Wind resource and site assessment - Wind Power Generation and Wind Turbine Design - page 82

An optimised layout can avoid excessive gradients related to the fi rst three phe-
nomena by staying clear from steep slopes, obstacles and allow a suffi cient spacing
between the wind turbines.
The gradient is often determined by measurements. Care has to be taken when
the measurement height is lower than hub height since the wind shear exponent is
a function of height, thus a gradient measured at lower height is not representative
for higher heights. Be also aware of the fact that because the measured wind speed
differences can be relatively small, the resulting measurement uncertainties can
become quite signifi cant. The preferred method in these cases is to model the wind
speed profi le using fl ow models.
6 Site classifi cation
6.1 Introduction
When designing wind turbines a set of assumptions describing the wind climate
on-site is made. As the robustness of a wind turbine is directly related to the costs
of the machine, a system has become common of grouping sites into four different
categories to allow cost optimisation of the wind turbines. The site class depends
on the mean wind speed and extreme wind speed at hub height referred to as IEC
classes [25]. Furthermore three different turbulence classes have been introduced,
for low, medium and high turbulence sites. The term extreme wind or U ref is used
for the maximum 10-min average wind speed with a recurrence of 50 years at hub
height. Please note that U ref is not related to the mean wind speed.
6.2 Extreme winds
Selecting a suitable turbine for a site requires knowledge about the expected extreme
wind in the form of the maximum 10-min average wind speed at hub height with a
recurrence period of 50 years ( U ref according to the IEC 61400-1 [ 25 ]). While local
building codes frequently generalise the amplitude of extreme events across large
areas, on-site measured wind data allows for a more precise site-specifi c predic-
tion of the expected 50-year event. A number of methods offer the possibility to
estimate the on-site 50-year maximum 10-min wind speed from available shorter
term on-site data. The IEC 61400-1 does not prescribe any preferred method for
this purpose though.
Under certain assumptions, a Gumbel analysis [26] can be applied to predict the
50-year extreme wind speed on the basis of the measured on-site maximum wind
speeds. If these assumptions are correct, then the measured maximum wind speeds
should resemble a straight line in the Gumbel plot.
Different methodologies are available to extract extreme events from a
short-term data set and further to fi t the Gumbel distribution to these extremes.
The parameters describing the Gumbel distribution are generally determined
from the linear fi t in a Gumbel plot. Hereby the so-called reduced variant