Computational fluid dynamics simulations of wind farms often use Reynolds-Averaged Navier Stokes (RANS) models or large eddy simulations (LES). RANS models do not accurately capture effects such as wake-atmospheric boundary layer (ABL) interactions, blade boundary layer dynamics and turbine-turbine interactions. LES can theoretically capture these effects but often does not because of the computational cost. This leads to errors in wind-farm layout and control, wind farm financing and electricity grid development. The ExaWind project at the National Renewable Energy Laboratory aims to use the newly developed hybrid RANS-LES model, Active Model Split (AMS) and exascale computing to capture often-neglected wind-farm dynamics. This work modifies AMS, which originally used the k-epsilon RANS model, to use the SST k-omega RANS model and simulate an ABL for the first time. Using the SST k-omega model avoids the k-epsilon model's need for a damping function near walls and its inaccuracy for adverse pressure gradients and the k-omega RANS model's sensitivity to freestream velocity. The ExaWind project will use the model developed in this work, AMS with SST k-omega for the ABL, to simulate a set of wind turbines in a realistic ABL while resolving the blade boundary layer.
University
Stanford University