Detection and segmentation of mechanical damage in concrete with X-Ray microtomography

Abstract

X-Ray computed microtomography (micro-CT) is an imaging technique that has seen increasing application in the field of experimental mechanics for its ability to take non-destructive, precise, 3D measurements of material morphology. The microstructure of a heterogeneous material with distinct constituent phases can be segregated using micro-CT as a result of the differences in X-Ray attenuation between the various phases. However, detection and segmentation of different features within the same phase requires more complex methods beyond the segmentation by X-Ray attenuation rate. For example, cracks and porosity are recognized as the same phase by X-Ray attenuation-based segmentation methods. However, these two microstructural features are the result of different processes, as pores are inherent to the microstructure of cementitious materials, while cracks are often considered to be material damage. This study aims to compare a novel quasi-3D crack detection technique with two other techniques reported in previously published studies, with the goal of finding the most effective technique for identifying cracks/damage in concrete specimens subjected to controlled intermittent dynamic mechanical loadings. The results show that while the previously reported techniques work well for detecting cracks in concrete specimens with discrete porosity and/or large interconnected crack networks, they were not as effective in distinguishing small cracks from clusters of voids. The quasi-3D geometric segmentation technique proposed in this study is shown to perform well across all crack length scales even in concrete specimens with interconnected porosity.