IMPROVED MODELLING OF WAKES: EXPERIMENTAL STUDY AND EXPERIMENTALLY-ANCHORED MODEL

APPROACH
In the present work detailed measurements of wind turbine wakes under different ambient turbulence conditions (zero and low ambient turbulence) are made in the ETHZ dynamically-scaled wind turbine test facility; the low ambient turbulence is relevant to offshore wind farms. The measurements, over the extent of the rotor plane, extend from the near-wake region of the wind turbine to the far-wake region. Experimental observations and free shear flow theory are then used to develop a new wake model. The new model substantially improves predictions of power losses due to wakes.

RELEVANCE
In offshore wind farms, wind turbine wakes can be responsible for 5%-15% of the power losses of the whole wind farm's power output. Thus, there is compelling imperative to accurately model the expected wind turbine wake profiles, so that the attendant power losses can be minimised and the turbine spacing can be optimised, in order to ensure profitability of the wind farm.

INNOVATION
(1) By using water as the flow medium, the ETHZ facility more closely reaches full-scale non-dimensional parameters than in a comparable-sized wind tunnel. (2) An interchangeable, turbulence-generating grid upstream of the wind turbine model to generate ambient turbulence; thus the evolution of the wind turbine wake in zero ambient turbulence and in low ambient turbulence, which is relevant for offshore conditions, can be examined. (3) The evolution of wind turbine wakes up to 6.5 rotor diameters downstream is detailed. Free shear flow theory is combined with the experimental observations to formulate a new wake model.

RESULTS AND CONCLUSIONS
Measurements of wakes are compared to predictions using existent wake models. It is shown that existent wake models perform poorly. Furthermore, the evolution of wake is also shown to be strongly dependent on the level of the ambient turbulence. Increasing the ambient turbulence enhances the mixing-out process within the wake, and the near-wake region has a substantially smaller streamwise extent. The observations of the experiment and free shear flow theory are used to formulate a new wake model. Then, the performance of the model is evaluated against previously reported full-scale wind turbine measurements. The new wake model has an average error of 3.6% in the predicted performance, which is a significant improvement compared to the more commonly used a Gaussian profile (6.5% error) and flat "top-hat" profile (12.2% error).

Download Abstract