Numerical prediction of transverse cracking and delamination in fiber-reinforced laminates by using a two-scale cohesive finite element approach

Abstract

Transverse cracking is commonly experienced by fiber-reinforced composites subjected to multi-axial loading, especially for laminates with different ply orientations. This microscopic damage mechanism causes a decrease in their overall stiffness properties, and may promote delamination between adjacent plies, which is usually associated with their premature failure. Many works have clearly shown the role of interactions between microscopic and macroscopic damage sources in determining the overall strength properties of such materials, highlighting the need for fully detailed models. However, the huge computational cost of the related numerical simulations has motivated the introduction of different multiscale models for failure analysis of fiber-reinforced composites under transverse loading, using in combination with fracture models to represent fiber/matrix debonding and matrix cracking. Such models, including the popular computational homogenization-based and multilevel domain decomposition-based approaches, allow both damage percolation and boundary layer effects to be adequately captured, but are quite complicated to be implemented within the most common commercial finite element codes. In this work, a simpler although reliable two-scale approach is presented, able to predict in a very accurate and efficient way the transition from diffuse micro-cracking to localized transverse macro-cracks in fiber-reinforced composites. The proposed approach is based on a diffuse cohesive interface model, adopted to derive a microscopically based traction-separation law to be used within a purely homogenized nonlinear model for simulating matrix cracking. Numerical computations are performed with reference to small fiber-reinforced components subjected to complex loading paths. Finally, comparisons with reference solutions obtained via direct numerical simulations are presented to assess the validity of the proposed approach.