A NOVEL DISCRETE, MESOSCALE MODELING FRAMEWORK FOR THE SIMULATION OF THE DAMAGING AND FRACTURING BEHAVIOR OF COMPOSITES

Abstract

Unidirectional (UD) and 2D/3D textile composites are increasingly being employed in systems acrossed many industrial sectors including aerospace, automotive, and wind energy. This is due to the excellent specific mechanical properties and tailorability of composites, paving new avenues for structural optimization and weight savings. One of the challenges with the simulation of the mechanical response of composite materials is that their damage mechanisms depend strongly on the material micro- and mesostructures. Phenomena such as fiber micro-buckling and kinking in compression or fiber scissoring and matrix microcracking in shear in UD composites are only a few examples. Homogenized continuum models that describe these mechanisms are extremely mathematically complex, lack generality, and can only be used to fit experimental data. In fact, the constitutive equations compensate for not modeling fibers and matrix explicitly by introducing several complex equations and fitting parameters of unclear physical meaning. This makes model calibration extremely cumbersome and limits the predictive capability of the model. In reality, the modeling of damage and fracture in composite materials does not have to be complex if the physics of the micro-and mesostructures is simulated explicitly. This is the goal of the Discrete Model for Composites (DM4C), a novel discrete mesoscale modeling framework that simulates the mechanical behavior of UD and textile composites. Specifically, this framework is only based on physical laws and does not depend on element erosion to simulate fracture. As it will be shown in this presentation, in DM4C, fibers, groups of fibers, and tows are simulated explicitly as Timoshenko beam elements while the matrix is described by vectorial constitutive laws defined on the facets of a tetrahedral mesh anchored to the nodes of the beam elements. These vectorial laws describe both the elastic and inelastic behavior of the matrix, including the traction-separation laws governing the fracture process and the friction between facets governing the compressive behavior. Thanks to the facet-based formulation, fracture is modeled in a discrete way and the need for element erosion can be avoided. Furthermore, since fibers and matrix are now simulated explicitly, the constitutive laws of each material can be physics-based, simple, and with clearly defined material parameters. To demonstrate the predictive capability of the proposed framework, simulations of several typical damage mechanisms in composites will be compared to experimental data such as shear band formation in transverse compression, fiber micro-buckling and kinking in longitudinal compression, and sub-critical matrix microcraking in off-axis layers.