2011 年 6 巻 3 号 p. 342-356
To obtain basic understanding for the development of atmospheric turbulence assessment technique for wind-energy application, we examine the performance of turbulence simulation using the large-eddy simulation technique, focusing on grid dependency of predicted turbulence statistics. We test two different types of model codes, one derived from a numerical weather prediction (NWP) model and the other from a computational fluid dynamics (CFD) model. Both model types have advantages and disadvantages while applied to the atmospheric boundary layer with complex terrain, and it is the purpose of this study to examine their capability for use in simulating wind-farm turbulence. The first simulation uses an NWP model for an ideal atmospheric flow and the other uses a CFD model for flows over complex surface. The horizontal grid spacing ranges from 50 m to 300 m. The results show that a horizontal grid spacing of 50 m for both model types can reasonably capture the energy containing eddies and represent coherence structures and turbulence statistics, such as intensity, anisotropy and spectra of wind fluctuations. This study provides a guideline for using numerical simulations for turbulence assessment at wind turbine locations. It also suggests that the combination of NWP and CFD models may provide a better approach to assess atmospheric turbulence, for example by using an NWP model with fine grids to provide turbulent inflow boundary conditions for a CFD model.