True triaxial tests have been carried out on two porous sandstones, Coconino (n=17.5%) and Bentheim (n=24%) to investigate the effect of the intermediate principal stress (σ2) on compressive strength, failure-plane angle and failure mode. These rocks were selected because they are similar to sandstones forming hydrocarbon reservoirs, as well as to those being considered for CO2 sequestration. In both rocks the increase in strength as σ2 rose above a given σ3 was found to be similar to that established in crystalline rocks, but the maximum rise in strength was considerably smaller. Failure mode in Coconino sandstone was in the form of a steeply inclined shear band, or fault, at lower σ3, with the slope becoming gentler as σ3 increased, culminating in multiple parallel and conjugate bands at the highest σ3 attempted (=150MPa). Bentheim sandstone failure mode experienced a similar evolution in a narrower range (σ3 between 0 to 80MPa), but as loading continued, failure was in the form of shear-enhanced compaction bands with angle dropping to ~ 45° and lower, and reaching 0° at σ3 =150MPa (creating pure compaction bands). Failure-plane angle in both rocks increased by up to 15° as σ2 was raised above a given σ3.
The University ofWisconsin true triaxial testing apparatus has been used for studying the effect of the intermediate principal stress (σ2) on rock brittle failure and failure-plane angle (the angle between the normal to failure plane and the direction of the major principal stress, σ1). To date studies have been conducted on Westerly granite (Haimson & Chang 2000), KTB amphibolite (Chang & Haimson 2000), Long Valley Caldera hornfels and metapelite (Chang & Haimson 2005), TCDP low-porosity sandstone/siltstone (Oku et al. 2007), and SAFOD granodiorite (Lee & Haimson 2011). All tested rocks, except for the ultra-fine grain Long Valley metapelite and hornfels, exhibited substantial compressive strengthening (expressed by σ1 at failure, or σ1,peak) as σ2 was raised above a constant σ3.TCDP siltsone was the only rock tested through the entire σ2 range, showing that the increase in compressive strength reached a peak at about halfway between σ2 =σ3 and σ2 =σ1, so that σ1,peak was back to its level when σ2 =σ3 as σ2 approached σ1. These results reveal the deficiency of the Mohr-Coulomb criterion, which neglects the role of σ2, predicting only the lower limit of rock strength for a given σ3.