A nonstationary inflow turbulence generator driven by divergence modified stochastic wave-based spectral representation method

Accurate generation of nonstationary turbulence inflow is essential for refined analyses of structural wind effects under exceptional wind events. The stochastic wave-based spectral representation method (SWSRM) is widely used for multi-dimensional wind field simulation. However, its application to large eddy simulation (LES) remains challenged by the continuity condition and the efficient generation of turbulence on large-scale non-uniform inlet meshes. To address these limitations, a divergence-free non-uniform fast Fourier transform (NUFFT)-enhanced SWSRM is proposed for nonstationary turbulence inflow generation in LES. The framework incorporates a divergence-free modification to satisfy the flow continuity condition and combines the NUFFT algorithm to efficiently simulate turbulent wind velocities on non-uniform grid points. The proposed method is validated through LES of stationary turbulence in an empty atmospheric boundary layer domain and around the Commonwealth Advisory Aeronautical Research Council model. Furthermore, the proposed method is extended to efficiently generate nonstationary turbulent inflow. The results demonstrate that the divergence-free modification preserves the continuity condition near the inlet while maintaining the target turbulence statistics and spatial coherence characteristics. For both stationary and nonstationary inflow conditions, the simulated wind characteristics agree well with the prescribed target values. The turbulent power spectrum remains consistent with the target spectrum at low frequencies, whereas deviations gradually increase in the high-frequency range during downstream development due to LES filtering effects. Overall, the proposed method provides an efficient and physically consistent framework for high-fidelity analyses of structural wind effects under nonstationary wind conditions.