An Experimental-Computational Correlated Study for Describing the Failure Characteristics of Concrete across Two Scale Levels: Mixture and Structural Component

Abstract

The failure characteristics of concrete, which are often brittle, are complex due variability in mix design, heterogeneity of the final man-made composite product, and the complexity associated with describing the corresponding mechanical response across different scale levels. Numerous experimental methods as well as numerical models have been developed to characterize the mechanical behavior of cementitious composites, but the major of these methods have focused on describing bulk response and are not well suited to characterize localized phenomena. Recent advances in the areas of multi-scale modeling and computational mechanics have shown promise for improving current capabilities, but these approaches also require experimental validation. This manuscript explores the extension of Digital Image Correlation (DIC) to fully characterize the behavior of concrete across different structural scales. The investigation leverages results from an experimental testing program at both mixture and structural member scale levels to evaluate the performance of two representative plasticity-based numerical models commonly used to describe the failure characteristics of concrete subjected to various states of stresses. The experimental study consisted of a series of compression, split tensile, and flexural tests. For the numerical models, the finite element method (FEM) was used to simulate concrete specimens at different scale levels. A comparison of the experimental and numerical results demonstrated that the numerical models are capable in predicting the ultimate capacities and global responses of the tested specimens. The minimum discrepancy between the results was observed in the pure compression tests, with less agreement observed in the presence of tensile stresses (i.e. split tensile and flexural tests at both scale levels). This can be attributed to the limitations of the selected material models in describing the tensile behavior of concrete beyond the elastic limit as well as the current shortcomings associated with numerical analyzes and their capabilities in describing the localized behavioral features such as crack initiation and propagation. Results from this investigation highlight the potential of DIC as a non-contact measurement technology to improve the performance of existing material models for traditional civil engineering materials, but also underscores its capabilities in development of new constitutive models for the next generation of innovative high performance materials.