- |
Concert – Control and uncertainties in real-time power curves of offshore wind power plants
![]() |
Gregor Giebel DTU Wind Energy, Denmark CONCERT – CONTROL AND UNCERTAINTIES IN REAL-TIME POWER CURVES OF OFFSHORE WIND POWER PLANTS Abstract ID: 307 Poster code: PO.144 | Download poster: ![]() |
Presenter's biography
Biographies are supplied directly by presenters at WindEurope 2016 and are published here uneditedDr. Gregor Giebel is Senior Scientist at DTU Wind Energy in Risø. His main topic is short-term prediction of wind energy and integration in the grid. He is the Operating Agent of the newly built IEA Wind Forecasiting Task, and collaborates on standardisation within IEC and SGIP. During his 20 years in wind power, he also looked into wind resource estimation, the use of drones for atmospheric measurements, and condition monitoring for the drive train.
Abstract
Concert – Control and uncertainties in real-time power curves of offshore wind power plants
Introduction
Modern wind farms need to be grid-friendly in times where wind power produces a sizeable amount of the total electricity in the grid. Their controllability plays a crucial role, to be downregulated when there is too much power in the grid, or to provide some upregulation potential when there is danger of too little power in the grid. For the latter case, the calculation of the possible reserve power needs an accurate assessment of the power available from the wind power plant given the current wind conditions. During the last 4 years, the ForskEL PossPOW project developed the necessary real-time power curve for offshore wind power plants, and verified it with some experiments. The follow-up project Concert now enhances the PossPOW real-time power curve with the impact on loads, with an uncertainty calculation, and with additional experiments. Concert also uses PossPOW to find an optimal control strategy to distribute the set-points on the turbines with regard to the total power.
Approach
The real-time power curve algorithm PossPOW will be used as a basis for a calculation of the local wind speed and turbulence at the turbine positions within a downregulated wind farm. The improvements attained during the project will be tested with dedicated experiments at offshore wind power plants of project partner Vattenfall, ideally containing wind turbines of project partner Siemens Wind Power. The convolution of the for actual use employed forecasts’ with the algorithm’s uncertainty is investigated. Finally, the improved wind power plant control will be tested for actual trading at Vattenfall’s trading office in Sweden.
Main body of abstract
The algorithm takes the wake effects between turbines into account to estimate the wind farm scale real-time power curve. The Concert project aims to build up a comprehensive uncertainty estimation methodology for this previously developed real-time power estimator to evaluate the contribution of the wind farm available power uncertainty to the uncertainty of the forecasted available power. The preliminary data based uncertainty quantification approach (UQ) is submitted as another abstract entitled “Uncertainty Quantification of the Real-Time Reserves for Offshore Wind Power Plants” by Gocmen et al. The results of the objective UQ will then be compared to the traditional UQ and propagation methods (e.g. Monte Carlo techniques, Galerkin approaches). The uncertainty module in the Concert project will be finalised by the implementation of the machine learning techniques to reduce the uncertainty and further enhance the PossPOW algorithm and the real-time power curve estimation.
After the available power algorithm is enhanced, a smart spatial distribution of the down-regulation criteria, i.e. set-points, within the turbines in a wind farm or across several wind farms in a region is possible. Spatially distributing the set-points reduces the correlation be-tween each down-regulation and thereby the correlation of their uncertainties. As independent uncertainties should compensate each-other, that would reduce the combined uncertainty of the down-regulation. That optimization algorithm to reduce uncertainties will then be elaborated in considering the power production and the loads as well as the uncertainty. The load calculation will be based on and verified with the Dynamic Wake Meander model, but for performance and real-time reasons will probably employ a pre-calculated look-up table.
The power optimisation and other control objectives are the topic of another poster by Kazda et al.
Conclusion
The poster presents a new Danish national project, enhancing the PossPOW real-time power curve with a full uncertainty calculation, with a load parameterisation amenable for real-time control purposes, and with optimal control for maximum power extraction or minimum loads. The results will be tested at operating offshore wind farms and for trading on the electricity exchange.
Funding from the Danish programme ForskEL is acknowledged.
Learning objectives
The PossPOW project has gotten a successor, Concert, enhancing the PossPOW real-time power curve in several aspects.
Aim of Concert is to distribute set-points within a wind farms for optimal power production, load reduction or minimisation of uncertainty.
Follow EWEA on: