Presentations

Abstract

Hectrometric-scale large-eddy simulation (LES) of atmospheric flow is becoming widely adopted in industry for wind resource assessment (WRA) and energy yield assessment (EYA). Especially in context of wind farm simulation, this approach is often met with reservations in the academic community due to the coarse horizontal resolutions (0.5-1 rotor diameters) applied. The validity of such resolutions for wind farm simulation is disregarded based on theoretical arguments and/or highly idealized or "golden day" simulations. In this work we argue that although these arguments are valid, their relevance in the simulation of wind farms in real world, atmospheric flows is limited.   More specifically, for a 1 GW wind farm, a set of 30 representative days has been simulated at resolutions between 133 and 25 m. Additional simulations with turbines for which the thrust was set to 0 allow for a comparison of free stream conditions and provide unwaked (or 'gross') production numbers. Results indicate that the impact of the grid spacing on the wind farm power production and wake losses is limited. Splitting out the wake losses as a function of wind farm wind speed and direction indicates that the coarser LES simulations follow the finest grid spacing of 25 m well, although for some wind speed and direction bins differences exist. Differences in downstream wind farm wake effects are small.   When zooming in from the wind farm to turbine level, the impact of grid-spacing becomes more apparent. Mainly in single-wake cases, it is apparent that coarser LES smears out the individual turbines wakes, leading to over-prediction of power production of the downstream turbine. In contrast, when a turbine is affected by the superposition of multiple upstream wakes, the difference between the different LES resolutions vanishes. Since the majority of the turbines is waked by multiple upstream turbines most of the time, the impact of these single wake cases on the overall production and wake loss numbers is limited.