Rapidly predicting the effect of tool geometry on the wrinkling of biaxial NCFs during composites manufacturing using a deep learning surrogate model

A deep learning surrogate model is developed to rapidly predict the wrinkling patterns of a biaxial non-crimp fabric (NCF) layup for any given tool geometry during forming. The underlying dataset finite element simulations used investigate effect on severity. trained able make reliable predictions at very low computational cost, suitable design optimisation. Results indicate that certain geometrical features have greater impact than others. In particular, forming NCFs over geometries with draft angles tends result in smaller wrinkles.