The paper compares different state-of-the-art methodologies for partial load control of wind turbines. The authors conclude that an advanced Tip Speed Ratio-tracking scheme cannot outperform the traditional kw2 controller in terms of power maximization. The advanced controller can, however, help to reduce loads, while reducing the bandwidth of the controller.

The paper is very well written, well structured and does not require larger revisions. Please find a few comments in the pdf attached and consider them for the revised version of this paper.

Major comments:

• This paper provides a detailed analysis of torque control tuning with a primary focus on the frequency domain.  Specific guidance for tuning the torque control is not provided or any insight into secondary effects, like drivetrain loads and the ability to avoid tower frequencies. 
• The main conclusions of the authors are already known, both controllers track the optimal TSR and the TSR tracking controller gives more flexibility; these are the design objectives of each controller.  The conclusion that higher torque control (and WSE) gains lead to tighter control, with more torque variation and slightly more power output seems obvious. I'm not sure so much analysis is needed to prove that. 
• When is the torque control bandwidth "high enough" and there are diminishing returns in the power gains?  How are drivetrain and tower loads affected?  If the reference rotor speed is changed to avoid a natural frequency, how high does the torque control bandwidth need to be?  I would have liked to learn the answer to these questions so that users of the WSE-TSR controller have specific guidance for how to tune them.
• This paper is on torque control, not general model-based wind turbine controllers as the title suggests. 
• Structural loads are mentioned but not presented anywhere.  "Torque variation as a proxy for loads" could be made more specific.  What structural loads are affected by the torque control tuning?
• For a wind energy science audience, there is a lot of mathematical jargon and nomenclature.  To appeal to a wider audience, frame your problem in wind energy terms.  Some examples are
  • around L340, where MOO doesn't need to be explained in such mathematical detail
  • In L344, "operational conditions" are described. What exactly are they?
• I would have liked to see power variance as the cost or perhaps structural loads; this data is available in the simulations.

Minor Comments

•  Abstract
  • Quantitative results would be appropriate here.
  • Does OpenFAST count as high-fidelity simulations? Probably not, so perhaps be specific.
• Intro
  • L29: k\omega^2 may be common in research articles, but probably not on actual turbines
  • L50: is this always the case or just for that rotor in that study?
• Section 2: Theory
  • L149: constant \lambda_* or single \lambda_*?
• Fig 6:
  • It's difficult to tell the black lines apart
• Are the controllers with a high bandwidth stable?  The resonance would lead me to believe they are not.  Are there stability margins associated with these controllers?  What is the significance of the resonant peak?  It seems like it would degrade performance if excited.