Wind farm noise compliance assessments require measurement of the wind farm noise under full operating conditions. Normal wind farm operation typically includes periods where one or more wind turbines may be unable to operate, due to maintenance requirements, grid restrictions, or other factors. For a large wind farm with many wind turbines, or in situations where cumulative noise impacts from other wind farms need to be considered, there may be very few periods where all wind turbines at the wind farm, and any neighbouring wind farm, operate simultaneously. In such scenarios it is rarely practicable to simply take noise monitoring data from only the periods where all wind turbines were operating, as this may involve excluding a significant proportion of the noise monitoring data set, which would result in a considerable increase in the duration of noise monitoring that is necessary to obtain the required size of data set. To reduce the proportion of data that needs to be excluded, noise modelling can be used to determine whether the combination of wind turbines operating at any particular time would have resulted in a materially lower noise level than would have occurred with all wind turbines operating. Based on such modelling, non-representative noise measurement data can be excluded from the assessment. This paper discusses a data exclusion methodology that has previously been used for wind farm noise assessments in Australia, and investigates the effect that various modelling simplifications and exclusion criteria could have on the noise assessment outcome, based on a case study of noise monitoring data from 18 locations at two large Australian wind farms. For the case studies investigated it was found that the effect of excluding periods of noise data where there were non-operational wind turbines was minor, and the exclusion of such data could possibly be an unnecessary step in practical noise assessments for large wind farms, when using the assessment methodologies in use in Australia.
展开▼