Power Curve Working Group Overview Peter Stuart Senior Technical Manager Tuesday 06th October 2014

Power Curve Working Group Motivation

What does the PCWG aim to do? The power curve working group (PCWG) formed to answer a specific question: What power will a wind turbine generate in the full range of atmospheric conditions seen in the ‘real world’? Wind Speed Power Power Curve Is a power curve based on just wind speed and density the ‘whole truth’?

Turbine Performance in the Real World Measured Power Deviation Matrix: Deviation vs TI and Wind Speed The real world has… low wind speed high wind speed low turbulence Wind Speed Turbulence intensity high turbulence low wind shear high wind shear and many combinations of the above (on same site at different times)… Wind conditions change all the time, and so does turbine performance.

Power Curve Working Group: Who are we? The power curve working group (PCWG) is a balanced and broad industry group encompassing Developers, Consultants, Manufacturers and Academics/Researches. RES, Vattenfall Crown Estate Dong, Iberdrola SSE, RWE EDF, EON, ESBI, Mainstream, Scottish Power DNV GL Natural Power AWS True Power Sgurr Wind Guard Barlovento Anemos-Jacob Prevailing Vestas Senvion GE Suzlon Siemens Nordex NREL DTU LLNJ IWES CIRCE ORE Catapult Leosphere Romowind ZephIR Vaisala The group aims to examine ways of improving understanding of wind turbine energy yield in ‘real world’ conditions. Peter Stuart (RES), Mike Anderson (RES), Alan Derrick (RES), Andrew Tindal (GLGH), Staffan Lindahl (GLGH), Mark Young (DNV), Oisin Brady (Natural Power), Michael Brower (AWS True Power), Ralph Torr (Sgurr), Rasmus Svendsen (Vestas), Ioannis Antoniou (Siemens), Peder Bay Enevoldsen (Siemens), Thomas Blodau (REPower), Henk-Jan Kooijman (GE), Wiebke Langreder (Suzlon), Gareth Craft (Crown Estate), Rebeca Rivera Lamata (Dong), Luis Prieto Godino (Iberdrola), Daniel Paredes Beato (Iberdrola), Daniel Stevens (SSE), Temi Gocheva (RWE), Patrick Moriarty (NREL), Rozenn Wagner (DTU), Axel Albers (Wind Guard)& Jan-ake Dahlberg (Vattenfall). Openness is a key principal: Proceedings of all meetings publically available at: http://www.ewea.org/events/workshops/resource-assessment-2013

Understanding the Problem and Each Other

Statement of the problem The power function of a wind turbine is dependent on wind speed, density, vertical wind shear, vertical wind veer, turbulence intensity, directional variation and inflow angle. Wind Speed Power Power Function Density Wind Shear Wind Veer Turbulence Inflow Angle

Nomenclature: Types of Correction Corrections should be applied for ‘real world’ conditions which are different to those for which a power curve is representative. These corrections fall into two categories: Type A: Adjustments made to reflect changes in the energy available for conversion across the rotor in a ten minute period due to ‘non-standard conditions’. Type B: Adjustments made to reflect changes in the conversion efficiency due to ‘non-standard conditions’. Available Energy Turbine Behaviour

Nomenclature: Types of Correction Proxy methods: relate production to a parameter which is broadly associated with changes in performances. Acknowledge that the underlying mechanisms may not be identified, but pursue regardless if methods demonstrably improve predictions. e.g. production loss based on low turbulence intensity. Analytical/physical methods: apply methods based on understanding of underlying physics/statistics. May involve use of additional measurements such as LiDAR e.g. Rotor Equivalent Wind Speed and Turbulence Correction defined in 61400-12-1.

Common Understanding in Practice Recent PCWG discussion: ‘There is a gap between the proxy and analytical methods, in particular in the low wind speed & low turbulence quadrant. Type B effects are suspected as the reason for the difference’ Measured Power Deviation Matrix: Deviation vs TI and Wind Speed

Round Robins: Understanding the Methods We Already Have

Round Robin Exercises The PCWG has held a series of round robins aims at improving common understanding of methods define in 61400-12-1: Rotor Equivalent Wind Speed Turbulence Correction (Turbulence Renormalisation) Equivalent Wind Speed Including Veer Site Specific Power Curves (coming soon) The Round Robins have been invaluable in prompting detailed discussion in how best to apply these methods. The final product of the round robins is a publically available ‘consensus analysis’.

Round Robin: Application of REWS Method Pretty good agreement!

