Non-crimp fabrics geometrical variability and its influence on composites cure

A framework is developed and implemented to characterize and model in plane fibre misalignment in Non-Crimp fabrics (NCF). Image analysis based on fast Fourier Transform and correlation analysis is used to characterise in plane fibre misalignment in a carbon fibre ±45º NCF. It is found that fibre misalignment is significant showing high anisotropic spatial autocorrelation with the major direction of autocorrelation coinciding with the direction of the non-structural stitching of the fabric. The spatial autocorrelation structure of the fabric is modelled using an autoregressive two-dimensional stochastic process, the Ornstein- Uhlenbeck (OU) sheet. The resulting stochastic field is simulated using the Cholesky factorization, the spectral decomposition and the Karhunen-Loeve expansion (KLE). The three discretization techniques are compared in terms of accuracy and efficiency.