Abstract
In Germany, the expansion of the wind energy in accordance with the species protection can
only push the goals of German Energy Transition. In the present scenario, technical solutions
(camera and radar) does not have sufficient scientific database on assessment of reduction of
risk of collision of endangered species in German wind parks. Therefore, wind industry faces
operational restrictions on existing wind turbines as well as reduced access to onshore wind
potential area.
In beginning of 2018, WestfalenWind GmbH and Lackmann Phymetric GmbH took the initiative
to test and evaluate SafeWind - a camera-based automated system by Biodiv-Wind SAS,
France in their wind parks. At Lichtenau Hassel Wind Park, three SafeWind systems are
installed at location (WEA 9, WEA 11 and WEA 13) and have been in operation since
August 16, 2018. The SafeWind system works on principle of real time detection of flying object
within intrusion area and triggering of regulation control of wind turbine, when the target is
within the threshold limit. The aim of the test and evaluation of the SafeWind system was to
create the decisive scientific database for its acceptance as an adequate mean to reduce the
risk of collision of bird species like – red kite and black stork.
For uniformity in interpretation of the results, the evaluation protocol was adapted as per
guideline published by KNE. In 2019, Dr. Karl Heinz Loske conducted the field study (between
March and October) at Lichtenau Hassel Wind Park as an independent expert, Ph.D. Ecology,
and submitted the report. The report from the expert is used as an input for the evaluation of
the SafeWind system. Due to practical constraint on the field observation, only two wind
turbines, WEA 11 and WEA 13 are considered for the study. The limiting parameters for the
detection capacity (i.e. detection range) and regulation control (i.e. reaction range) were
determined through the repeated Robird drone experiment. The dataset for the estimation of
species-specific acquisition rate and collision risk probability were generated by applying the
limiting parameters to the field observation data (LRF data) from Dr. Loske. In the dataset, it
was assumed to consider the data between the installation height of camera and tip of the rotor
blade (12 o’clock position) of respective wind turbines.
The detection range and the reaction range around the wind turbine were evaluated as 270 m
and 172 m for red kite, and 337 m and 215 m for black stork from the Robird experiment. The
acquisition rate is percentage of flying tracks detected by the SafeWind system out of all the
flying tracks within detection range of respective species from the LRF data. At the rotor altitude
of the wind turbine within the detection range, the acquisition rate for red kite is maximum, 84%
at WEA 11 and 93% at WEA 13. In case of black stork, it is 100% only with one wind turbine
(WEA 11) but may not be meaningful due to less number of the observation points. Below the
Testing and evaluation of SafeWind – a bird protection system at Hassel Wind Park 3
rotor altitude of the wind turbine and within the detection range, the acquisition rate for red kite
is comparatively lower, 83% at WEA 11 and 76% at WEA 13. In general, the acquisition rates
were expected to reduce gradually outside the detection range. However, this reduction is
neither systematic nor homogeneous and linear. For example, the acquisition rate for red kite
is surprisingly increasing with the distance in case of WEA 11 and reach up to 88% in the
zone 2 (270 m – 300 m). The hypothesis is that the other factors than distance, altitude and
position of bird may influence the acquisition rate. While investigating the non-acquisitions, it
was observed that the topography and kind of background (vegetation, other wind turbines,
deep dark sky) around the wind turbine may influence the acquisition rate. The topography and
the background have a relation with the installation height of camera (In our case, the
installation height was selected due to the turbine-specific constraint). The site-specific
adjustment on the installation height of camera thus can improve the acquisition rate of the
system.
The collision risk probability is the measure of effectiveness of reaction of the SafeWind system
from the reaction range. A worst-case scenario, direct flights of each track of red kite at their
average speed from the reaction range distance towards the wind turbine rotor and their
inability to anticipate the moving object, were assumed for the calculation of the collision risk
probability. When the time taken to reach to the rotor is less than the time needed to bring the
rotor to the safe rotational speed (idle speed = 50 kmph of blade tip speed), respective tracks
are declared, hypothetically, as a collision with the rotor. Considering the reaction range
adopted for this experiment, the collision risk probability for red kite is determined as 9% at
WEA 11 and 12% at WEA 13. These probabilities of risk of collisions appear highly
overestimated. When the reduction factors like varying operational speeds of the wind turbine,
biological avoidance behaviour of red kite can eliminate the overestimated risk of collision.
Based on learning from the SafeWind recorded videos, the red kite has obvious anticipation
behaviour while passing through the rotor. For quantifying the anticipation behaviour, birdvehicle
collision studies for turkey vultures (DeVault, 2014) are referred which resulted in
reduction of probabilities to 3% at WEA 11 and 8% at WEA 13 due to anticipation behaviour
of birds of prey below 90 kmph speed of the moving object. Moreover, empirical data of Bird
Sentinel from Biodiv-Wind shows that collision events of red kite in France (n=7) and in
Germany (n=1) detected by the SafeWind up to now were only recorded when blade tip speed
was above 130 kmph. Considering this, the collision risk probability in our case is 0%. Indeed,
the collision of any bird species is reported neither during the study period nor from the day
the SafeWind system is in operation.
To sum-up, the use of the SafeWind system at wind turbines, which has the maximum
acquisition rates (84% at WEA 11 and 93% at WEA 13 in the critical risk of collision zone)
within the detection range and the low theoretical collision risk probability (9% at WEA 11 and 12% at WEA 13) within the reaction range for red kite, shows the significant reduction in the
risk of the collision. Despite the limited numbers of observations for black stork, the reduction
in the risk of collision is also applicable. Such system can also reduce the risk of collision for
other species like common buzzard, black kite and kestrel found around the study area. At last,
the outcome of the evaluation will enable the authorities to decide on accepting the SafeWind
system as an adequate mean to reduce the risk of collision of the impact sensitive bird species
from the wind turbine.