In the analysis and design of concrete structures, the assumption is often made that concrete is a homogeneous and isotropic material. Although these assumptions have suitable applications, consideration of local events, such as the fracture and failure of concrete, requires consideration of the random microstructure of concrete. This proposal outlines a research program that will link the microstructure and the mechanical behavior of concrete. The research will be developed along three lines: theoretical, experimental, and stochastic. The basis of the theoretical analysis is the moving-window generalized method of cells (GMC) composite model, which allows for consideration of local failure initiation such as debonding between aggregates and hydrated cement paste. The GMC model, when used in conjunction with finite element models, will predict fields of local stress, strain and displacement that would develop in a given sample of concrete, either before or after the onset of damage, directly from digital microstructural images. For experimental aspects of the research, digital image correlation (DIC) will be employed to determine the displacement fields in concrete resulting from mechanical loads. DIC uses correlation between pixel alignments in undeformed and deformed images of a specimen to determine surface displacement fields. The results of DIC will be used to calibrate the GMC model. A number of sto-chastic methods will enhance the experimental and theoretical techniques, including probabilistic characterization of local failure events and stochastic simulation of mechanical properties and local maximum stresses and strains in concrete. The final steps of the project will include extension of the models to three dimensions, and characterization of the reliability of concrete structures based on the results obtained from the analytical and experimental models of local damage.
