A dynamically adaptive lattice Boltzmann method for predicting turbulent wake fields in wind parks

Abstract

Wind turbines create large-scale wake structures that can affect downstream turbines considerably. Numerical simulation of the turbulent flow field is a viable approach in order to obtain a better understanding of these interactions and to optimize the turbine placement in wind parks. Yet, the development of effective computational methods for predictive wind farm simulation is challenging. Sufficiently resolved computations of the Navier-Stokes equations on moving meshes can become prohibitively expensive, disk or line actuator models simplify the fluid-structure interaction drastically, and most vortex methods struggle representing wake interaction phenomena correctly. As an alternative approach, we are currently developing an adaptive lattice Boltzmann method (LBM) for turbulent weakly compressible flows with embedded moving structures that shows good potential for effective wind turbine wake prediction. The LBM is a very inexpensive time-explicit scheme that is known to add comparably little numerical dissipation to computational results and is thereby well suited for (very) large eddy simulation. Since the method is formulated in an Eulerian frame of reference and on a Cartesian grid, even moving boundaries can be considered rather easily. However, in order to mitigate the inherent error due to limited resolution along non-Cartesian structures, the application of local mesh adaptation is of crucial importance. In implementing a dynamically adaptive LBM with moving boundaries, we rely on our software system AMROC that was originally developed to supplement shock-capturing finite volume methods with mesh adaptation and level-set-based embedded boundary techniques. The presentation will describe all crucial components of the numerical method and discuss first test computations. Among other configurations, simulations of the wake fields created by multiple Vesta V27 turbines will be shown.