| |
By Javier Villanueva, Wind Energy Department, National Renewable Energy Center of Spain (CENER), Pamplona, Spain
View the pdf of this Supplement
The site in northern Spain considered in the study
Harvesting wind from mountainous
regions is an appealing
option in many countries,
where flatter land is often put to
housing, industrial, or other community
use, and citizens prefer that
power generation facilities of any
kind not mar the nearby landscape
for most people on most days.
Several software programs are currently
available to assess the wind
power generation potential of a
given site, but they are not, as a
rule, well suited to steep terrain.
CFD is now emerging as an alternative
tool for wind source assessment
and forecasting, and its growing
popularity is due, in part, to its ability
to simulate wind conditions for
any type of terrain.
Velocity contours on a plane slicing through the highest
peak in the landscape show a variety of wind speeds near
ground level
Velocity contours are more uniform in regions where the
topographical variation is less
In a recent project carried out at
the National Renewable Energy
Center of Spain (CENER), a CFD
study of a region in Navarre, south
of Pamplona in the northern part of
the country, was performed. Using a
square expanse of land, 14 km on a
side, as a footprint, the atmosphere
was modeled to a height of 7 km.
Digitized contour data, with one
node every 50 m, was used to replicate
the surface of the land for the
CFD model. The quad paving
scheme in GAMBIT was used for the
surfaces, and a volume mesh of 1.5
million hexahedral elements was
built. Using FLUENT, customized
boundary conditions for velocity (a
logarithmic wind profile) and turbulent
kinetic energy and dissipation
were applied using user-defined
functions (UDFs). Through the use
of wall roughness factors characterized
by different length scales, different
vegetation patterns on the
land were simulated using the wall
function. Four meteorological masts
located at the site were used to collect
wind speed and direction data
at several heights for a period of one
full year. Pair-wise correlations from
two measurement stations were
used to validate the CFD predictions
for average wind speed and power
density. The results were also compared
to the predictions of one of
the simpler wind analysis programs
typically used in the industry.
The results showed that the average
absolute error between the CFD
predictions and measured values for
wind speed was 4.4%, while it was
6.6% for the simplified wind analysis
program. For the power density,
the average absolute errors were
4.9% and 13.4%, respectively, for
the CFD and wind analysis program
predictions. For both numerical
methods, the error was found to
increase with the spacing between
the stations used for the correlation.
This increase was greater for the
wind analysis program, however,
suggesting that as the distance
between meteorological masts
increases during site assessment
studies, it is more beneficial to use
CFD than a simplified code.
The methodology developed
can now be used with confidence to
evaluate the wind conditions in the
vicinity of each turbine position
planned when a wind farm is
designed. It is especially recommended
when accuracy in the predictions
is important (to minimize
the financial risk), when there are
few meteorological masts near the
planned turbines, and when the terrain
is complex. Models are currently
under development to test different
turbulence models and include
wake effects.
|
|
|
