Are steady-state wake models and lookup tables sufficient to design profitable wake steering strategies? A Large Eddy Simulation investigation

Abstract

We leverage Large Eddy Simulation (LES) of wind farms to study the performances of an open-loop wake steering controller. This global controller is elaborated based on Lookup Table (LUT) generated using FLORIS. Our objectives are to highlight the potential problems that can arise from model errors due to its steady-state formulation and from difficulties to properly define the ambient flow state. We consider two wind farm configurations: a first theoretical layout of six machines, and a second layout representing a commercial wind farm that is currently studied for wake steering strategy, the King Plains wind farm. We represent our wind farm environment using LES within which the turbines are modeled as actuator disks. In addition to the standard local controller, the wind farm utilizes a LUT to compute the optimal yaw offset angles. Before its integration into the controller, this LUT is first populated using the yaw optimizer integrated into FLORIS. The primary input parameters for the LUT are derived from the freestream atmospheric conditions, which are parametrized in terms of TI, hub height wind speed, and wind direction. In this work, we investigate different strategies for estimating the ambient wind direction fed as input to the LUT during the control and their impact on the dynamics of the yaw actuation. While results confirm the potential of this type of controller, they also highlight the importance of a proper sensing to determine the freestream state.