Posters

Abstract

As the number and size of wind farm clusters grow, so does the need to understand the full spectrum of airflow phenomena and their interaction with real atmospheric conditions. Whiffle has developed an operational Large Eddy Simulation (LES)-based weather forecasting model that runs on Graphics Processing Units (GPUs). Wind turbines are represented as actuator-disks, so global blockage and wake effects, and their interaction with real meteorological conditions, are inherently present in the model. Even with the GPU implementation the existing LES model is limited in terms of domain size, which makes incorporating the effects of mesoscale orography, land-sea interfaces or cluster effects challenging. The latest break-through in model developments contain major improvements in the boundary conditions of the LES. Previously, simulations had periodic boundary conditions and information from the large-scale weather model was supplied as dynamic tendencies for the state-variables. This setting has now been replaced for a configuration where a mesoscale model is run in which the LES domain is nested. The implementation adds a new scale of flow phenomena to the LES simulation altogether; the mesoscale phenomena. Events like land-sea breezes and gap winds are too small to be captured in a 30-kilometer-resolution reanalysis model and too large or too far away to be included in the LES domain. The mesoscale model has boundary conditions from the large-scale weather model and runs concurrent with the LES, whilst providing it with inflow conditions via a meso-to-microscale coupling. The mesoscale model is, in fact, the same model as the GPU-resident LES but with a turbulence parametrization rather than a sub-grid scheme. Contrary to the traditional LES model, which operates for spatial resolutions between 10 and 100 meters, the model in meso-mode typically performs well for resolutions up to 2 kilometers. These features allow for effective nesting of multiple meso- and/or LES domains, depending on the terrain complexity and use case. Expanding from LES into the meso-domain is a big step towards mapping the complete spectrum of meteorological conditions and events onto relevant wind farm performance metrics, in one single model run. Complex sites like land-sea transitions, extremely non-homogeneous orography, or a combination of those, no longer require artificial configurations with periodic LES but have now become more straightforward while still computationally feasible. This presentation explains the details of the model development and the results in terms of application to onshore and offshore sites.