Anna Wellmann

wellmann(at)irmb.tu-bs.de

In Germany, wind power accounts for the largest share of renewably generated electricity [1]. In the future electricity generation using wind power is to be further expanded. The service life of wind turbines is a decisive factor for the sustainability and economic efficiency of electricity generation. Various (wind) loads, which depend on the different flow situations of the wind turbines, can strongly affect their lifetime. Currently used simulations of flow fields around wind turbines focus primarily on the analysis of large-scale flow structures. In this context, wind turbines have only been considered in a highly simplified way, using methods such as the Actuator Line Method (ALM) [2]. In this context, wind turbines have mostly been studied using simplistic methods such as the Actuator Line Method (ALM). These simulations are only of limited use for a detailed estimation of the loads acting on the components of a single wind turbine.

In this research project, high-resolution simulations of the flow on the components of a wind turbine are performed using the Lattice Boltzmann Method (LBM). The LBM is based on a discretization of the flow domain using a grid. The simulation of rotating objects, such as the rotor of a wind turbine, requires an adjustment of the grid during the simulation. A promising approach is the simulation of the rotor by means of a grid which moves synchronously to the rotor [3]. The surrounding area is represented by a static grid. A particular challenge of this approach is the robust and accurate numerical coupling of the moving grid with the static grid.

The explicit nature and data locality of LBM favor highly efficient massive parallelization using graphics processing units (GPUs). In this research project, the GPU version of the VirtualFluids software package is used. High-resolution simulations of large flow domains require memory and computational resources beyond the capacity of a single GPU. For this reason, high-performance computers with multiple GPUs are used in this research project. Thus, another focus is on optimizing the scaling properties of VirtualFluids for these computer architectures.

With the help of the mentioned developments, individual wind turbines will be simulated in high resolution and the loads acting on wind turbines due to the surrounding air flow will be analyzed. A more profound knowledge of the loads acting on the wind turbines can contribute to a better comprehension of the aging processes and facilitate a subsequent optimization of the service life of wind turbine.