Abstract
Bird and bat mortalities caused by interactions with wind turbines is a critical concern that requires addressing for conservation purposes. Deploying a low cost sensor array will be instrumental during site permitting, conducting impact assessments of existing wind farms, and assessing efficacy of wildlife mortality mitigation or wildlife deterrent technologies. While carcass surveys are the standard method for measuring wildlife mortality for onshore sites, the method is inadequate due to factors such as carcass removal. For offshore wind turbines, there is no industry adopted method for evaluating wildlife mortality. A near-real-time detection system could quantify wildlife interaction rates of both onshore and offshore wind facilities. This US Department of Energy funded project covers the development and testing of a multi-sensor instrumentation package capable of detecting avian and bat interactions with the blades, nacelle, and tower of a wind turbine. The onboard, integrated sensor package includes contact microphones, accelerometers, visual and infrared spectrum cameras as well as bioacoustic monitoring. Infrared or visual image recording are necessary for event confirmation and taxonomic classification. Simulated impacts using tennis balls were successfully recorded in tests on the wind turbines at the North American Wind Research and Training Center, Mesalands Community College, New Mexico and the National Wind Technology Center, National Renewable Energy Lab, Colorado, proving the system's operability. Accelerometers were shown to be the more reliable sensor while contact microphones were shown to be the more sensitive sensor. Results also revealed the requirement of mounting both sensors on each blade for reliable detection. A 1296 x 972 pixel resolution was recognized as an acceptable camera setting for the focal length scale to perform species identification on a medium sized sea bird. Acceptable camera positions were found on the nacelle and on the tower near the ground with both looking at the lower blade sweep area. A custom computer was assembled to handle the network data. The data volume requiring manual review was reduced by incorporating event-based triggering ring buffers into the system's software structure. The system will be capable of long term, unattended deployment by improving the automatic event detection algorithm and robustness of the system's software architecture.