A probabilistic model of acoustic emissions generated during compression test of cementitious materials for crack mode classification

This article presents a study on crack mode classification in cementitious materials under uniaxial compression using gaussian mixture modeling (GMM) of acoustic emissions approach. To implement a retrofitting method to an in-service concrete structure, a prior knowledge about the type of crack developed in the concrete structure is useful. Because, occurrence of AE events during fracture process in solids is random, a probabilsitc method has been required to classify the AE sources related to different types of cracks. In this study, a monotonically increasing unconfined uniaxial compressive load has been applied on different cylindrical specimens of plain cement concrete cast with maximum coarse aggregate size of 20 mm, 12.5 mm and cement mortar to study crack classification. It has been observed that the slope of the line separating the AE data clusters belonging to tensile and shear cracks is more steep for the concrete specimen containing 20 mm maximum coarse aggregate when compared to the concrete specimen containing 12.5 mm maximum coarse aggregate and cement mortar. This indicated that as the coarse aggregate size in concrete increased, the generated AE events related to shear cracking decreased. Also, the AE based b-value has reached minimum at the peak load. At the time of failure, AE related to shear cracking has increased sharply.