A Comparison of Tensile Testing Techniques and Predictive Correlations

Abstract The engineering properties of a rock foundation are key parameters in the analysis of dams as they inform an understanding of how a structure will perform under a loading condition of interest. One important, but difficult to accurately evaluate, engineering property is the tensile strength of rock. As the direct tensile strength (DTS) testing of intact rock is non-trivial and expensive, the tensile strength is typically inferred from more common indirect methods, such as the Brazilian tensile strength (BTS) test, akin to the splitting tensile strength test performed on concrete, or using correlations to other mechanical properties, such as the crack initiation (CI) strength, m i parameter from the Generalized Hoek-Brown failure criterion, or the Mode I fracture toughness. This research leverages a robust laboratory testing program utilizing a state-of-the-art digital image correlation (DIC) system to understand how the different platen shapes and the use of a soft cushion affect BTS results, improve the understanding about the relationship between DTS and BTS tests (with a specific focus on optimizing the test arrangement), and explores correlations between other laboratory-measured properties (e.g., CI and m i ) and the DTS. The DTS was difficult to measure directly, but was found to be closely related to the BTS when testing was performed using curved platens, with or without a cushion; this testing arrangement also resulted in a failure mechanism that aligns with the underlying assumptions from Hondros (1959); ISRM (1978); and ASTM (2016). When analyzing the BTS data from different platen arrangements, it is important to use a relationship that incorporates the loading arc angle to ensure data are comparable as the BTS is a function of the loading arc angle in the full stress solution. The use of CI or m i to estimate the DTS was found to be unreliable with the BTS providing the closest approximation