Thank you for writing this article in a clear and well organized manner.  I really liked how it was organized into sections exploring different parameters of the experiment.  With that there are some minor points of confusion that I think could be cleared up to improve the article:

Page 3 line 75: you refer to "the square root of the sum-of-squares" method with no citation.  If it is the same as (Crespo et al., 1996), then I would suggest moving that citation to the end of the sentence.  If it is from something else, then a citation should be added.

Page 5 line 106: the bounds of 0° to +30° are considered for yaw optimization with no mention of negative yaw angles.  Some explanation should be given as to why negative yaw angles are not considered in this article.

Page 5 lines 107 & 113: The word "greedy" is used to describe the Boolean method.  This is confusing to me because the word "greedy" has been traditionally used to mean "business-as-usual" in wind plant optimization.  I would consider sticking to that convention and use a different word here to describe the Boolean method.

Page 6 line 135 & figure 2 title: you mention “we optimized turbine rows…”, which seems contradictory to the figure title “Turbines in-line”.  I think what you are trying to say on line 135 is “we optimized an individual row of turbines varying from 10 – 50…”.  Please consider changing this sentence to avoid confusion.

Page 8 paragraph starting at line 165: It is probably worth mentioning here how often a re-optimization of yaw angles needs to occur for a production wind plant during operation.  This would further make the case for a need to have a computationally efficient algorithm as the optimization time would really start to add up with re-optimizations for the continuous method when compared to the Boolean method.

Page 9 lines 196 & 197: you refer to a wind plant arranged in a grid and contrast this against an in-line arrangement.  To me a grid is several sets of in-line turbines, so they don’t really contrast one another.  Something needs to be done to clear this up.  I do agree that an in-line arrangement does contrast a random arrangement.  Also “in line” here should be changed to “in-line”.

Section 4.2: It was surprising to me that the 285°/345° and 300°/330° pairs of wind directions do not give identical results as the pairs are mirrors of one another about 315° for a regular square grid arrangement.  Perhaps this is because only 0° to 30° optimized yaw angles are considered.  Some mention should be made about this in this section as to why the results are not identical.  Additionally, if it is because of only optimizing positive yaw angles, then it should be reiterated here why negative yaw angles were not considered in this article.

Page 14 line 253: Here it is stated that seven random wind plant layouts were generated.  Why not more? Why not less?  It should be mentioned why seven was chosen for the results presented in this section.

Page 15 figure 10: A thought I had after seeing this figure is that it would be great if a scatter plot was included in this paper showing “computational time ratio” vs. “optimized power ratio” from the two right plots in figure 10.  This figure could show an aggregate of results from the scenarios across the article.  I think that this scatter plot would greatly add to this paper by showing how much power improvement you get at the cost of how much additional computational time, which is one of the main points that this article is trying to make.

Again, overall, nice job!

General comments

The manuscript contains a description and numerical assessment of an algorithm for yaw angle optimization for wind plant wake steering. The authors claim to propose a new Boolean optimization method based on a greedy approach. The algorithm is not accompanied by any theoretical considerations regarding convergence and the ability of the algorithm to find feasible or optimal points. Hence, it should be classified and referred to as a heuristic and not as a method. Secondly, given the vast literature on heuristics for nonlinear 0-1 optimization problems it is surprising that the authors (i) do not include any references on that topic, and (ii) choose such a trivial algorithm and do not investigate other alternatives.

Based on the literature review and the simplicity of the problem formulation (e.g. lack of constraints) one can wonder if the considered approach reaches state-of-the-art in the field. It is of course sometimes relevant to use simplified models, but one of the main arguments in the introduction is to achieve a realism. The authors should clarify.

Several of the numerical experiments are based on artificial wind farms or wind scenarios that are unlikely to appear in a real world application, e.g. random turbine positions and turbines perfectly in-line with the wind and a single wind speed. Even if this is good for reproduction of the results, it is dubious if these examples can be used to make general conclusions that are applicable to real world wind farms. The authors are encouraged to extend the numerical experiments with these comments in mind.

One of the main conclusions from the manuscript is that the proposed greedy algorithm is much faster than a traditional method, in this case SQP applied to a continuous version of the problem. There are several issues with this conclusion. Firstly, the two algorithms attempt to solve different problems and meet different requirements and are as such not comparable in a fair way, particularly when it comes to computational effort. Secondly, the implementation of the call to the SQP method is based on finite difference approximations of the gradient of the objective function. This is well known to have potentially very large implications on the computational time per iteration, the robustness of the algorithm, the achieved accuracy, and the number of iterations when the number of variables increase. How much the implementation choices affect the outcome is not reported. It is therefore possible that the differences in computational time are entirely attributed to the implementation rather than the method itself.

Specific comments

The authors claim to propose a Boolean formulation. This formulation is not formally stated in the manuscript. The optimization problem that is stated is in fact the continuous problem on line 90. Please clarify.

The authors argue that the proposed optimization formulation is novel. However, the literature review mentions that others have considered several discrete choices of angles. It seems that the Boolean approach would be a special case. Please clarify.

The statement on page 8 line 167 “… time is seen to increase exponentially with increasing design variables.” is not properly motivated and most likely not correct. It is much more likely that the increase in time is polynomial given the type of problems and the method employed. The authors should confirm that the increase is indeed exponential or revise/remove the statement.

Thank you for the interesting article. I think it is well written and the topic is clearly explained.

Some minor suggested points could improve it:

Line 22: 'models' to be substituted by 'experiments' (models can be also numerical, like yours)

Line 71: written that FLORIS has been used for wind plant layout optimization research. Add references.

Lines 129 to 131: I think the sentence "For all of the results in this section, the freestream wind speed was set at 10 m/s, which is below the rated wind speed of the wind turbine we used. Past studies have shown wake steering to be most effective below the rated wind speed."  belongs to ch. 2, so I would move it to the end of this chapter.

Two comments on the same sentence: 1. Explicit that the freestream wind is assumed uniform; 2. 'Past studies', give the references.

 Line 142: the performances decrease after a pick. Please comment why, in the text.

Two general comments on the multiple figures: 1. I think it would be better to add a), b), c)...on the subfigures, making it easier to refer to them in the text (instead of 'top left subfigure' it becomes 'Fig. 3a'). That might not be needed for all the multiple figures, but only those subfigures you need to refer to in the text (e.g. Fig. 5 is ok as it is). 2. I think the title at the top of the figure is distracting (ex. Fig. 4 '50 turbines in-line: varied turbine spacing'); if needed move it to the caption.

Overall, good job!