Exploring the link between microstructure statistics and transverse ply fracture in carbon/epoxy composites

Damage evolution within polymer matrix composites (PMCs) is difficult to characterize, as variability arises in the material microstructure from manufacturing and in the local strength of the constituents, which is inherently statistical. Transverse failure of unidirectional plies is particularly critical in composite laminates, as it is often a precursor for other, more catastrophic failure modes such as delamination and fiber breakage. Although it is possible to attempt to orient fibers primarily in the directional of external loads, avoiding transverse stress concentrations is unattainable, and this will often lead to failure within the ply. The possible interaction of failure mechanisms makes obtaining reliable strength predictions under general loading difficult. This project is dedicated to understanding how the underlying statistics affect the failure of unidirectional composites, and to develop tools to analyze real unidirectional composite microstructures. The first part of this study is focused on reconstructing micrographs of experimental composites and looking at various statistical metrics. We create a computational tool for predicting initial debonding sites based on geometry alone. From there, we create “virtual” microstructures that mimic some of the statistics that describe the real microstructure. We then perform some mesoscale simulations that focus on the cohesive failure of the interfaces between the fibers and the surrounding matrix material. Finally, an analytical material sensitivity formulation is derived and implemented using the direct differentiation method implemented in a C++ Interface-enriched Generalized Finite Element Method (IGFEM) framework. Emphasis is placed on extracting the sensitivity of the transverse failure response with respect to the two material parameters that characterize the interfacial cohesive failure: a critical stress (σ_c) and a critical opening displacement (δ_c)