Posters

Abstract

Horizontal extrapolation of on-site measurements across onshore sites is traditionally performed by running linear flow models or RANS-based models for a predefined set of (sectorial) flow cases. Although extensively used and validated, these methods break down in case of complex sites (linear flow models) or require site specific tuning and expert knowledge to operate (RANS).   Further, by construction, these methods cannot account for the diurnal and seasonal variability of the wind and atmospheric stability conditions on site. Also, as wind farm sites are becoming increasingly large, the meso-scale heterogeneity in the background wind conditions cannot be neglected anymore.  Meso-scale models provide a physically sound way of modelling diurnal and seasonal variations and larger wind gradients, but lack the resolution for capturing the impact of micro-scale orography and land-use heterogeneity. To overcome these shortcomings, we show how a large eddy simulation (LES) modelling framework that incorporates both meso-scale weather conditions and micro-scale details of the terrain and land-use can be leveraged to create accurate and long-term representative wind resource maps. Simulations were performed across various wind farm sites across the world and the accuracy of the model was assessed by comparing the observed and modelled wind speed ratios between several masts over the site. Accurately predicting the ratios, or speed-ups, over the site is a crucial step in the industry’s current practices that still rely heavily on on-site measurements. Additionally, it is shown that the diurnal, seasonal and sectorial dependence of the wind speed ratio is accurately captured by the simulations, as a result of the full-physics modelling approach.  Finally, we show how with a smart selection of a limited number of days, the simulated wind speed ratios can be made representative of the long-term wind climate for a fraction of the computational cost of simulating the full time period.