Simulation and Prediction of Wakes and Wake Interaction in Wind Farms

    Research output: Book/ReportPh.D. thesis

    531 Downloads (Orbit)

    Abstract

    The highly turbulent wake and the wake interaction of merging wakes between multiple wind turbines are modelled using Large Eddy Simulation (LES) in a general Navier-Stokes solver. The Actuator Line (AL) technique is employed to model the wind turbines, and the aeroelastic computations are fully coupled with the flow solver. The numerical simulations include the study of the far wake behind a single turbine, three idealised cases of infinitely long rows of turbines and finally three infinite wind farm scenarios with different spacings. The flow characteristics between the turbines, turbine performance, and principal turbulent quantities are examined for the different scenarios. The study focuses on the large coherent structures and movements of the wake behind and between wind turbines. The large coherent structures are analysed using Proper Orthogonal Decoposition (POD). POD constitutes the basis for two proposed dynamic wake models of the turbulent wake deep inside large wind farms. The first model is based on a direct reconstruction using POD, while the other model (REDOMO) is based on an additional reduction by only including the most dominant frequencies. The flow fields derived from the two wake models are assessed and verified by comparing turbine performance and loads to those derived from the flow extracted from the numerical simulations. The most comprehensive model yields excellent agreement for small and intermediate turbine spacing, while the simpler version is unable to resolve the complex dynamics due to severe temporal filtering. The models have difficulties capturing the more extreme and spurious events for larger turbine spacings. The performance is also compared to stochastically generated Mann turbulence, which gives better results for larger spacings. The comparison also reveals how much information should be retained by the POD models to add more value than simply applying homogeneous turbulence as inflow.
    Original languageEnglish
    PublisherWind Energy Department, Technical University of Denmark
    Number of pages170
    Publication statusPublished - 2014

    Fingerprint

    Dive into the research topics of 'Simulation and Prediction of Wakes and Wake Interaction in Wind Farms'. Together they form a unique fingerprint.

    Cite this