most part, there are areas where typing errors have crept in or word choices cause confusion. A few examples: line 75 (also), line 241 (I’m not sure unimodal is the right word here), line 309, line 348, line 479, line 629, line 819 and so on. Recommend careful proofreading once again and rewording the unclear sentences.

Closing comments:

In general, this is a well-written and researched paper and will advance the literature on CFD based design approaches for blade tip devices. The current study can be considered a companion study to the ones by Zahle et al. (2018) and Barlas et al. (2021) where the authors also tackle the problem of load neutral blade tip extensions.  The paper addresses questions relevant to the scope of WES and presents a framework for using a CFD based design optimization process. I recommend that with minor revisions (addressing the concerns listed above), the paper be accepted for publication in WES.

The authors presented a very interesting and valuable study on optimization of WT blade tip. Particularly, it seems that the main contribution of the research is on proposing a novel methodology which can be implemented in the process of optimization. Through employing that process, 12 design variables which is a considerable number of variables in comparison with previous investigations on blade tip shape, have been taken into account. Their results show that for the optimized geometry of the blade tip, the power output of the turbine has been increases 1.2% while there is no excessive bending moment at the tip location, i.e. top 10% of the blade length. The manuscript is very well written and structured. The authors have provided a very comprehensive literature review with regard to the corresponding research area. However, there are few issues need to be resolved and clarified to improve the article.  The comments are summarized as follows:

• Since the literature review is extensive and it occupies a big portion of the manuscript, the reader might be confused about the novelty of the paper at the end of “literature review” section. I would suggest to re-state the novelty and contribution of the study at the end of this section.
• At page 18, line 535, it is mentioned that the flow over blade has been considered to be fully turbulent (which is true!). It would be more informative to include the physical justification behind that assumption.
• It is indicated in the manuscript that steady-state flow modeling has adopted to solve the equations. However, as you confirm, tip vortices are unsteady phenomenon in nature and thus steady simulations might affect significantly the results. How do you justify this issue? Is there any other previous investigation that clearly addresses that effect?
• In figure 3, it would beneficial for the reader to see the boundary conditions in the figure where the domain is displayed.
• The operating conditions of the given turbine has not been clearly presented. For instance, it would be great to include the power curve of the turbine (Cp ~ TSR), so the reader can identify that the rotational speed at which your simulations are performed, is lower than the optimum TSR or higher. TSR as a governing parameter of the fluid flow around the turbine, significantly influence the flow structures at the blade location since it determines the angle of attacks experienced by the blade at different sections.
• Finally, although the focus of this study is on the methodology and its effectiveness, it would be crucial to validate the numerical results against any available data. Particularly, because the authors concluded about the correctness of the results obtained from the optimization process, i.e. 1.2% increase in power output. Since the simulations are not performed in unsteady-state mode and the results have not also been validated, the increase in the power obtained from the optimization process might not be reliable.