The paper by Sebastiani, Bleeg and Peña addresses the influence of wind farm blockage on power performance measurements. They demonstrate through CFD simulations that met mast and nacelle lidar measurements of turbine power curves in a wind farm are affected in a material way by the blockage from the tested turbine and from the other turbines in the array. This introduces a bias in the measured power curve relative to the case of an isolated turbine. The authors propose and test a correction methodology, which can remove the effect of the wind farm and adjust the measured power curve to what would have been measured, if the turbine was operating in isolation.

They further introduce a second correction step, which removes the influence of the turbine’s induction zone on the wind speed measurement. The result is a power curve where the power is a function of the freestream wind speed infinitely far from the turbine.  

It is also demonstrated that the procedure can give meaningful power curve measurements in wakes.

Both corrections are in principle to be determined by numerical simulations, but the authors show through simulations of nacelle lidar measurements that measuring close to the rotor, 0.5 rotor diameters (D) upstream, can eliminate the need for a wind farm simulation. The remaining element of step 1 of the correction is the ratio of the wind speeds 2D upstream and at the rotor disk (or 0.5 D upstream if measured) for the turbine operating in isolation. This can be modelled, but the authors also suggest measuring this at the test site where the type certification is done using the same nacelle lidar setup to be used in the power curve verification at the wind farm.

Since power curve measurements do not generally find an underperformance relative to the manufacturer provided power curve, I would argue that the manufacturer's power curve is not specified at infinity, but at the distance range of 2-4 rotor diameters specified by the IEC standard, most likely 2.5D which is the recommended upstream distance in the standard. Likely, the power curves issued by manufacturers are generated to agree well with measurements taken at 2.5D.

I would therefore argue that correction step 1 (Eq. 1) is the only one we need to worry about for a power curve verification test (PCV). This test is meant to validate the power curve provided by the manufacturer and is typically done based on wind speed measurements at 2.5D.

As the authors show, these measurements are affected by the wind farm blockage. This will lead to a bias if this effect is ignored. Since the wind blockage tends to reduce the wind speed, this bias will make the power curve appear to be over-performing. Since the PCV is a contractual procedure linked to a performance warranty, this represents a commercial challenge.   

The second correction step (Eq. 2) is necessary for EYA calculation where the freestream wind speed is the known characterisation of the wind resource. For this a version of the power curve which depends on the freestream wind speed is needed. But it is not needed for PCV if the manufacturer's PC is given at 2.5D for a turbine in isolation, as I suspect. Rather, with this interpretation the second step is done not as part of a measurement campaign, but as a component of the EYA modelling and hence it can be accomplished using modelling tools.

I do not expect the authors to completely agree with the above, but I think there is sufficient uncertainty between these two views of what the manufacturer’s power curve represents and hence the necessary steps needed to correct it that I wish for them to include a discussion of this in the paper. This is central to the corrections they devise and test in their work.

As a general remark why are the simulated measurements in the paper at a distance of 2D? The standard IEC recommends a distance of 2.5D and consequently most PCV campaigns use this distance to my knowledge.

Finally, I would like for the authors to comment on the use of engineering models for the single turbine and wind farm blockage as alternatives to the more expensive, but presumably more accurate CFD model. Have they tested a simpler model and compared the impact of blockage on the power curve with that from CFD?

Beyond the general comments above I have these specific suggestions:

