Active Acoustic Monitoring of Laboratory Hydraulic Fracture Growth under Triaxial Confinement

Laboratory experiments of fluid-driven fracture propagation are a great tool to investigate the mechanisms driving fracture growth. They allow for a controlled environment where the controlling factors such as confining stresses and pore pressure, elastic constants and fracture toughness are well known or estimated before the experiment. It is then possible to investigate the influence of a few selected experimental parameters on the fracture propagation, and to tie the experimental results with numerical models in oder to validate them.

Here we describe an experimental setup consisting in a reaction frame fitted with three pairs of flat jack that are independently controlled. This lets us perform injections in cubic rock samples with dimensions 250 x 250 x 250 mm, under true triaxial conditions with confining stresses up to 20 MPa. The injection is done with a high-pressure injection pump for a maximum pressure of 50 MPa and a flow rate ranging from 1 μL/min to 90 mL/min, using injection fluids with various viscosities. This gives us a wide range of injection conditions, including toughness and viscosity dominated regimes, for injections lasting on the order of tens of minutes.

We monitor the fracture growth using an active acoustic setup consisting of 64 piezoelectric transducers — including 10 shear ones — organized in 32 sources and 32 receivers. The transducers are positioned in platen against all six faces of the sample, in order to record waves transmitted through, reflected and diffracted by the fracture. Our recording system allows to quickly excite sequentially all 32 sources while recording on all 32 receivers, in order to save a full active acoustic snapshot of the fracture every few seconds. With this information we estimate the fracture size and opening at every step of the growth.

We then present experimental results for fluid-driven fracture propagations in homogeneous isotropic rocks but also in anisotropic (VTI) slate, and compare the results with predictions from numerical models.