Progressive damage and failure response of hybrid 3D textile composites subjected to flexural loading, part II: Mechanics based multiscale computational modeling of progressive damage and failure

A mechanics based multiscale computational model is presented to predict the deformation, damage and failure response of hybrid 3D textile composites (H3DTCs) subjected to three-point bending. The geometry of the textile architecture was incorporated in a mesoscale finite element (FE) model, while the H3DTC was homogenized at the macroscale. The mesoscale model is a collection of repeat unit cells (RUCs) that are composed of different types of fiber tows embedded in a surrounding matrix. Matrix microdamage was modeled by a (pre-peak) nonlinear stress versus strain response, using a modified 12 deformation theory of plasticity incorporating a secant-modulus approach. Fiber tow pre-peak nonlinear response was computed using a novel, two-scale model, in which the subscale micromechanical analysis was carried out in closed-form based upon a unit cell of a fiber-matrix concentric cylinder. Consequently, the influence of matrix microdamage developing at the microscale manifests as the progressive degradation of fiber tow stiffness at the mesoscale. The smeared crack approach (SCA) was employed to model the post-peak softening of the constituents due to failure, including matrix macro-cracking, tow kinking, and tow breaking. This method offers a mesh objective result by relating the post-peak softening response to a traction-separation law that is associated with each failure mechanism through a characteristic length. Thus, the total energy release rate during failure in a continuum element is related to the fracture toughness of the material. The load-deflection responses, along with the progressive damage and failure events, including fiber tow kinking and rupture, are successfully predicted through the proposed computational model. In addition, the textile architecture-dependent effect, observed in the asymmetric H3DTC5, is also captured, demonstrating the predictive capability of the proposed modeling scheme. Since all the inputs are from the constituent level, the model is useful in understanding how the macroscopic response of H3DTC5 is influenced by textile architecture and constituent properties. The experimental studies are presented in Part I of this two-part sequence (Zhang et al., 2015). (C) 2015 Elsevier Ltd. All rights reserved.