Rotor Loading Characteristics of a Full-Scale Tidal Turbine
Round 1
Reviewer 1 Report
The article reported the measured rotor loads on a 400 kW tidal turbine in real sea conditions. Important findings are obtained. On one side, results obtained during ebb tidal conditions were found to agree well with theoretical predictions rotor loading. On the other side, the measurements during flood were lower than expected. These findings indicates the necessity to quantify the characteristics of the turbulent flows at sea sites during the entire tidal cycle to ensure the long-term integrity of the tidal turbines to be deployed. I suggest to accept the paper for publish on Energies after the following minor corrections be done.
In page 2, line 59, it could be better to review the tidal turbine experimental research in detail with the focus on mechanical loading.
In page 6, line 160, it could be better to include some schematic diagrams to show where every gauge and sensor is located on the turbine.
In page 6, line 168, for the pchip extrapolation method, is it Piecewise Cubic Hermite Interpolating Polynomial (PCHIP)? If it is, it is an interpolation procedure, the dash curves in Figure 5 should go through every point they used? Why it isn’t like this in Figure 5?
In page 7, Figure 5, why at radial positions about 0.7 m, the measurements seems far away from other measurements?
In page 13, line 339, for the Blade Element Momentum (BEM) based model, what conditions are you used for the tidal turbine? For example, what’s the input for the BEM model?
In page 14, line 357, why fewer measurements were obtained during flood conditions? Is there any methods which can be used to avoid this?
It could be better to insert anther figure to show smoothed time series which reduce the high frequency noise of the time series in Figure13.
Author Response
Reviewer #1
The article reported the measured rotor loads on a 400 kW tidal turbine in real sea conditions. Important findings are obtained. On one side, results obtained during ebb tidal conditions were found to agree well with theoretical predictions rotor loading. On the other side, the measurements during flood were lower than expected. These findings indicate the necessity to quantify the characteristics of the turbulent flows at sea sites during the entire tidal cycle to ensure the long-term integrity of the tidal turbines to be deployed. I suggest to accept the paper for publish on Energies after the following minor corrections be done.
The authors thank the reviewer for his/her comments that have aided to improve the quality and clarity of our manuscript. Below we address the questions raised.
In page 2, line 59, it could be better to review the tidal turbine experimental research in detail with the focus on mechanical loading.
More detailed review of previous experimental campaigns has been added in lines 60 – 68:” Mycek et al. [3] studied the impact of different turbulent flow conditions in the loadings and wake of a single turbine. Blackmore et al. [5] quantified the effect of various turbulent intensity and length scale conditions on the thrust and bending moment on a single turbine and revealed that larger turbulence length scales increase load oscillations. Milne et al. [6] studied the variation of bending moment oscillations inducing single periodic loadings on a single turbine with various amplitudes and found that dynamic stall triggered load unsteadiness. Payne et al. [11] investigated the spectral distribution of blade loadings on a single turbine operating under low and high turbulent conditions, finding that ambient turbulence largely affects the loads spectra as blade passing frequency-related phenomena smoothed.”
In page 6, line 160, it could be better to include some schematic diagrams to show where every gauge and sensor is located on the turbine.
The reviewer is correct as this description needs further details. Unfortunately, no pictures were taken from these on the turbine. Alternatively, we have added the following in line 168: “the gauges were located at mid chord length”. We hope this is enough and no diagram is further needed.
In page 6, line 168, for the pchip extrapolation method, is it Piecewise Cubic Hermite Interpolating Polynomial (PCHIP)? If it is, it is an interpolation procedure, the dash curves in Figure 5 should go through every point they used? Why it isn’t like this in Figure 5?
The full description of pchiphas been included in the manuscript for clarity. The curves do not go through every point because some of the measurements were discarded from the curve fitting process, or had weights applied in the curve fit. It was necessary to do this because some of the measurements were consistently spurious, e.g. at radial position 1.7 m. This is explained in the text, while the original measured values have been kept in the Figure 5 for justification.
In page 7, Figure 5, why at radial positions about 0.7 m, the measurements seems far away from other measurements?
We understand that the reviewer is referencing to the measurements at radial position 1.7 m. If so, see the response to the previous comment for an explanation.
In page 13, line 339, for the Blade Element Momentum (BEM) based model, what conditions are you used for the tidal turbine? For example, what’s the input for the BEM model?
The BEM simulations were performed under steady conditions (constant and uniform flow field with a fixed turbine rotational speed). This has been clarified in lines 366 - 367.
In page 14, line 357, why fewer measurements were obtained during flood conditions? Is there any methods which can be used to avoid this?
Unfortunately, there are fewer measurements during flood conditions because these tests were performed just once in each tide. During the flood test, the turbine operator also started the test later in the tidal cycle, yielding fewer measurements than during ebb. At all other times, the turbine was controlled to operate in its normal condition. The authors did attempt to reduce the flood averaging period to get more data points on the curves, but this did not increase the range of the results. Hence the ebb and flood averaging periods were kept the same for consistency.
