Currently, hundreds of bats are killed annually by wind turbines, many of them endangered. The spinning blades kill bats, as the bat is unable to notice the moving blade before it is too late. To help protect bats, wind turbines are shut down during the migration season and are limited in construction type, reducing power output. A study by the Department of Agriculture revealed that bats avoid areas of ambient acoustic noise. Previous experiments with ultrasonic pest deterrents on wind turbines have produced inconsistent results. The problem stems from the signal not being able to pass the sweep area of the wind turbine blades.

My idea was to attach an ultrasonic transducer to the blade which would theoretically let the sound be heard from further away. To do this I used a pair of earbuds attached to a blade to transmit a 12 kHz signal generated by my phone. Another phone was used at various positions 60 cm from the turbine to sample the sound at 44.1 kHz.

An analysis of the data recorded, by taking the first second of the last ten seconds and running it through a FFT analysis, reveals a 12 kHz peak. The peak size (strength of signal) varied depending on the position as well as the frequency. The variations may originate from the spinning blades, but its existence is conductive to deterring bats as a signal must be inconsisten to insure that bats don't become used to the sound.

Using a signal within the hearing range of bat would work based on the data collected. A real-life application would have to meet certain requirements: energy efficiency, being noninvasive to the life of people nearby, and a variance in signal and frequency to insure bats don't become used to the sound. There isn't a significant energy use by generating the signal in comparison to the energy produced by the turbine. Since bats hear in a range above humans, the proposed solution has no effect on people living nearby. The last point, the variance in signal would require further study on bats as well as the effect of spinning the sound signal. It is possible though to artificially induce the variance.