Presentations

Abstract

With increasing hub heights of new wind turbines, measurement heights required for wind resource assessment are increasing. Data availability from remote sensing instruments typically decrease with height, often not reaching 90% availability recommended by MEASNET. The impact of missing data on the accuracy of long-term extrapolated wind speed has been quantified at 29 locations in Norway, Sweden, and Finland. Two long term correction methods were used on 282 1-year timeseries with above 90 % data availability. 15 types of artificial data gaps were applied to each timeseries to quantify their impact on the error of the predicted long-term wind speed. 5 of the artificial data gaps simulate the data availability of a WindCube v2.1 lidar between 175 m and 300 m AGL. A comparison with 4 other long-term correction methods was done with a representative subset of data. The performance of the long-term correction methods was found to be good with all the lidar cases. Compared to the control long-term correction runs without artificial gaps (97 % data availability), the increase in the wind speed error is below 0.1 % for the simulated 225 m AGL data availability (69 %). For the simulated 300 m AGL availability (47 %), the increase in wind speed error is 0.35 %. Performance is also good with single months of missing data and data missing at regular intervals. Notable reduction in the accuracy of the long-term corrected wind speed can be seen with 3 or more months of consecutively missing data. For long-term correction, 90 % data availability threshold should not be used to discard measurement data, but instead to quantify the uncertainty of missing data. This in line with the new IEC 61400-15-2 standard which is under preparation.