An experimental study on rotor-wake interactions of wind turbines
Doctoral thesis
Permanent lenke
http://hdl.handle.net/11250/2614550Utgivelsesdato
2018Metadata
Vis full innførselSamlinger
Sammendrag
Interactions of wind turbine wakes with downstream turbines can reduce a wind farm’s
power production and increase loads on the individual turbines. For the purpose of
wind farm optimization, different aerodynamic approaches to modify the performance
and wake flow of one or two model wind turbines have been tested in a number of wind
tunnel experiments.
In a first set of measurements, different modifications of the rotor design to limit wake
effects are studied. Herein, the effect of the blade number on the wake development
is studied by comparing the wake properties behind 2- and 3-bladed model wind
turbines. Also, the influence of the rotational direction is investigated by comparing
the performance of an aligned two-turbine array with co- and counter-rotating rotors.
Moreover, the effect of winglets on the performance and vortex interaction in the wake
is assessed. For this purpose, a new rotor with aerodynamically optimized winglets has
been designed. The performance of the rotor is compared to a reference rotor without
winglets and effects on the vortex interaction and velocity recovery in the wake are
investigated.
The second set of measurements investigated the control of the model wind turbines
by intentional yaw misalignment. Therefore, the wake flow behind a yawed turbine
exposed to different inflow conditions is measured, while also the power and loads on
a two-turbine array are analyzed for varying separation distances, lateral offsets and
yaw angles. Selected test cases are furthermore provided for validation purposes of
CFD codes. In a Blind test experiment, performance and wake data are compared to
computational results from external groups.
All the experiments have been carried out in the closed-loop wind tunnel at NTNU
in Trondheim. The wakes were investigated for uniformly distributed and sheared
inflow velocity profiles with different turbulence intensities ranging from 0.23% to
10.0%. During the project different rotor designs from 2- to 3-bladed rotors, all with
a diameter of D = 0.9 m, are investigated. The velocities in the wake are measured
using a 2-component laser Doppler velocimetry system or a Cobra probe, which is used
to extract phase-averaged information from the wake flow.
The potential of the blade number and opposite rotational directions in turbine
array are found not to have a significant potential for the optimization of a wind farm.
While not affecting the mean velocity distribution, the blade number is observed to
influence to turbulence peak levels in the wake. An opposite rotation of the downstream
turbine is assessed only to be effective for very small turbine separation distances,
where the energy contained in the wake swirl of the upstream turbine can be extracted.
The design of aerodynamically optimized winglets could rise the power coefficient CP
of a single rotor by 8.9%, whereas the thrust coefficient CT only increased by 7.4%.
Winglets are furthermore found to accelerate the tip vortex interaction in the wake,
leading to a local shear layer enlargement and earlier wake recovery. In a wind farm,
rotors with winglets extract more energy and leave a similar amount of kinetic energy
in the wake for potential downstream turbines. Yaw control is found to have the
largest potential for the optimization of wind farms. The total power of an aligned
two-turbine array is assessed to increase up to 11% by deflecting the upstream turbine’s
wake laterally though an intentional yaw misalignment. However, yaw moments on
yawed turbines and turbines operating in a partial wake are observed to increase,
showing the importance of considering loads for yaw control. Finally, the comparison
of experimental data to numerical predictions in the Blind test confirmed the strength
of codes based on Large-Eddy Simulations (LES) in predicting mean velocity and
turbulent kinetic energy levels in the wake precisely.