It could be better to insert another figure to show smoothed time series which reduce the high frequency noise of the time series in Figure13.
We thank the reviewer for this suggestion. However, Figure 13 as is aids to visualise the high- and low-frequency oscillation in the forces which is intrinsic of turbulent tidal flows. We consider that adding a new figure might make the discussion of these results quite dense, shading the main outcomes.
Reviewer 2 Report
This paper presents a turbine inflow and loads dataset from a tidal turbine deployed offshore of Wales, UK. The paper is well written, well organized, and the content is valuable to the global tidal energy industry. I suggest the paper be published after minor revisions to address the following comments:
- Please explain how the method for estimating blade loads is similar to or different from standard practices for calculating this variable. Line 104 states that "A methodology was developed to infer blade root bending moments from radial measurements". Does this mean no one has ever done this before? Are there not accepted approaches for doing this?
- Lines 102-104: presumably you don't have 'rotor rotation position' data, because then you would know that blade 2 was 'pointing up' in this case? Is this kind of observed difference in bending moments typical? In other words, do the 1m moments on a single blade vary from 30 to 70 kN.m every rev? If so, I suggest revising this text to state that these differences are normal. Right now it reads as though the blade 2 moments are abnormal.
- Figure 2 is not what I typically call a "tidal ellipse". The IEC -201 standard calls these 'joint velocity and direction probability distributions', but I've also heard them called "current rose distributions".
- Figure 7: Turbulence intensity is often considered 'invalid' below some threshold mean velocity (e.g., 0.5 m/s). In these cases the turbulence isn't energetic enough to be worthy of investigation, TI is only high because you're dividing by a small U. It may be worthwhile to select such a velocity and indicate the time periods where the flow is above that speed in both panels of Figure 7.
- Figure 8: I suggest removing the TI profiles (panels c and d) for slack conditions. They are not particularly meaningful, and are somewhat misleading (i.e., somewhat might interpret this as "wow the turbulence is extremely high at this site").
- Figure 8: There seems to be two layers in the flow: a high-shear bottom boundary layer, and a relatively constant speed upper layer. During ebb the BBL is ~15-20m thick, during flood it is ~12m thick. Do you have any hypothesis as to why this would be? Is it bottom roughness? Is there stratification? Other ideas? Also, does this suggest there may be significant advantages to positioning the turbine higher in the water column?
- Lines 202-307: More details on how spectra were calculated are required. What was the length of your spectral window? Was a window function applied? How many ensembles went into each spectral plot? What is the significance of the -1 slope, and the -5/3 slope (add references)? Any hypothesis for the spectral peak at 10^-1 Hz in the Ebb spectra?
- Section 4.3: This is very interesting and well presented. I think this is the most important piece of this paper. On line 351 you state that "further refinement is required to improve the derivation of Fxroot". Can you provide details on what kinds of 'refinement' might be effective?
- Figure 14: Again a clearer description of how spectra were calculated is needed. Are the spectra calculated identical to Figure 9 (same window, etc.)? Furthermore, it is uncommon to see a large spike in the spectra at the highest frequencies. If it is real this suggest that high-frequency signals exist in the measured variable, and therefore that aliasing of signals above the measurement frequency could be occurring. However, I can't help but wonder if this is noise in the measurement. Are there ever time-periods when this peak does not exist? Does this peak exist in the calibration dataset? A better description of the source of this peak is needed.
Author Response
Reviewer #2
This paper presents a turbine inflow and loads dataset from a tidal turbine deployed offshore of Wales, UK. The paper is well written, well organized, and the content is valuable to the global tidal energy industry. I suggest the paper be published after minor revisions to address the following comments:
The authors are grateful to the reviewer for the time spent revising our manuscript which, after their consideration, have notably helped to improve it. In the following, the response to the comments:
- Please explain how the method for estimating blade loads is similar to or different from standard practices for calculating this variable. Line 164 states that "A methodology was developed to infer blade root bending moments from radial measurements". Does this mean no one has ever done this before? Are there not accepted approaches for doing this?
The following text was added in lines 199 - 204 with references to provide more information on measuring rotor loading: “This procedure for calculating thrust differs from that typically used in scale-model studies [17] or field testing of smaller rotors [12], where load cells are placed on the turbine support structure and a correction is applied to remove the drag on the frame, leaving only the axial force on the rotor. This involves measuring the drag on the support structure without the rotor, something that was not practical for the turbine in this study and would not provide any information on the individual blade loading and bending moments.”
- Lines 182-184: presumably you don't have 'rotor rotation position' data, because then you would know that blade 2 was 'pointing up' in this case? Is this kind of observed difference in bending moments typical? In other words, do the 1m moments on a single blade vary from 30 to 70 kN.m every rev? If so, I suggest revising this text to state that these differences are normal. Right now it reads as though the blade 2 moments are abnormal.
