Efficient simulation of two-spatial dimensional turbulent wind fields based on the factorization of random functions

Refined modelling of turbulence in two-spatial dimensions is necessary for reliability assessment of large-size structures such as wind turbines. Stochastic wave-based spectral representation (SWSR) method is a widelyused tool. However, the correlation of turbulent components is usually neglected in it. Based on the SWSR, a multi-dimension (n-D) multi-variable (n-V) continuous temporal-spatial field model is thus proposed in this study. The low efficiency exist in this model, which are also unavoidable in others when simulating turbulence in multi-spatial dimensions. One reason is the large number of sample functions in the Monte Carlo (MC)-based method. The other is the n-D matrix-based Fast Fourier Transform (FFT) algorithm in single sample generation. Therefore, an efficient method is given to improve efficiency in both two aspects. The former improvement is realized through defining multi-index dimension reduction (DR) expressions in the proposed model. The latter is based on the 1-D vector-based FFT algorithm, the key to which are the decoupling of n-D matrices and factorization of multi-index DR expressions. A numerical simulation of a stochastic wind field in two-spatial dimensions is designed to demonstrate the validity of the proposed method. Results show its improvement of efficiency in both views of sample sets and single sample generation.