Presentations

Abstract

Wind projects rely on accurate and reliable wind resource data but they can be unavoidably limited in time and space. Extrapolating them to the long-term by combining with wind modelling results is one of our main aims. In this context, the presentation introduces a methodology for turbulence calibration in long-term high-resolution synthetic time series. Although models can describe site characteristics in terms of trends and yearly fluctuations, their accuracy can be significantly enhanced by calibration with real-world measurements. This process integrates actual data structures into the models, leading to long-term turbulence time series that can be used for event classification and the description of extreme episodes. Nevertheless, turbulence calibration has not been widely implemented in the wind industry, primarily due to the scarcity of long-term turbulence reference data. This situation presents a significant challenge, as we embark on what is, to our knowledge, the first industry-scale attempt to address this gap. Selecting high-quality, long-term reference data is essential for an effective calibration. To this end, the Weather Research and Forecasting (WRF) model is used for the downscaling of ERA5 data up to the microscale, ensuring that our data not only covers the standard atmospheric variables but also provides a comprehensive and well-grounded turbulence description. This dataset surpasses standard long-term references such as reanalysis data and mesoscale models, especially in capturing local flow characteristics such as wind speed and turbulence. Our approach aims to further refine turbulence by incorporating insights from actual measurements. Utilising a combination of statistical and machine learning techniques, our method offers a cost-effective alternative to accurately depict long-term turbulence characteristics. The calibration involves improving the modelled wind speed and its standard deviation, trying to isolate systematic biases on the modelled turbulence. These biases can vary across different scenarios, including seasonal variability and shifts in sectors of wind direction. Furthermore, the calibration process also considers the additional challenge of ensuring that the resulting turbulence is consistent with physical patterns observed in nature. Our presentation will outline the calibration methodology, detailing the most relevant steps of our approach. Special emphasis will be placed on offshore locations, which are not only prone to significant inaccuracies in modelled turbulence but also often suffer from a lack of measurements. Finally, through extensive validation at over 50 sites worldwide, we demonstrate the robustness and reliability of our methodology for turbulence description in wind resource assessment.