Interaction between Hydraulic Fracturing Process and Pre-existing Natural Fractures

Abstract

Hydraulic fracturing is employed as a stimulation treatment by the oil and gas industry to enhance the hydro-carbon recoveries. The rationale is that by creating fractures from the wellbore into the surrounding formations, the conductivity between the well and reservoir is significantly increased and the hydro-carbon flow is therefore stimulated. The hydraulic fracture is initiated and driven by pressurizing a bore-hole section via fluid injection. Despite its half century’s practice, hydraulic fracturing treatments sometimes fail to increase the well productivity. One prominent reason is that there are pre-existing (natural) fractures in connection with the wellbore or in the way of the hydraulic fracture propagation. If the natural fracture is opened by the injection fluid, the borehole pressure will decrease as if a hydraulic fracture break-down has taken place. If a hydraulic fracture is intercepted by natural fractures that layer the formation, its dimension could be restricted in only one of the layers. In both of the cases, the fluid losses via the natural fractures could mislead the interpretation of the bore-hole measurements. Laboratory tests were design to investigate the interaction between the hydraulic fracturing process and natural fractures’ fluid infiltration. To characterize the natural fractures’ mechanical and hydraulic properties under normal confining and shear, we performed shear and flow tests with rock samples cleaved into layers. To characterize the hydro-natural fracture interaction, we performed the injection fracturing tests on the layered samples. We interpreted the test results by correlating the possibility of the hydraulic fracture crossing-over natural fractures with the test conditions. We concluded from the test results that if the natural fracture is closed by the normal stress significantly higher than the minimum in-situ stress, it is then likely to be crossed by the hydraulic fracture. Tests were facilitated by a tri-axial set-up in combination with an acoustic monitoring system. The tri-axial loadings maintained the samples’ in-situ stresses which determined the hydraulic fracture orientations. The monitoring system tracked the fracture propagation by converting the diffraction signal arrival times to the fracture tip locations via a velocity model. Comparisons between the recovered fractures from the acoustic monitoring and sample postmortems showed agreement. In modeling the fracture initiation and propagation, we introduced a quasi-static numerical algorithm that coupled the 2D fluid domain and 3D matrix domain via boundary conditions. The model was extended to include a layer interface (natural fracture) transverse to the bore-hole. As a result of the fluid infiltration into the interface, the coupling between the layers was weakened. The displacement loss across the interface was related to the weakening process when the hydraulic fracture met the interface. The model results showed discontinuities across the interface in the fracture width and injection fluid distribution, which indicate the impact of the natural fracture infiltration to the hydraulic fracture development. In conclusion, the model was a novel approach to simulate such a complicated process and the comparison between the modeled and test results showed agreements.