Micro-failure Analysis of Direct and Flat Loading Brazilian Tensile Tests

The difference between rock tensile strength measured by a direct tensile test and a Brazilian test is always attributed to external testing conditions and the natural heterogeneity of rock. There are few studies focusing on achieving a substantial understanding and quantifiable analysis of this difference. Methodologically, the relationship between the dominant frequency characteristics of acoustic emission and micro-failure has been preliminarily verified in the direct tensile test. To further explore the relationship between micro-failure types and dominant frequency characteristics and to quantitatively study the relationship between micro-failure and macro-mechanical behaviour, direct tensile tests and Brazilian disc tests were carried out to determine rock tensile strength. Furthermore, the acoustic emission technique was used to monitor the micro-failure of rock during the loading process. Detailed temporal and spatial distribution, as well as spectrum analyses of the acoustic emission signals, were conducted. Based on the statistical analysis of dominant frequency of AE signals, comparisons were made in rock tensile strength and micro-failure components from direct tensile tests and Brazilian tests. Then, a new parameter of strength factor was established to evaluate the contribution of micro-cracks produced by micro-tensile and micro-shear failures to the strength indirectly. The parameter also reflects the statistical relationship between micro-failure events and macro-failure process. By virtue of this parameter, it has been demonstrated that the difference between direct tensile strength and Brazilian splitting strength results from the higher proportion of micro-shear failure in the Brazilian test. Finally, the ideal tensile strength under the specific condition was obtained based on the statistical relationship between the strength and micro-failure component and suggested as a conservative design parameter for rock engineering.