A series of monotonic uniaxial compression tests under quasi-static loading conditions on Hawkesbury sandstone specimens were conducted. To accurately measure pre-peak and post-peak stress-strain behaviour, lateral strain-controlled test at a prescribed constant loading rate were undertaken. To study localised deformations of specimen, 3D Digital Image Correlation (DIC) technique was used. 3D DIC is a non-contacting optical method for strain measurement that uses two-digital cameras to acquire images of the undeformed and deformed shape of an object. Field of strains and strain localisation in the rock surface were analysed by 3D DIC method and coupled with the respective stress levels applied to the rock. By using 3D DIC technique relatively large strains developed in the post-peak regime within localised zones were accurately measured. Field strain development in the rock samples suggests that strain localisation takes place progressively and develops at a lower rate in pre-peak regime, otherwise it is accelerated in post-peak regime associated with the increasing rate of strength degradation. The results show that a major shear failure plane, due to strain localisation, becomes noticeable only long after peak stress.
The mechanical properties of intact rock extracted from the stress-strain curve in uniaxial compression tests have major implications on civil engineering projects, mining engineering and mineral exploration related operations. In this sense, the complete stress-strain characteristics of intact rock are relevant in the understanding of the total process of rock deformation. The complete stress-strain curve for both class I and II rocks (Wawersik & Fairhurst 1970) can be successfully obtained, depending on the brittleness of the rock, by implementing a closed-loop servo-controlled loading system having as feedback signal to control the applied axial load a prescribed constant lateral-strain rate (Hudson et al. 1971). This method has been by far most widely used to investigate the post-peak behaviour of quasi-brittle materials (Hudson et al. 1971, Fairhurst & Hudson 1999); the other method is based on a linear combination of stress and strain (Okubo & Nishimatsu 1985).
Improper measurement of strains in the post-peak regime can lead to improper characterisation of the post-peak behaviour of rock useful to quantify postpeak fracture energy (Bazant 1989, Markeset & Hiller-borg 1995, Jansen & Shah 1997) and rock brittleness on the ground of post-peak instability (Tarasov & Potvin 2013, Tarasov & Randolph 2011). Conventional strain measurement of rock specimen in uniaxial compression includes setting up external devices, e.g. Linear Variable Displacement Transducers (LVDT) and direct-contact extensometers or strain gauges. In this respect, a major drawback using LVDT includes bedding error measurements (Taheri & Tani 2008). Although free from bedding errors, extensometers and strain gauge measurements are yet limited to a fixed gauge length and specific points of bonding, respectively. In addition, they may not capture entirely strains found in post-peak regime. In this case, extensometer misalignment and strain gauge damage are the major issues particularly when progressively growing cracks and localisation take place in the rock surface. Therefore, accurate non-contact strain measurements, via three-dimensional digital image correlation (3D DIC), becomes relevant to study strains in particular in post-peak regime.