Triaxial tests are conducted to determine the relationship between confinement and axial compressive strength. Depending on the confining stress applied to a core specimen, the failure process changes depending on the rock type’s internal composition (i.e. porosity, flaws, stiffness heterogeneity, grain shape, etc.). These failure process changes are not typically considered when planning a triaxial testing program or when processing triaxial test data. This paper summarises the changes in failure process that occur depending on the confining stress level for various rock types and outlines a procedure for processing triaxial data depending on the confining stress level for the determination of the Hoek-Brown strength envelop parameter mi and confidence intervals. For this, a triaxial and uniaxial dataset from Bingham Canyon is presented for a quartzite. The dataset is exceptionally complete in terms of the number of tests conducted (total of 217 uniaxial and triaxial tests) and the detail in test data quality and characterization of the specimens. The results show that triaxial data requires mi values outside (mi >50) the typically assumed range (mi =50). These high mi values appear to be needed to fit data at high confining stress levels (larger than about UCS/10). Based on the discussion in this paper; (1) the selection of confinement levels for testing purposes should include sufficient data in the confinement range of 0 to UCS/10 and UCS/10 to UCS/2; and (2) two sets of strength curves may need to be considered depending on the problem being assessed. One curve is valid for confinements of 0 to UCS/10 (representative of strengths near excavation boundaries) and the other for >UCS/10 to approximately the Mogi line (representative of strengths, for example, in wide pillars or mine abutments), after which a third envelop is needed. These changing envelop requirements are a result of changing failure mechanisms.


Recent developments in understanding brittle failure processes in hard brittle rocks (UCS>25MPa) has lead the authors to the realization that there are several deficiencies in how triaxial data is being treated to arrive at peak strength parameters. Furthermore, it is necessary to properly describe the uncertainty in test data. This paper aims at providing an understanding of how failure processes change as confinement increases, how this affects the failure envelop and thus how data should be interpreted to arrive at meaningful engineering parameters. Most importantly, as recently introduced by [1] it is necessary to differentiate strength envelops that are to be used for low confinement dominated problems (inner shell) and for those used to describe the behavior of rock at high confinement (outer shell).

The goals of this paper therefore are: (a) to review current practice for processing of triaxial data; (b) identify failure mechanisms that occur during loading at increasing confinement levels for different rock types and how they cause changes in the curvature of the strength envelop; and (c) provide suggested procedures for the formulation of triaxial data testing programs and the processing of triaxial testing results.

This content is only available via PDF.
You can access this article if you purchase or spend a download.