Multiscale computational modeling of damage evolution in viscoelastic particulate composites with growing cracks.

摘要

Heterogeneous viscoelastic solids are a complex class of materials and accurate prediction of their mechanical response remains a challenge. An even more challenging task is the prediction of failure in structural components made from this class of materials. One of the primary recognized failure modes for heterogeneous solids is the development of new internal boundaries in the form of cracks. In this form of failure, multiple cracks on widely varying length scales can interact to produce sufficient energy dissipation to cause total destruction of the component. Furthermore, components that possess multiple length scales happen often in nature, such as composite materials used in the aircraft industry, geologic media, and asphaltic roadways. This suggests the need for improved models besides costly experimentally based design procedures, making it propitious to utilize multiscale algorithms.;This dissertation presents a multiscale computational model for predicting damage evolution in viscoelastic particulate composites. The model is based on continuum mechanics and is implemented into a time-marching multiscale finite element formulation that employs a viscoelastic cohesive zone model to predict rate-dependent damage evolution in the form of hundreds of cracks. The algorithm solves problems on three simultaneous length scales: microcrack (micro-scale); microstructure (local-scale) and the structural component (global-scale). This approach models widely different length scales using separate but two-way coupled algorithms, thus obviating the necessity to model every crack/heterogeneity at the global scale. The solution for each length scale is linked to the neighboring length scale by a homogenization theorem.;Therefore, models that can accurately predict the evolution of damage and the ultimate failure event, though complex, would appear to be useful for design purposes. This dissertation includes a detailed description of the methodology followed by a few example problems to illustrate the approach. The model is also applied to predict life of asphaltic pavements, subjected to different sets of design variables, such as aggregate volume fractions and shape, asphalt layer thickness and truck loads.