Round Robin: Application of Turbulence Correction Method Not so good agreement!

Round Robins The disagreement in the Turbulence Correction Method demonstrated that people find this method relatively difficult to apply. To address this issue a consensus implementation of the Turbulence Correction has been developed. Dropbox\PowerCurveWorkingGroup\Consensus Analysis The consensus analysis uses a pure excel implementation of the turbulence correction method (Excel array formulas) A ‘guided tour’ of the consensus analysis was presented at the September 2014 PCWG meeting. An excel consensus analysis of the Rotor Equivalent Wind Speed Method is also available in DropBox.

Inner-Outer Range Proposal

Stakeholder Interaction: Inner and Outer Range Concept Performance < 100% (on average) Inner Range Performance = 100% (on average) Turbulence Shear

Stakeholder Interaction: Inner and Outer Range Concept The inner-out range concept has been proposed as a pragmatic framework for offering a ‘two-tier’ performance warranty. AEPReference(u): The reference AEP per wind speed bin. AEPmeasured(u): The measured AEP per wind speed bin. F(u): The fraction of data in the outer range per wind speed. R: (Warranty level in Outer Range) / (Warranty level in Inner Range) Proposal document is publically available at: http://www.ewea.org/events/workshops/wp-content/uploads/2014/03/PCWG-Inner-Outer-Range-Proposal-Dec-2013.pdf The proposal has already been used contractually by some working group members.

Inner-Outer Range Proposal

PCWG Open Source Analysis Tool The first release of the (Version 0.5.0) of the PCWG Open Source Analysis tool is now available to download. The code is provided without warranty under the terms of the MIT software license (see attached for more details). The tool itself can be downloaded at: https://sourceforge.net/projects/pcwg/files The tool has been benchmarked against the Excel Consensus Analysis of the working group Round Robin Exercises. The tool is open source and working group members are encouraged to contribute. Those interested can access the project source code on GitHub: https://github.com/peterdougstuart/PCWG

PCWG Tool: Calculation of Power Curve Deviation Matrices Example Analysis of combined dataset of 5 turbines of the same type Deviation after Turb Correction Deviation before Turb Correction

Data Sharing

Data Sharing The PCWG currently has 5 datasets which have been supplied to the group by its members. The existing datasets have been invaluable, but much more could be achieved if more data was available. In particular analysis like the power deviation matrices require a substantial amount of data to generate something meaningful. If you have any data which you can share please get in touch! The ideal dataset is a power performance test dataset with both masts and LiDARs. Traditional mast based power performance datasets are also highly useful.

What Next & Conclusions

Site Specific Power Curves Round Robin What Next for the PCWG Publication of companion document for Turbulence Correction Consensus Analysis i.e. a guide to applying this method. Site Specific Power Curves Round Robin Publication of document expressing the ideal format for communicating power curve information. New features in Open Source Analysis Tool e.g. site calibration module. Innovative data sharing exercise: find new ways to bring data together Examine correction methods for other parameters e.g. inflow angle Drill into ‘gap’ between proxy and analytical methods (Type B effects).

This is an issue the industry has the power to solve! Conclusions The PCWG has been formed to find practical approaches to deal with the issue of turbine performance in real world conditions. The PCWG has reached a consensus that the power function of a wind turbine is dependent on wind speed, density, vertical wind shear, vertical wind veer, turbulence intensity, directional variation and inflow angle. The PCWG is exploring corrections for ‘real-world’ wind conditions and new methods of stake holder interaction in order to give a more realistic expectation of turbine performance. These corrections will help the wind industry further improve/refine the accuracy of its energy yield predictions and thus improve investor confidence. Further collaboration between manufacturers, developers and consultants is required to improve communication of power function information and explore corrections for real world conditions. This is an issue the industry has the power to solve!

Current Status Power Curve Working Group Roadmap Definition Solution / Evolution Conclusion Meeting 1 Meeting 2 Round Robin 1 Meeting 3 Round Robin 2 Meeting 4 Final Meeting Define what’s the problem we are trying to solve. Identify possible solutions Trial solutions Feedback on solutions. Compare experiences & lessons leant. Trial refined solutions Feedback on refined solutions. Finalise conclusions Publication of journal paper by working group. Is problem is solved? Current Status Identify refined and/or alternative solutions Should problem be redefined? Publication of guideline document. Iterate solutions as required… Publically disseminate presentations and minutes Time Dec 2012 Mar 2013 Apr - May 2013 May 2013 Jun – Sep 2013 Dec 2013 Jun 2014