We present an application of data analytics and supervised machine learning to allow accurate predictions of the macroscopic stiffness and yield strength of a unidirectional composite loaded in the transverse plane. Predictions are obtained from the analysis of an image of the material microstructure, as well as knowledge of the constitutive models for fibres and matrix, without performing physically-based calculations. The computational framework is based on evaluating the 2-point correlation function of the images of 1800 microstructures, followed by dimensionality reduction via principal component analysis. Finite element (FE) simulations are performed on 1800 corresponding statistical volume elements (SVEs) representing cylindrical fibres in a continuous matrix, loaded in the transverse plane. A supervised machine learning (ML) exercise is performed, employing a gradient-boosted tree regression model with 10-fold cross-validation strategy. The model obtained is able to accurately predict the homogenized properties of arbitrary microstructures.

A cohesive-zone approach is used to study the interaction between an approaching crack and a particle embedded in a matrix material as a function of the mismatch in elastic and fracture properties. Crack-particle interaction is a crucial issue governing fracture behavior of particle-dispersed materials. Special attention is given in the present work to the effect of the mismatch in fracture properties, namely fracture strength and energy, which has not been fully-explored in the literature. Based on extensive finite element simulations using cohesive elements, the basic fracture mechanisms governing the crack-particle interaction are identified, namely particle fracture, crack deflection and interface debonding. The details of the cracking sequences are elucidated and the role of secondary cracks is highlighted. The effect of pre-existing flaws on the fracture behavior is analyzed both for flaws inside the particle as well as flaws on the particle/matrix interface. Several flaw configurations in terms of size, orientation and location are considered. In addition, the effect of the mismatch between the matrix and the interface fracture properties is also considered for a wide range of adhesive characteristics. The results of the simulations are summarized in the form of several fracture maps for different configurations, whereby the main fracture mechanisms are identified in regions inside a two-dimensional space of strength and toughness mismatch between the particle and the matrix. It is observed that the mismatch in the fracture properties usually plays a more dominant role on the crack trajectory than the mismatch in elastic properties in a particle-dispersed system. Pre-existing flaws/defects in the particle and the interface are found to be one of the principal controlling factors that alter the crack propagation characteristics. These results can be used as a guideline for designing particulate composite system with a preferred fracture mechanism, namely matrix cracking, interface debonding or particle fracture.

A cohesive zone-based constitutive model, originally developed to model fracture, is extended to include a healing variable to simulate crack healing processes and thus recovery of mechanical properties. The proposed cohesive relation is a composite-type material model that accounts for the properties of both the original and the healing material, which are typically different. The constitutive model is designed to capture multiple healing events, which is relevant for self-healing materials that are capable of generating repeated healing. The model can be implemented in a finite element framework through the use of cohesive elements or the extended finite element method (XFEM). The resulting numerical framework is capable of modeling both extrinsic and intrinsic self-healing materials. Salient features of the model are demonstrated through various homogeneous deformations and healing processes followed by applications of the model to a self-healing material system based on embedded healing particles under nonhomogeneous deformations. It is shown that the model is suitable for analyzing and optimizing existing self-healing materials or for designing new self-healing materials with improved lifetime characteristics based on multiple healing events.