Fluid-Structure Interaction for Wind Turbines in Atmospheric Flow

Christian Grinderslev*

*Corresponding author for this work

Research output: Book/ReportPh.D. thesis

19 Downloads (Pure)

Abstract

This thesis presents work conducted within the field of aero-elasticity of wind turbines in atmospheric wind flow. This involves investigation of the complex and chaotic turbulent flows in the lower atmosphere along with the structural response of the wind turbines. To do so, fluid-structure interaction (FSI) simulations are conducted, which combine high fidelity flow modelling through computational fluid dynamics (CFD) with structural response modelling to capture the coupled effects between wind and structure. This method becomes increasingly relevant for wind turbines, as these are continuously being designed larger and more flexible than ever before. This means that the aero-elastic stability of the blades becomes essential to consider, and that current engineering aerodynamic models are used outside the validated envelop. In order to enable these FSI simulations, various existing CFD capabilities are combined to consider both relative motion of the rotor and surroundings along with blade flexibility, through the overset grid method and the FSI coupling respectively. The overset method has further been enhanced with a tighter coupling between overlapping grids, through coupling of pressure gradients. Finally, to consider realistic atmospheric flow, a hybrid turbulence model have been developed and tested, enabling the capabilities of simulating turbulent flow in a large range of scales; from the small blade boundary layer scales to the large turbulent scales of the atmospheric boundary layer (ABL).

As initial steps, the individual capabilities have been validated before being combined in more complex studies. The FSI framework is first validated by simulations of an experimental blade test of a relatively short wind turbine blade of ≈14m length, conducted in the experimental test facility of the Technical University of Denmark. The study shows excellent agreement between simulations and measurements, and finds that the confinement of the test facility floor and walls has insignificant effects on the aerodynamics during the test. The study likewise calibrates proper values of blade section drag coefficients, which are needed as inputs to the engineering models, often used for fatigue test planning, and are underestimated in current models. Further, three studies are conducted of a 2.3 MW wind turbine rotor with ≈40m blades, placed within the ABL, with experimental measurements available for validation. First, a validation study of the CFD method is conducted with the rotor present in sheared laminar wind flow including tilt and extreme yaw error. It is found that the CFD solver and chosen methods capture well the loading of the rotor, and important setup considerations and corrections needed for subsequent studies are identified. After independent validations of both FSI framework and CFD setup and methods, confidence is gained in the capabilities. More complex simulations combining the two are therefore conducted to study the aero-elastic effects of wind turbines in complete atmospheric flow. Here, FSI simulations including flexibility of the rotor, in the formerly treated complex flow scenarios are conducted with an improved CFD setup. This, along with lower-fidelity aero elastic simulations using the simpler blade element momentum (BEM) aerodynamics to clarify the differences seen between FSI and the less costly engineering models. The importance of good airfoil data and corrections for the engineering model is clarified, as the BEM based simulations do not capture well the loads due to early stall predictions. The stall is not found in neither measurements nor FSI simulations, which agree well. The impact of considering the flexibility is small for the specific relatively stiff rotor, seen in both loading and resulting flow wakes behind the rotor computed in the FSI simulations. Finally, another step on the ladder of complexity is taken, as inflow turbulence is added to the simulations, using the novel turbulence model developed. Here, a large eddy simulation (LES) model developed for atmospheric flows is combined with the improved delayed detached eddy simulation (IDDES) model for the separated flow, e.g in the rotor wake. This model enables geometrically resolved simulations of wind turbines inside the atmospheric flow, without an excessive need of grid resolution near the rotor surface. A neutrally stratified turbulent flow is simulated with and without the rotor present in the flow. For rotor calculations, cases with and without flexibility of the blades are simulated to investigate the aeroelastic effects. A large load dependency is found on the incoming turbulence, whereas flexibility of this specific rotor shows little effect, due to its relatively stiff design, as seen in the aforementioned study as well. The developed hybrid turbulence model does not yield significantly different results than the well established IDDES model, but is expected to do so for more complex flows and/or larger flow domains. Future studies are, however, needed to explore this.

In general, this project investigates and demonstrates methods of various complexities for the study of wind turbines in ABL flow conditions. This, from the efficient BEM based methods to the high fidelity FSI simulations using the overset mesh method and including turbulent ABL flow modelling. The findings lead the way of future investigations of modern wind turbine designs, with blades of more than 100m, for which accurate consideration of flexibility and the ABL will be of high importance. These high fidelity studies can give new insights in the complex aero-elastic phenomena which can occur between wind and structure, and in the worst case lead to instabilities with fatal structural consequences. Likewise, simulations will be necessary in order of aiding the development of the efficient engineering models used in industry, which are reaching their limits as shown in the conducted studies. To do this, FSI simulations are convenient, as experimental data are rarely available in the large scales considered.
Original languageEnglish
Place of PublicationRisø, Roskilde, Denmark
PublisherDTU Wind Energy
Number of pages170
DOIs
Publication statusPublished - 2020

Fingerprint Dive into the research topics of 'Fluid-Structure Interaction for Wind Turbines in Atmospheric Flow'. Together they form a unique fingerprint.

Cite this