A Peridynamics-FEM Approach for Crack Path Prediction in Fiber-Reinforced Composites

Abstract

Peridynamics (PD) theory provides a means to investigate crack propagation behavior in solids, without the need for re-meshing the domain or other special treatments necessitated by conventional FEM approaches. As such, it has been increasing applied for modeling failure in different applications. In this work, we investigate crack propagation behavior in fiberreinforced composite (FRC) structures using a Peridynamics approach that builds on finite element method (FEM) solution kernels. The PD bonds representative of the composite are modeled as truss elements which takes into account the corresponding fiber/matrix micromechanics in definition of its constitutive behavior. The PD-FEM framework is implemented using a commercial FEA package ABAQUS, via custom defined application programming interface (API). Subsequently, we employ it to investigate crack propagation behavior in 3D composite structures subject to various types of loading conditions. Crack propagation simulations are performed on 3-D composites with fibers oriented in different directions per ply (unidirectional fiber lay-up within each ply) and multidirectional composite (woven, multiply). The developed framework is able to predict crack propagation behavior based on applied loading conditions and nature of pre-existing defects. A comparison between simulated results and available literature data is also undertaken.

Peridynamic computational homogenization theory for materials with evolving microstructure and damage

“…Peridynamics (PD) [1] has been used to solve a wide range of engineering problems since its introduction, and particularly, it is increasingly used to study the behaviour of composite materials [23][24][25][26][27][28][29][30][31][32]. When modelling composite materials with a complex microstructure resulting from manufacturing processes and oftentimes random distribution of constituent particles [33], it is a common practise to replace highly heterogeneous properties with effective material properties to avoid the costly computational requirement of numerical simulation that will directly resolve information from all relevant scales for the entire computational domain.…”