We agree with the reviewer that the text might lead to confusion and unfortunately there was no rotor position sensor on the turbine. The change in bending moments shown in Figure 5 agrees with the instantaneous values in Figure 13, in which at some given times blades can feature quite different values. To clarify this, we have added in lines 193 - 195: “The differences of instantaneous loads between blades are constantly observed during the turbine rotation, as it is shown later in Figure 13.”
- Figure 2 is not what I typically call a "tidal ellipse". The IEC -201 standard calls these 'joint velocity and direction probability distributions', but I've also heard them called "current rose distributions".
This has been re-labelled as “tidal rose” as also used in literature.
- Figure 7: Turbulence intensity is often considered 'invalid' below some threshold mean velocity (e.g., 0.5 m/s). In these cases the turbulence isn't energetic enough to be worthy of investigation, TI is only high because you're dividing by a small U. It may be worthwhile to select such a velocity and indicate the time periods where the flow is above that speed in both panels of Figure 7.
We agree with the reviewer that turbulence intensities below a given threshold do not challenge the turbine, which it is already highlighted in the original manuscript, lines 258 - 260. The time instants when velocities are lower or equal to 0.5 m/s are now included in Figure 7 and indicated in the text in lines 254 – 255: “Shaded grey areas in Figure 7 outline periods of slack water with flow velocities below a velocity threshold of 0.5 m/s”.
- Figure 8: I suggest removing the TI profiles (panels c and d) for slack conditions. They are not particularly meaningful, and are somewhat misleading (i.e., somewhat might interpret this as "wow the turbulence is extremely high at this site").
Profiles have been removed as suggested.
- Figure 8: There seems to be two layers in the flow: a high-shear bottom boundary layer, and a relatively constant speed upper layer. During ebb the BBL is ~15-20m thick, during flood it is ~12m thick. Do you have any hypothesis as to why this would be? Is it bottom roughness? Is there stratification? Other ideas? Also, does this suggest there may be significant advantages to positioning the turbine higher in the water column?
The boundary layer thicknesses during different tidal phases are attributed to bed roughness effects and flow velocity variation. Boundary layer theory accounts for these two factors in the logarithmic distribution of velocities so we attribute our results are mainly due to these former factors. As the reviewer points out, there would be indeed an obvious advantage if the turbine would be placed near the free-surface, e.g. floating turbine, but the analysed device was bottom-fixed.
- Lines 202-307: More details on how spectra were calculated are required. What was the length of your spectral window? Was a window function applied? How many ensembles went into each spectral plot? What is the significance of the -1 slope, and the -5/3 slope (add references)? Any hypothesis for the spectral peak at 10^-1 Hz in the Ebb spectra?
This comment is very useful, thank you. We have added the following text in line 294: “The pwelchfunction was used to compute the spectra, dividing the signal into 20 ensembles”. The -1 slope is commonly found in boundary layer flows resulting from the largest scales, whilst the -5/3 slope corresponds to the inertial subrange. We have included references to Chamorro and Porte-Agel (2009 Boundary-Layer Meteorology 132: 129-149) and Ouro and Stoesser (2019, Flow, Turbulence and Combustion). The peak at 0.1Hz might be due to roughness-induced turbulent structures which occur at larger frequencies than the largest flow scales, as found in Ouro and Stoesser (2019), but this cannot be verified with the field data available.
- Section 4.3: This is very interesting and well presented. I think this is the most important piece of this paper. On line 351 you state that "further refinement is required to improve the derivation of Fxroot". Can you provide details on what kinds of 'refinement' might be effective?
Text has been added in lines 384 - 291 to state that the methodology could be improved by performing a secondary calibration and by moving the sensors at the problematic 1.7 m radial positions. However, since both of these recommendations require the physical removal of the turbine, the robustness of the existing arrangement could be improved by further modification to the weighting applied to measurements, such that the agreement between Fxrootloads calculated on the compression and tension surfaces of the blades improves. Similarly, the agreement between blade-to-blade Fxrootaveraged over a long enough period should improve.
- Figure 14: Again a clearer description of how spectra were calculated is needed. Are the spectra calculated identical to Figure 9 (same window, etc.)? Furthermore, it is uncommon to see a large spike in the spectra at the highest frequencies. If it is real this suggest that high-frequency signals exist in the measured variable, and therefore that aliasing of signals above the measurement frequency could be occurring. However, I can't help but wonder if this is noise in the measurement. Are there ever time-periods when this peak does not exist? Does this peak exist in the calibration dataset? A better description of the source of this peak is needed.
The exact same procedure as in the calculations of spectra in Figure 9 was followed. This has been clarified in line 418. The reviewer is right that such a large energy value is not frequent. However, in these tests the data were sampled at 16 Hz which nearly coincides with a harmonic of the rotational frequency, thus both effects can lead to such a peak. This has been clarified in line 433. Alternatively, as indicated in the original manuscript, Payne et al. (2019, Journal of Fluids and structures 83: 156 – 170) also observed such high-peaks at harmonics of the rotational frequency. Hence, we think calibration is not the issue here.
