Acoplamento entre modelos de otimização geométrica de sistemas particulados e a micromecânica dos campos médios para análise multiescala de compósitos cimentícios multifásicos

Abstract

On a macro scale, concrete can be considered a homogeneous material and has even been tested as such. However, when its scale is reduced, it exhibits inherent phases at every level. Modeling concrete has always been a major challenge due to the complexity of this cementitious material, yet proposals such as homogenization by micromechanics are always made. Mean-field micromechanics is based on the homogenization of two-phase composites consisting of inclusions immersed in an infinite matrix. At the macroscopic level, concrete can be understood simply as a composition of inclusions (coarse aggregates) in a matrix (mortar) and is often modeled in this macro configuration. However, it is known that there are other phases that need to be evaluated to obtain a result closer to laboratory tests, such as the interfacial transition zone. In addition to modeling, it can be said that the construction industry is trying to maximize the properties of concrete, with an initial idea to seek the packaging of particle systems (inclusions) that minimizes the dispersed phase (matrix). Given this open field, the present work develops a multiscale methodology that evaluates the accuracy of biphasic models and multiphase models with experimental tests from the literature. In order to maximize the mechanical properties of concrete, a methodology is developed that combines optimization models of particle systems with models of homogenization of composite materials. The entire above strategy has been built on an object-oriented framework consisting of a set of classes designed to solve decoupled problems: homogenization and packaging, as well as problems involving the two common methods. The developed framework considers design patterns and an architecture that allows diversification of the problems without changing the base coding, using the paradigms of inheritance, encapsulation and polymorphism included in this type of programming architecture. The results obtained with the multiscale modeling and the coupling between the techniques confirm the maximization of the mechanical properties in the optimization of the particle systems.