Injection from Well Experiments in Axisymmetric Triaxial Cells: Example of Hydraulic Fracturing

Laboratory experiments involving the injection of fluids through rocks under well-controlled conditions can help better constrain the processes taking place at depth during CO2 geo-sequestration in porous reservoirs, or hydraulic fracturing of tight gas/geothermal reservoirs. This article reports the testing of a new method for injecting a fluid through a cased borehole drilled in a cylindrical rock sample subjected to in-situ stress conditions in a conventional axisymmetric triaxial stress vessel. We use rock samples 1.5 inch in diameter in which a 2-mm wellbore can be drilled along or orthogonal to the sample's axis, which is also the direction of maximum principal compressive stress. The casing is made of 1/16 inch perforated stainless steel tube, cemented with a high-strength epoxy. We report here the results of 3 experiments with samples of American Black granite. Water is injected downhole (at the centre of the sample) through a wellbore drilled along or orthogonal to the sample's axis. In this tight rock, the injection pressure reaches a peak and leads to hydraulic fracturing of the sample in both configurations. In addition to the injection parameters (injection pressure/volume, flow rate), the impact of the injection operation is monitored thanks to 16 to 20 ultrasonic transducers attached to the lateral surface of the cylindrical sample, and operated in both active and passive modes. In the case of a sample with horizontal well we observed high stresses around the well that can be attributed to the relatively small size of the sample. As a result, one should not use this configuration to model fluid injection into rock formations in the field. On the other hand in the tests with a vertical well we observed a low breakdown pressure compared to predictions assuming a homogeneous and infinite elastic medium. We attributed this to the relatively large grain/flaw size compared to the well diameter, and the probable stress heterogeneities associated with the epoxy-sealed sections of the well. Finally, we observed a dependency of the P-wave velocity on the state of confining stresses applied to the sample. We argue that this dependency can be, at least partly, explained by the progressive deformation of the rock/sensor interface, rather than solely by the intrinsic stress dependency of the tested rock (e.g., micro-cracks closure with increasing mean compressive stress). We can also report that combined active and passive ultrasonic monitoring allows for an adequate reconstruction of the spatio-temporal evolution of the micro-seismicity induced by hydraulic fracturing.