This paper brings together several elements related to the aerodynamic predictions for kites in airborne wind energy systems, using results and methods from multiple related publications.

* The title is too broad, as the parametrization of the airfoil section seems to apply only/mostly to leading-edge inflatable kites, not to ram-air kites for instance, so the title should mention this limitation.

* The title is somewhat misleading, as the "soft" aspect is not included in this paper, there is not aero-elasticity. This is also present in the manuscript and in the conclusion.

* The introduction with the step-wise build up and history of the lifting line method is interesting and suitable for a PhD thesis, but does not support the key message of this paper and is not required to understand what is being done in the presented work.

* The machine learning model is now built separately for each Re, which is strange, as Re will constantly be varying during operation so also in realistic simulations. This should be changed by making a combined model, probably after gathering new data, or mentioned clearly as a significant limitation of the applicability.

* The PCHIP interpolation for the angle of attack is a strong limitation, because that means a user or application should always extract the data from the model at the sampling points and then fit them before actually using the data. Smooth and physical behaviour should be built into the machine learning model by construction, e.g. by having coefficients of a polynomial for the data as output instead of the actual data.

* The actual contribution and goal of this paper are drowning in too much other information that is interesting as such, but not contributing to the message.

* Line 29: Give a range for the lift-to-weight ratios.

* Line 35: "But" is a coordinating conjunction and should only be used at the beginning of sentences to signal a shift in thought.

* Line 69: Kutta (1902) => (Kutta, 1902)

* Line 182: Fritz (2024) => (Fritz, 2024)

* Line 226: Introduce abbreviation AIC here.

* Line 275: Motivated => and which is motivated

* Line 283: w_i should be introduced after equation (19)

* Line 328: Is this at 20% starting from the TE? Or 80% in the current formulation?

* Line 334: The sizes that were coming out of the sensitivity analysis should be summarized here.

* Line 338-9: It seems contradicting that one chord length is used in the z-direction and at the same time unit depth. Furthermore, it is not clear what the physical distance (in multiple of chord lengths) covered by the 201 cell layers is.

* Line 346: Mention the remaining error

* Line 353-5: It is strange to mention documentation of 2025 for an older OpenFOAM version

* Line 393: The importance of the variables could be quantified using Sobol indices

* Line 411: What did the flow field in the outliers look like? How come there is a strong variation in their results despite all being converged?

* Line 464-5: This fits better with line 445

* Line 482: The counterclockwise only applies on the pressure side (but that is also where the recirculation occurs here)

* Line 487: encounters both three => encounters three

* Line 580: V => U_\infty

* Line 587: The last sentence is repetition from the line above.

* Line 639: Missing point at end

* Line 711: < 0 => > 0

* Figure 17: Indicate also unstable regions due to other criteria, e.g. in the right subplot with a second color

320: Are there any physical considerations for the seam angle and the dynamic search? Any physical insights on the effects of the curvature distribution on the flow field?

335: I would replace "computational cost" with something like "the effort required for case set-up" as computational cost generally directly refers to the solution cost.

335: I would also refer to the number of elements in the mesh.

340: Shouldn't the number "eight" be written as 8 since it is a parameter? Same with "five".

360: I would reword "vector inlet velocity" as "inlet velocity vector"

375: "converge to 1e-4" would suffice without saying "within"

380: What does "initial convergence" vs. "final value" mean here? The point about rapid convergence is not clear.

405: Are these cases generated automatically within the design space? What is the computational resource requirement for each case based on the grid size?

415: "..smaller than the tube radius"

480: "A corner eddy closer to the..". I would reorganize this sentence.

485: "Increasing alpha from 2 to 8 leads to a more concentrated and..."

485: "One encounters three pressure side surfaces". You do not need to say "both" here.

585: Are the results sensitive to the perturbation step size?

 

Overall, there are some sentences that I would rephrase for better grammar and better conveyance of the ideas in the manuscript. Also, I see way too many commas, not all of which are necessary. Cleaning up the writing should improve the quality of the manuscript.