A very nice and complete work, pointing out relevant shortcomings in the leading implementations of DWM and putting a legitimate doubt on the use of DMW for fatigue modelling. The work is high quality, showing relevant analyses and pointing out the key observations clearly. Conclusions are backed up by results, and limitations of the study are discussed openly and fairly.  

Abstract: OK

1 Introduction:

I think the first sentences (line 17 and 18) mischaracterize the history of wind energy a little, unless “recent decades” includes the 70’s and 80’s, when the first large-scale wind parks were built in Denmark and the US [1].

2 Methodology:

Equation 1 there is something wrong with the notation. The summation from k=1 to N_k suggests N_k is the number of load ranges, but later N_k is also the number of cycles at load range S_k. Please correct the notation, e.g. to a summation from k=1 to K, and adjust the phrasing in line 127 if needed. Further, please define m as the material Wohler coefficient.

IFE model: The superposition of wake deficits is based on Zong et Al, but the summation of wake added turbulence is based on root sum squared. This is an interesting choice and could be pointed out in the text.

3 Results

In lines 485 to 490, frequency content above 1p is discussed. However, no mention is made of the 3p frequency, indicator of asymmetric rotor loading, which for case a) shows up prominently in the LES PSD, but not an any of the DWM models. This is consistent with the observation made on tower loads in lines 512 to 515. These observations are in line with previous literature on asymmetric loading in DWM [2]. A mention of 3p loading could be made in the discussion at line 720, where asymmetric flow in the wake is discussed. The addition of the PDS plots for tower yaw moment in the aligned case should provide further evidence for this phenomenon and would be a valuable addition to the work.

In line 588 we find “The reduced loads due to an increased wind speed”, should it not be due to a “decreased” wind speed? Otherwise, more clarificaiton is needed with respect to what you are comparing against here, ans which figures support this statement.

4 Discussion

In the limitations, the impact of a fully rigid structure should be discussed. Tower load spectra are normally dominated by eigen-frequencies and their harmonics. While the rigid tower allows to analyze the aerodynamic forcing in isolation, overall conclusions on tower DEL may be skewed as the interaction between tower Eigenmodes and DWM is neglected. This should be mentioned.

5 Conclusions: OK

Editorial commends

Line 146: Unnecessary comma.

Lines 182-185: Please revise this sentence.

Line 201: Please revise incomplete sentence: “necessary to model.”

Line 492: Missing a space after comma.

Line 786: No “-“ in fatigue damage.

[1] Möllerström E, Gipe P, Ottermo F. Wind power development: A historical review. Wind Engineering. 2024;49(2):499-512. doi:10.1177/0309524X241260061

[2] V Bernard et al 2024 J. Phys.: Conf. Ser. 2767 092092

This manuscript systematically compares the numerical results of three DWM frameworks developed by different institutions, benchmarking them against LES outcomes. The study considers a row of four wind turbines that are either fully aligned or partially misaligned with the inflow wind direction. Additionally, various inflow turbulence intensities are considered. The evaluation of DWM implementations is comprehensive, covering time-averaged velocity profiles, TKE profiles, wake center positions, fatigue loading, and more. The authors show that the wake characteristics predicted by all three DWM frameworks agree reasonably well with LES. However, notable discrepancies remain, which they attribute to differences in the implementation of sub-models.

Overall, the topic is highly relevant to Wind Energy Science. Intercomparisons of DWM implementations developed by different institutions are important for assessing whether the approach of DWM is robust. Moreover, the turbine configurations considered here are appropriate for testing the effectiveness of DWM, especially given its intended role in providing economical estimates of steady and unsteady turbine loads under wake conditions.

That said, I have several major concerns that should be addressed before I can recommend publication:

1. Although the overall storyline of the manuscript is clear, many sentences contain grammatical issues or awkward wording. This hinders readability and sometimes makes the intended meaning ambiguous. I have highlighted a number of these language issues in the attached document, though this list is not exhaustive. A thorough language edit is required to improve clarity and the flow.
2. I strongly recommend rewriting the “Conclusion” section (with the possible exception of the first paragraph). At present, large portions of this section read more like an “Introduction,” discussing general features of DWM-based models rather than emphasizing the key findings of the current study. As a result, the main contributions of this work are underrepresented. Moreover, it would be more informative to frame the discussion directly around the impact of individual sub-model implementations rather than linking the results primarily to the institutions. This approach would make the “Conclusion” a more self-contained section, reducing the need for readers to revisit the main body of the paper. It would also highlight insights that are of greater interest to the research community, since the effects of sub-model implementations are more scientifically important over the names of the code developers.
3. Since the authors used different aeroelastic solvers to model the structural dynamics, it is unclear whether the observed differences in fatigue load predictions can be attributed solely to the DWM models themselves (sections 3.1.5 and 3.2.5). The influence of the aeroelastic solver should be discussed more explicitly, or the authors should clarify to what extent solver-specific effects may have contributed to the reported discrepancies.
4. As stated in the final paragraph of the “Introduction,” one of the major aims of this study is to “(2) to investigate how differences in sub-modelling strategies —such as wake meandering formulations, velocity-deficit profiles, multi-wake superposition methods, and wake-added turbulence treatments—affect model performance.” However, the manuscript only compares the three DWM frameworks as complete packages. By this mean, multiple sub-models differ simultaneously, making it difficult to attribute discrepancies to specific sub-modeling choices, perhaps even raising the risk of subjective judgements. I recommend that the authors at least verify some of their claims by conducting sensitivity tests in which a single sub-model is swapped within one DWM framework. Such controlled examinations would allow the impact of individual sub-model implementations to be highlighted and quantified, making the paper’s conclusions stronger and more valuable for guiding future development of DWM-based frameworks.
5. At this moment, I disagree with the explanation given in lines 335–345. The authors claim that a steeper time-averaged streamwise velocity gradient in the lateral/vertical direction, combined with wake meandering, results in elevated streamwise velocity fluctuations (sigma_u). In my view, this reasoning is flawed. I think that sigma_u would be affected by the instantaneous velocity gradient in combination with wake meandering, not by the time-averaged gradient. In fact, frequent wake meandering events would tend to smooth out the time-averaged velocity gradient, while still increasing the magnitude of fluctuations. This seems contradictory to the explanation provided in the manuscript, and the authors should clarify or reconsider this interpretation.

I have some other minor comments:

1. I recommend adding grid lines to the figures, particularly those showing velocity and sigma_u profiles, to improve readability and make it easier for readers to interpret the data.
2. I suggest that the authors include a few sets of instantaneous velocity contour plots, either in the main text or in an appendix. To be specific, I am especially interested in the streamwise velocity slice of y-plane at hub height. Since the inflow conditions for the DWM and LES simulations have been conditioned to be the same (with the exception of DWM_DTU), such a comparison would be meaningful. It could also qualitatively illustrate the distinct features of the different models in a way that complements the statistical results.
3. I recommend, though perhaps subjectively, the authors to improve the readability and aesthetics of the figures.

There are also some minor notes in the attachment.