Evolution of Modern Wind Turbines Part A: 1940 to 1994 - Wind Turbine Technology: Fundamental Concepts of Wind Turbine Engineering - page 137

a few local measuring locations. These computer models are used today to lay out wind
power stations, although their accuracy varies depending on the complexity of the terrain
(PNL 1980, 1981].
On an even iner scale, the character of both the wind inlow into a turbine rotor and
the wake behind it are critical. Knowledge about the wake is needed to determine effects
on downwind turbines in wind power stations, such as energy losses and increases in
structural loads, as a function of wind turbine spacing. The turbine wake structure is also
important in assisting analysis of the low pattern through the rotor. Improved mathematical
models of turbine wakes have resulted from wind tunnel test results and ield measurements.
The latter can be extremely demanding.
Knowledge of the characteristics of the wind inlow is also critical to understanding the
performance, dynamics, and structural loads imposed on the wind turbine. Early analyses
assumed a steady wind across the turbine rotor disk, allowing for wind shear (variation with
elevation) and tower shadow (a sector downwind of the tower with reduced speed). Such
inputs to structural analyses predicted mean loads fairly well, but badly underestimated
cyclic loads caused by wind turbulence . This in turn led to underestimation of fatigue
damage. Extensive research was undertaken in the 1980s to understand the expected
turbulence spectra under various climatic, terrain, and wind velocity conditions.
Vertical and horizontal planar arrays of anemometers (Fig. 3-25), in which rings of
anemometers are located upwind or downwind of the rotor swept area, have provided exper-
imental data from which the turbulence experienced by a blade element can be synthesized.
Whirling arm tests, utilizing instrumented booms rotating as if they were a turbine blade,
directly measured local wind velocities as an airfoil section would actually see them. Kites,
smoke tests, and sonic techniques have also been used. By 1986, empirical turbulence
models became available which signiicantly improved the ability to analyze rotor fatigue
loads through knowledge of wind inlow details smaller than the rotor itself.
Figure 3-25. A vertical plane array of anemometers near the 25-kW research HAWT
at the NREL test station. By sequentially sampling the data around this ring of anemo-
meters (a process called “rotational sampling”), inlow turbulence experienced by a moving
rotor blade can be measured. ( Courtesy of the National Renewable Energy Laboratory )