The Whiffle LES model

The lowest portion of the atmosphere, called the atmospheric boundary layer (ABL, it stretches to about 1 km height), is defined by its strong interaction with the Earth’s surface. Through friction, heat exchange, evaporation, and terrain effects, the air flow in the ABL is turbulent: it is filled with ever-changing vortices, whirls, and eddies. These structures range in size from millimeters up to kilometers, and the particularly large ones are often marked at their tops by clouds. At night, when the sun has set and the surface cools, the turbulence in the ABL often fades and makes way for stable and calm atmospheric conditions.

Two contrasting LES simulations of the atmospheric boundary layer. Left: convective conditions cause turbulence, and create cumulus cloud. Right: stable conditions cause a calm, stratified air flow. The colors at the bottom visualize the wind, the colors in the vertical cross-section show potential temperature.

The details of the air flow in the ABL matter, not only for our experience of the weather, but also to great extent for renewable energy production. The strength of the turbulence, for example, sets the importance of wake effects in wind farms; and naturally, the number and size of those clouds topping the largest eddies influence the production of a solar farm. Furthermore, wind turbines themselves can influence the wind and thereby affect the power production of surrounding turbines. In order to quantitatively assess the interaction between the atmosphere and (in the case of Whiffle Wind) wind turbines, we need a detailed model of the atmosphere.

Large-eddy simulation (LES) is such a model. Its foundation was laid in the 60’s and 70’s in the scientific meteorological community, and the technique has become the designated tool to research physical processes in the ABL. In essence, LES is a computational fluid dynamics model, that simulates the flow of air on a computational grid based on the laws of physics, where the grid boxes are about 10 m to 100 m in size. With that setup, it can explicitly simulate the turbulent structures that fill the ABL. Because of the grid spacing, however, it only simulates the ‘large’ eddies that are larger than 10 m to 100 m, hence the name. They contain the bulk of the total energy contained in the entire turbulence spectrum, and the effect of all smaller eddies is accounted for in a statistical way.

Over the years, the scientific community kept improving the LES models, in terms of both computational efficiency as well as physical accuracy. The LES model that Whiffle uses has its origins in the Dutch Atmospheric Large-Eddy Simulation (DALES), which was developed by a consortium of Dutch university research groups. The main features that define Whiffle’s LES are its functioning on graphical processing units (GPUs), which makes it fast; and its coupling to weather data, which gives the LES realistic boundary conditions. We also call this setup ‘real-weather LES’.

Whiffle Wind runs the LES for a specified domain, takes boundary conditions from the ERA5 historical weather data, and can include turbines, which influence the flow. The result is a detailed simulation of the atmospheric conditions during the run time, at a resolution of 100 m, including power production values of individual turbines.