the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 23 Jul 2021
Approaches for predicting wind turbine hub-height turbulence metrics
Abstract. Hub-height turbulence is essential for a variety of wind energy applications, ranging from wind plant siting to wind turbine control strategies. Because deploying hub-height meteorological towers can be a challenge, alternative ways to estimate hub-height turbulence are desired. In this paper, we assess to what degree hub-height turbulence can be estimated via other hub-height variables or ground-level atmospheric measurements in complex terrain, using observations from three meteorological towers at the Perdigão and WFIP2 field campaigns. We find a large variability across the three considered towers when trying to model hub-height turbulence intensity (TI) and turbulence kinetic energy (TKE) from hub-height or near-surface measurements of either wind speed, TI, or TKE. Moreover, we find that based on the characteristics of the specific site, atmospheric stability and upwind fetch either determine a significant variability in hub-height turbulence or are not a main driver of the variability in hub-height TI and TKE. Our results highlight how hub-height turbulence is simultaneously sensitive to numerous different factors, so that no simple and universal relationship can be determined to vertically extrapolate turbulence from near-surface measurements, or model it from other hub-height variables when considering univariate relationships. We suggest that a multivariate approach should instead be considered, possibly leveraging the capabilities of machine learning nonlinear algorithms.
This preprint has been withdrawn.
-
Withdrawal notice
This preprint has been withdrawn.
-
Preprint
(8920 KB)
Interactive discussion
Status: closed
-
RC1: 'Reviewer comment on wes-2021-68', Anonymous Referee #1, 17 Aug 2021
The manuscript discusses an approach for estimating turbulence intensity (TI) and turbulent kinetic energy (TKE) from near-surface measurements in complex terrain sites. The selected topic is important and especially relevant for onshore wind energy applications with growing hub heights. For that, the authors show correlation plots between TI and TKE at distinct heights with chosen datasets combined in 100 bins with variables sizes and uniform sample size. A polynomial regression is than applied to the binned data. The authors conclude that no simple or universal relationship can be drawn from their univariate approach.
This manuscript has major issues.
The authors claim the goal is to assess to what degree TI or TKE at 80 m above ground (called “hub-height” by the authors) can be estimated via near-surface measurements. However, the proposed methodology is based solely on a correlation between variables. The validity of the presented regression plots to other heights in the exact location or other locations is most likely not applicable, especially in complex terrain. For the binned data, the error bars are not shown, which is important to assess how representative the regression functions are. For the polynomial functions, no regression coefficient is shown, which also hinders the analysis. The authors show that not even the TI vs. wind speed relation holds when using near-surface observations. Even for the heights and locations presented, the authors do not quantify the error in case the fitted polynomial functions were to be used.
In my view, the quantity and choice of datasets was unfortunate. The authors choose only three pairs of measurements (at 10 m and 80 m above ground) from two complex terrain experiments (Perdigão and WFIP2), which does not allow any general conclusion for the observed trends in complex terrain, apart from site-specific remarks or straightforward conclusions, such as a positive correlation between TI and TKE. Also, the relations between 10 m and 80 m at WFIP2 are done with measurements located more than 1 km apart, i.e., likely further than the upstream fetch of the 80 m observations. In terms of temporal coverage, the authors do not detail the length of the filtered dataset. Other sites could be used for this purpose, e.g., other experiments within the New European Wind Atlas (NEWA) project that have multiple masts with 3D ultrasonic anemometers sampling at 20Hz for a one-year period.
I cannot foresee any additional analysis within the lines of the proposed methodology that would justify the publication. The authors promote machine learning techniques as a potential solution for this research problem. However, the extension of the presented results using such tools would be, in my view, the scope for a completely new paper. Therefore, I suggest the rejection of this manuscript in its current state.
Citation: https://doi.org/10.5194/wes-2021-68-RC1 -
RC2: 'Comment on wes-2021-68', Anonymous Referee #2, 22 Aug 2021
Relationships between hub-height turbulence and ground-level atmospheric measurements would be helpful, no doubt. However, as experienced by the authors, there is no such thing. It would be excellent if simple power-law correlations could solve turbulence.
The manuscript is nothing more than a series of plots showing the curve fitting of the experimental data.
There is no justification for using and assuming one single mathematical function: power law.
Discussion of the results is feeble. For instance, why do coefficients and exponents differ among the various data sets?
There is no quantification of the success of the curve fitting.
Tower TSE09, in the valley, is not relevant for wind energy studies. The flow in the valley is much different from the flow near tower TSE04, as evidenced in the figures, and it is not reasonable to expect similar trends if they existed.
TI (turbulence intensity, eq. 1) was calculated over 10-minute intervals to match the current wind energy industry standard, and TKE (turbulent kinetic energy, eq. 2) was calculated over 30-minute intervals, a common choice to study atmospheric boundary layer processes. This practice is not consistent and, therefore, unacceptable.
The authors' main objective is to find correlations between hub-height turbulence and ground-level atmospheric measurements. However, why did they take this endeavour? Were they aware or had any signs of the success of their methodology under ideal conditions, i.e., flat terrain and neutral conditions? Of course, they were not, simply because there are none.
In conclusion, I cannot recommend the publication.
Citation: https://doi.org/10.5194/wes-2021-68-RC2 -
EC1: 'Comment on wes-2021-68', Andrea Hahmann, 05 Sep 2021
Dear Hannah Livingston and co-authors,
The reviewers agree that since there are no signs of the success of your methodology under ideal conditions, they are unlikely also in complex terrain. Thus, the main objective of the manuscript is unfounded. The two reviewers also pointed out other significant deficiencies:
- There is no justification for using and assuming one single power law as a mathematical function.
- The quantity and choice of datasets were unfortunate. Mainly when the WFIP2 measurements at 10 m and 80 m are more than 1 km apart, and the location of the Perdigão measurements is in the valley. There are also inconsistencies in the data treatment.
- The statistical relationships shown are oversimplified, without error quantification or error bars.
You are welcome to respond to all comments. However, the problems appear very improbable to be fixed upon revision. In that case, the manuscript will likely be rejected for publication in WES.
Best regards,
Andrea Hahmann
WES Associate EditorCitation: https://doi.org/10.5194/wes-2021-68-EC1
Interactive discussion
Status: closed
-
RC1: 'Reviewer comment on wes-2021-68', Anonymous Referee #1, 17 Aug 2021
The manuscript discusses an approach for estimating turbulence intensity (TI) and turbulent kinetic energy (TKE) from near-surface measurements in complex terrain sites. The selected topic is important and especially relevant for onshore wind energy applications with growing hub heights. For that, the authors show correlation plots between TI and TKE at distinct heights with chosen datasets combined in 100 bins with variables sizes and uniform sample size. A polynomial regression is than applied to the binned data. The authors conclude that no simple or universal relationship can be drawn from their univariate approach.
This manuscript has major issues.
The authors claim the goal is to assess to what degree TI or TKE at 80 m above ground (called “hub-height” by the authors) can be estimated via near-surface measurements. However, the proposed methodology is based solely on a correlation between variables. The validity of the presented regression plots to other heights in the exact location or other locations is most likely not applicable, especially in complex terrain. For the binned data, the error bars are not shown, which is important to assess how representative the regression functions are. For the polynomial functions, no regression coefficient is shown, which also hinders the analysis. The authors show that not even the TI vs. wind speed relation holds when using near-surface observations. Even for the heights and locations presented, the authors do not quantify the error in case the fitted polynomial functions were to be used.
In my view, the quantity and choice of datasets was unfortunate. The authors choose only three pairs of measurements (at 10 m and 80 m above ground) from two complex terrain experiments (Perdigão and WFIP2), which does not allow any general conclusion for the observed trends in complex terrain, apart from site-specific remarks or straightforward conclusions, such as a positive correlation between TI and TKE. Also, the relations between 10 m and 80 m at WFIP2 are done with measurements located more than 1 km apart, i.e., likely further than the upstream fetch of the 80 m observations. In terms of temporal coverage, the authors do not detail the length of the filtered dataset. Other sites could be used for this purpose, e.g., other experiments within the New European Wind Atlas (NEWA) project that have multiple masts with 3D ultrasonic anemometers sampling at 20Hz for a one-year period.
I cannot foresee any additional analysis within the lines of the proposed methodology that would justify the publication. The authors promote machine learning techniques as a potential solution for this research problem. However, the extension of the presented results using such tools would be, in my view, the scope for a completely new paper. Therefore, I suggest the rejection of this manuscript in its current state.
Citation: https://doi.org/10.5194/wes-2021-68-RC1 -
RC2: 'Comment on wes-2021-68', Anonymous Referee #2, 22 Aug 2021
Relationships between hub-height turbulence and ground-level atmospheric measurements would be helpful, no doubt. However, as experienced by the authors, there is no such thing. It would be excellent if simple power-law correlations could solve turbulence.
The manuscript is nothing more than a series of plots showing the curve fitting of the experimental data.
There is no justification for using and assuming one single mathematical function: power law.
Discussion of the results is feeble. For instance, why do coefficients and exponents differ among the various data sets?
There is no quantification of the success of the curve fitting.
Tower TSE09, in the valley, is not relevant for wind energy studies. The flow in the valley is much different from the flow near tower TSE04, as evidenced in the figures, and it is not reasonable to expect similar trends if they existed.
TI (turbulence intensity, eq. 1) was calculated over 10-minute intervals to match the current wind energy industry standard, and TKE (turbulent kinetic energy, eq. 2) was calculated over 30-minute intervals, a common choice to study atmospheric boundary layer processes. This practice is not consistent and, therefore, unacceptable.
The authors' main objective is to find correlations between hub-height turbulence and ground-level atmospheric measurements. However, why did they take this endeavour? Were they aware or had any signs of the success of their methodology under ideal conditions, i.e., flat terrain and neutral conditions? Of course, they were not, simply because there are none.
In conclusion, I cannot recommend the publication.
Citation: https://doi.org/10.5194/wes-2021-68-RC2 -
EC1: 'Comment on wes-2021-68', Andrea Hahmann, 05 Sep 2021
Dear Hannah Livingston and co-authors,
The reviewers agree that since there are no signs of the success of your methodology under ideal conditions, they are unlikely also in complex terrain. Thus, the main objective of the manuscript is unfounded. The two reviewers also pointed out other significant deficiencies:
- There is no justification for using and assuming one single power law as a mathematical function.
- The quantity and choice of datasets were unfortunate. Mainly when the WFIP2 measurements at 10 m and 80 m are more than 1 km apart, and the location of the Perdigão measurements is in the valley. There are also inconsistencies in the data treatment.
- The statistical relationships shown are oversimplified, without error quantification or error bars.
You are welcome to respond to all comments. However, the problems appear very improbable to be fixed upon revision. In that case, the manuscript will likely be rejected for publication in WES.
Best regards,
Andrea Hahmann
WES Associate EditorCitation: https://doi.org/10.5194/wes-2021-68-EC1
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 574 | 277 | 26 | 877 | 22 | 24 |
- HTML: 574
- PDF: 277
- XML: 26
- Total: 877
- BibTeX: 22
- EndNote: 24
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1