• 23 – maybe add “and warranted” to theoretical
• L 27 – strictly speaking the freestream wind speed is measurable, just not concurrently with the turbine power output
• L 28 – consider changing “is expected” to “has traditionally been expected”
• L 38 – also cite https://web.archive.org/web/20140512231016/http://www.sgurrenergy.com/wp-content/uploads/2012/12/Compression-zone-technical-paper-A4.pdf
• L 50 – also cite https://iopscience.iop.org/article/10.1088/1742-6596/2265/2/022001 and discuss how and why the results in that paper differ from those in Sebastiani 2022
• L 53 – also cite https://wes.copernicus.org/articles/6/521/2021/ and https://wes.copernicus.org/preprints/wes-2023-37/wes-2023-37.pdf
• L 65 - suggest changing “nacelle lidar measurements” to “simulated nacelle lidar measurements”
• L 99 – When using the average axial velocity across the rotor as a proxy for U_disk you are implicitly neglecting variations of the wind direction across the rotor and the effect that these might have on the power performance. Maybe comment on that
• L 129 – As mentioned above I do not believe that the PC and thrust curve is provided by the manufacturer as a function of the freestream wind speed, but regardless they need to be reformulated in terms of U_disk. The correction will just be smaller
• L 144 - What is the turbine rating and hub height? Include a plot of the power curve
• L 144 - Comment on how typical these spacing are. For offshore at least 3D is a very small spacing. Is it representative to have these spacings in a wind farm with 100 WTGs? What motivated this choice?
• Figure 2 – I propose to indicate rotor upper tip, lower tip and hub height
• L 181 - The assumption of horizontal homogeneity is violated in the turbine induction zone, but you only need to assume symmetry, ie that the lidar beams measure the same wind vector. This is nearly true if shear and veer are neglected and the rotor and hence the lidar is perfectly aligned to the wind direction. Please include comments on this simplification
• L 186 – You do not simulate the lidar probe volume. Please include an estimate of the size of this effect given the axial wind speed variation in the induction zone, the probe length and the measurement distance. Based on a rough estimate, is it reasonable to neglect it?
• L 190 – You neglect the vertical and lateral wind components. How accurate is this assumption? Why is it justified?
• L 192 – “When using the 2-beam lidar focused at 0.5D, the horizontal wind speed at hub height is used as Udisk,lidar” but the wind speed at hub height varies laterally due to the blockage, so what have you actually done? Which wind speed are you using?
• L 199 – Make it clearer that the standard deviation is the std over the 9 mast locations
• L 203 - Contrast the theta=-1 results with those of the single row in Sebastiani et al (2022) and in Bleeg and Montavon. Why are the Sebastiani results opposite those presented here and in Bleeg & Montavon for the wind direction perpendicular to the row? Is it because Sebastiani et al did not include the effect of the boundary layer?
• L 210 – I suggest adding “and hence the error bars” at the end of the sentence
• L 214 – why would you expect the power to be related to the wind speed raised to the power 2-3 and not simply to the power of 3? If you are trying to be more accurate and include the aerodynamic efficiency of the rotor as captures in the C_P curve, then I suggest adding a bit more detail. As it reads now it is simply confusing to anybody expecting the U^3 relationship. The details are also not important for the argument
• L 218 – Is the reduced correlation due to the separation between the mast and the rotor in a turbulent (fluctuating) wind field?
• L 219 – What does Figure 5 show? Mean power over front row WTGs or the sum of their power or a dot for every turbine and wind direction?
• L 258 – Again, this is up for discussion
• L 285 – I think the black squares near 7.1 m/s in Figure 12 represent the 20 front row turbines. This should be made clear in the text and in the figure caption
• L 287 - In wake the wind speed is highly inhomogeneous, and the two beams will see different radial wind speeds. The lidar will interpret this in the wind field reconstruction as a misalignment of the nacelle with the wind direction since it assumes a homogeneous wind speed. Is this not relevant because you are using the horizontal velocity = sqrt(ux^2+uy^2) and not the axial wind speed?
• L 320 – “However, the approach relies on the accuracy of the flow model “. I would add that it is not desirable to mix measurements and modelling for the purpose of a contractual PCV since this makes the resulting PC sensitive to details of the modelling setup. This is makes it much harder to define the PCV test in a contract and similarly challenging to define the uncertainty of the resulting power curve
• L 356 - Typically the hub heights will differ between the test site and the wind farm. This would further challenge the re-use of (Udisk/Ulidar)^I since this will be affected by ground clearance effects on the flow
• L 369 – Through the entire paper the reference distance is 2D, why is it 2.5D here?
• L 374 – “Field observations could further clarify the validity and utility of these corrections.” Be specific about how this would be done