Modelling Wind Turbine Wakes at Middelgrunden Wind Farm

Abstract

As part of the development of an offshore wind farm layout optimisation tool, this paper explores the accuracy and computational time of wake models applied to Middelgrunden Wind Farm outside of Copenhagen, Denmark. In this study, four years of data from 2001 to 2004 are used to test the applicability, accuracy, and computational time of the Jensen, Larsen, Ishihara, and a simplified version of the Ainslie Eddy-Viscosity wake models. This study has shown that the size of the directional sector used in the comparison and if that directional sector is applied to all turbines’ incoming wind velocities or just the northernmost greatly affects the results. From this it is found that the Larsen wake model provides the best balance between accuracy and computational time. It also shows that even a simplified version of a field model takes significantly longer to compute than an analytic model. This study has also shown that using directional sectors of ±15◦ these models perform similarly to previous studies at Nysted and Horns Rev indicating that the close spacing (2.4D) at Middelgrunden is not too close for the use of these models.