Effects of sampling frequency on the proper orthogonal decomposition based reconstruction of a wind turbine wake

Abstract

The effects of sampling frequency on the efficacy of the POD technique for reconstructing the flow field of a wind turbine wake are studied in a neutral atmospheric boundary layer (ABL) using large eddy simulations (LES) coupled with an actuator disk model. The objective of this paper is to study how the sampling frequency over a fixed period of time affects the POD modes. The results showed that sampling frequency has little influence on the convergence of POD modes and the extraction of large-scale coherent structures. Meanwhile, sampling frequency has negligible influence on the kinetic energy contained in each mode of the first 25 modes. When the mode number is greater than 25, it affects the distributed energy in each mode and the accumulative energy in a given number of modes, and the accuracy of the reconstructed flow field. In present study, the flow field reconstructed using the smallest sampling frequency can be as accurate as that with the largest sampling frequency, while the former can save about 86% of the computational cost. For a given number of POD modes, a smaller sampling frequency may lead to more accurate flow field reconstruction. The sampling frequency mainly affects the small-scale structures.