Abstract
Understanding the failure mechanism of the rock mass under the general stress state is of great importance for the safe constructions of the underground engineering. Here, a series of true triaxial fracture tests on the intact and pre-flawed sandstones are conducted. The failure modes of the sandstones are analyzed, and the multi-scale fracture characteristics of the basic types of cracks are identified. Moreover, a multi-scale observation and crack statistics based method for analyzing the failure mechanism of the rock is proposed, and the influences of the stress state and the pre-existing flaw on the rock failure mechanism are investigated. The results indicate that the rock failure mode is controlled by the true triaxial stress state and the pre-existing flaw. The crack quantities in the pre-flawed rocks are nearly always more than those in the intact rock. which indicates that pre-existing flaw has a significant promoting effect on the crack initiation and development. 7 types of basic crack in the rock failure modes are summarized. Based on the multi-scale fracture characteristics, the fracture mechanisms of the basic types of cracks are identified, and the fracture mechanisms of the seven basic type cracks are identified and divided into four categories. The quantity statistics of the cracks corresponding to different fracture mechanisms show that the rise of σ3 can significantly reduce the percentage of the shear cracks, while the rise of σ2 conduces to the increase of the percentage of the tensile crack. The pre-existing flaw has a promoting effect on the initiation of the tensile crack, however, the true triaxial stress is the decisive factor controlling the rock failure mechanism. In the discussion, the size effect of rock fracture and the correlation between true triaxial test and engineering application are analyzed. This work contributes to an improved understanding of the failure mechanism of rock and a potential means by which to guide the design and construction of underground engineering.