A Modified Overstress Model to Simulate Dynamic Split Tensile Tests and Its Experimental Validation

Abstract

The tensile strength of rock and rock-like materials is universally acknowledged to be much lower than their compressive strength and shearing strength (Meyers and Chawla 1999). Furthermore, the stability and reliability of rock structures are correlated remarkably well with the dynamic tensile strength of rock materials. Therefore, dynamic split tensile tests under high strain rate loads have been an active area of investigation in the field of rock mechanics. Carneiro (1943) first used Brazilian disc specimens to determine the tensile strength of brittle materials. Wang et al. (2004) improved these experiments by using flattened Brazilian discs (FBDs) instead to avoid stress concentration at the loading end. Wang’s method ensures that the initial crack only occurs in the central part of a specimen, which is vital for the validation of a split test. Recently, the split Hopkinson pressure bar (SHPB) has come into use as a general apparatus to determine the dynamic tensile strength of rock-like materials (Saksala et al. 2013; Xu et al. 2014). However, among the failure processes, crack propagation is difficult to observe, which makes it very difficult to judge whether or not the initial crack starts from the center. Finite element methods (FEMs) are widely used to simulate failure in brittle materials (e.g., Hang et al. 2015; Saksala et al. 2015). Compared with laboratory experiments, it is much easier and cost-effective to observe crack propagation in a simulation as long as an appropriate material model is built for the specimens. In this article, we build 3D models of FBD specimens in an SHPB apparatus and use these models to conduct numerical simulations of rock dynamic split tests. A modified overstress constitutive model is applied to describe the rock-like material. The failure patterns resulting from dynamic split tensile tests, as well as crack initialization and propagation, are studied with these simulations. The simulation results are verified by laboratory experiments using the SHPB and a high-speed camera.