The
large-scale implementation of wind energy is hampered to a large extent
by the unknown effects that wind turbines may have on the environment.
The collision of birds with turbines and the distortion of the migration
routes are in that respect points of great concern. The project ?WT-Bird , A low cost solution for
detecting bird collision? focused on developing a reliable and cheaper
method for counting bird collisions in order to obtain more insight
in the actual size of the problem, especially in offshore applications
for which no alternative detection method is available.
The report describes the work that has been performed to meet
the objective ?developing and demonstrating a system that determines
a bird collision against a wind turbine and with which it is possible
to determine the bird species?. The system had to meet among others
the following specifications:
·
the
system should be low cost in order to be competitive with manual counting
methods;
·
analysis
of recorded data by e.g. ornithologist should not be labour intensive,
meaning that only actual collisions should be recorded and no false
data should be stored;
·
the
system should be able to operate under all weather and visibility conditions;
·
the
system should operate in offshore wind farms for long periods in a reliable
manner and data should be accessible remotely.
The
assembly of the prototypes has been carried out in this project successfully.
They behaved robust and reliable during the field tests and only minor
problems have been identified. Two major problems have not been solved
completely in this project.
1.
Triggering
Birds can collide against rotors and towers in many different
ways. The tests that have been performed with shooting tennis balls
against the flat side of rotor blades and throwing sand bags against
the tower did not cover the entire range of possible bird impacts. Secondly,
the variety of possible microphone configurations and background noises
for especially larger turbines is bigger than originally expected. The
determination of a good algorithm for triggering should be further investigated.
Within the constraints of the project, more detailed investigations
were not feasible.
2.
Image recording and recognition
of bird
species
Already at an early stage in the project it was decided to
use the Mobotix camera a.o. because it had a normal lens and a low-light
lens, it could be accessed with commonly used web browser technology,
and especially because it was cheap. During the field tests, the shortcomings
of the camera became clear. The view during night-time was insufficiently
clear for recognition of species. Additional light sources were necessary
and the exposure time is too long, leading to too low sample frequencies.
A large part of the project has been spent on improving the visibility
during night-time but without being completely successful.
To
solve these problems, further R&D work is recommended. More tests
with dummies against larger turbines should be performed and the tests
with cameras should be continued. Moreover, much more long-term field
tests should be performed in close collaboration with ornithologists.
Terug naar overzicht.