We present an experimental and numerical study of the mechanism of hydraulic fracturing in unconsolidated rocks. We performed hydraulic fracturing experiments on loose sand samples under various confining pressures. We have developed a large biaxial set up in which cylindrical sand samples of dimensions 0.4×0.51 m can be tested under confining stress up to 40 MPa. We prepare a sand sample by pouring the sand into the sample vessel, which is filled with water to obtain a fully saturated sample. There is a borehole (injection system) with an open section in the middle of the vessel. The open section is closed during the sample preparation. After compacting the sample and applying the confining pressures, the injection system is opened and fluid is injected.
For modeling of the injection process we used a distinct element model in which we incorporated fluid flow that is coupled to the flow of the grains.
The injection tests resulted in either infiltration or generating fractures. Injecting viscous Newtonian fluid and cross-linked gel yielded in most cases only infiltration and wellbore expansion. Clear fractures were induced when injecting cross-linked gel with quartz flour and bentonite mud. Tests were run at a stress level between 0.2 and 20 MPa. At high confining stress we observed a strong tendency for shear fracturing during fracture initiation. The fracture initiation pressure was about 2.5-3.5 times the confining pressure. The stress level had a strong influence on the ratio of initiation pressure to confining stress. Also, it appeared that much higher pressures were obtained on dense sand samples, compared with less compacted sand.