Share this page on:

Conference programme 

Back to the programme printer.gif Print

Poster session

Lead Session Chair:
Stephan Barth, Managing Director, ForWind - Center for Wind Energy Research, Germany
Tuhfe Gocmen Bozkurt DTU Wind Energy, Denmark
Tuhfe Gocmen Bozkurt (1) F P Gregor Giebel (1) Mads Rajczyk Skjelmose (2)
(1) DTU Wind Energy, Roskilde, Denmark (2) Vattenfall Renewables Wind DK A/S , Fredericia, Denmark

Printer friendly version: printer.gif Print

Download poster(8.34 MB)

Presenter's biography

Biographies are supplied directly by presenters at OFFSHORE 2015 and are published here unedited

Tuhfe Gocmen Bozkurt is currently a PhD Student in Technical University of Denmark, DTU, working in the PossPOW project, Possible Power of Downregulated Offshore Wind Farms. She studied aerospace engineering in Middle East Technical University in Ankara and wrote her master thesis about wind turbine aerodynamics in Izmir Institute of Technology. Her PhD research is focused on the behavior of wind flow inside a wind farm, downregulation characteristics of wind turbines and estimation of available power.


Real-time available power estimation for offshore wind power plants


Available (or Possible) Power is the power that a turbine or a wind power plant would produce if it had not been down-regulated (or curtailed). In modern offshore wind power plants, available power estimation is becoming more of an issue due to increasing shares in wind and operating reserve calculations. Currently, the Transmission System Operators have no real way to determine exactly the available power of a down-regulated wind farm. The PossPOW project is addressing that need, aiming to develop an industry standard method for the real-time estimation of the available power of a wind power plant (see


Even though modern wind turbines have a SCADA signal called available power, the sum of those signals is more than the available power of the entire wind farm since the wake losses decrease during curtailment. Therefore, to calculate the real-time (1-sec) available power at the wind farm scale, the second-wise free wind speed at the turbine locations has to be estimated and advected using a wake model calibrated for the same resolution. In this paper, the details of that estimation algorithm together with the designated experiments in Horns-Rev offshore wind farm and hopefully the first validation results will be presented.

Main body of abstract

During down-regulation, the optimal power curve is no longer valid and the nacelle wind speed may induce unacceptable uncertainties especially for real-time calculations during shorter periods. Therefore, the method proposed takes power, pitch angle, and rotational speed as inputs and also requires the coefficient of power, Cp. The calculated rotor effective wind speeds were validated using 1-sec data obtained from the Horns Rev and Thanet wind farms, together with NREL 5MW simulations.
The estimated wind speeds of upstream turbines are read by the wake model to estimate the velocity deficit and calculate the possible power of the wind farm. However, most of the computationally affordable wake models are tuned for 10-min averaged data. Therefore, the GCLarsen wake model is re-calibrated for single wake cases and then extended to the farm scale considering the time delay and meandering. The preliminary results indicate that the wind velocity at the downstream turbines can be estimated with a maximum average error of 12% using the re-calibrated model on 1-sec dataset. When implemented during optimal operation, the algorithm provides a real-time wind farm power curve.
The test and validation of the algorithm is rather challenging since there is no actual measure of the available power for wind farm scale. However, we can benefit from the similarity in power production between the neighbouring rows in a simple layout like Horns Rev wind farm. The idea behind the validation and the experimental setup is described in Gregor et al. in the same proceedings.


The algorithm of the wind farm scale available power estimation in the context of the PossPOW project is presented. The results of the real-time wake model are evaluated only using the normal operation, where active power is equal to the available power. However, the algorithm will also be tested using the experiments described in Gregor et al. in the proceedings and hopefully the preliminary results of the actual validation of the offshore wind power plants available power algorithm will be presented.

Learning objectives
Available (or possible) power estimation
Real-time wake modelling
Real-time wind farm power curve