Wind direction estimation using SCADA data with consensus-based optimization

Annoni, Jennifer; Bay, Christopher; Johnson, Kathryn; Dall'Anese, Emiliano; Quon, Eliot; Kemper, Travis; Fleming, Paul

doi:https://doi.org/10.5194/wes-4-355-2019

Articles | Volume 4, issue 2

https://doi.org/10.5194/wes-4-355-2019

© Author(s) 2019. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/wes-4-355-2019

© Author(s) 2019. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 4, issue 2

Research article

|

20 Jun 2019

Research article |

| 20 Jun 2019

Wind direction estimation using SCADA data with consensus-based optimization

Jennifer Annoni, Christopher Bay, Kathryn Johnson, Emiliano Dall'Anese, Eliot Quon, Travis Kemper, and Paul Fleming

Download

Final revised paper (published on 20 Jun 2019)
Preprint (discussion started on 05 Oct 2018)

Interactive discussion

Status: closed

AC: Author comment | RC: Referee comment | SC: Short comment | EC: Editor comment

- Printer-friendly version

- Supplement

RC1: 'Review', Anonymous Referee #1, 29 Oct 2018
RC2: 'Results from Scopus', Anonymous Referee #2, 20 Nov 2018
RC3: 'Interesting Approach - a more thorough investigation required', Anonymous Referee #3, 23 Nov 2018
AC1: 'Author Comments to "A Framework for Autonomous Wind Farms: Wind Direction Consenus"', Jennifer King, 04 Feb 2019

Peer-review completion

AR: Author's response | RR: Referee report | ED: Editor decision

AR by Jennifer King on behalf of the Authors (04 Feb 2019) Author's response Manuscript

ED: Referee Nomination & Report Request started (22 Mar 2019) by Sandrine Aubrun

RR by Anonymous Referee #1 (15 Apr 2019)

RR by Anonymous Referee #3 (02 May 2019)

Suggestions for revision or reasons for rejection

Thanks for the clarifications and additions made to the article. The revised version of the manuscript is much improved. Although the very relevant discussion on turbulence sensitivity remains lacking, I hope that the future improvement of the approach will include that investigation. For this manuscript, I suggest to add it to the Future Work/Conclusions Section.

More detailed comments are listed below:
- Page 9, (regarding) Figure 2:
The definition of the error term here should be stated (for mean in this Figure, and Max in Table 1 later in the text), especially since seemingly 90deg error gives 15deg error for the top left plot of the sensors. If it is a "representative error" for the whole wind farm (i.e. Individual error divided by number of turbines) the reasons for that representation should be discussed. For the proposed use cases of the presented method (wake steering, faulty WD signal detection, etc.), it is important to know which turbines within the wind farm have erroneous signals. This is because some erroneous WD signals are more crucial than the others within the wind farm for e.g. wake modelling.
The definition of this error also leads to a peculiar behavior comparing the averaging methods. For example, although the weighted average has 3 out of 6 turbines matching to "True" value, it still has higher error than the Equal average which has all the turbines pointing to a different direction.

- Page 12, lines 5-8:
It is seen in Figure 6 error comparison that, the performance of the simpler methods (particularly Cluster averaging, as can be expected) are similar to the proposed method. This similarity and the advantage of the proposed method (under biased and/or faulty signals) is very briefly discussed here. Can we then say it might be recommended to use cluster averaging approach for non-biased or corrected data if simplicity and computational efficiency is a concern?

Page 15, lines 18-20:
Since there is one offset presented for the whole wind farm at Figure 10(b), should we read it as the individual offsets (which are varied across the wind farm) are normalized by the number of turbines? If that is the case, then it is more likely that a limited number of turbines would be misaligned for longer periods, instead of the whole wind farm being misaligned for more than 50% of the time. The sentences should be rephrased to clarify this.

Page 16 - Figure 9:
Would be nice to see the error distribution over the investigated period - possibly with the filtered wind speed (e.g. for wind speed > 5m/s) to see the temporal statistics of the error.

Hide

ED: Publish subject to minor revisions (review by editor) (03 May 2019) by Sandrine Aubrun

AR by Jennifer King on behalf of the Authors (14 May 2019) Author's response Manuscript

ED: Publish as is (15 May 2019) by Sandrine Aubrun

ED: Publish as is (16 May 2019) by Gerard J.W. van Bussel (Chief editor)

AR by Jennifer King on behalf of the Authors (24 May 2019) Manuscript

Short summary

Typically, turbines do not share information with nearby turbines in a wind farm. Relying on a single turbine sensor on the back of a turbine nacelle can lead to large errors in yaw misalignment or excessive yawing due to noisy sensor measurements. The wind farm consensus control approach in this paper shows the benefits of sharing information between nearby turbines by computing a robust estimate of the wind direction using noisy sensor information from these neighboring turbines.