Laboratory Imaging and Phase Field Modeling of the Interaction of Hydraulic Fractures with Well Cemented Natural Fractures

Induced hydraulic fractures in the field interact heavily with pre-existing natural fractures in the rock that are abundant in many formations. Most laboratory fracturing investigations in the literature consider pre-existing fractures as frictional interfaces with zero thickness. However, natural fractures in subsurface formations are often sealed with mineral cementing material of finite thickness. In this study, we present a novel experimental demonstration of the behavior of an induced hydraulic fracture as it approaches a cemented natural fracture utilizing a two-dimensional (2-D) hydraulic fracturing cell. Sheet-like test specimens are cast with natural fractures of varied mechanical properties, thickness, and relative position to a fluid injection port. Plaster is used as the specimen matrix. The filling material for hard natural fractures are cast using hydrostone while soft natural fractures are cast using a mixture of plaster and talc. Several tests are performed to characterize the mechanical and flow properties of these materials. A novel method for casting the specimen matrix and filling material of the natural fracture is described and used to enable strong bonding between the natural fracture and specimen matrix. The test specimen is placed between two thick, transparent plates and constant, anisotropic far-field stresses are applied to the specimen. Fracturing fluid is injected in the center of the specimen and the induced fracture trajectories in several experiments are captured with high-resolution digital images. We show a clear tendency for the induced hydraulic fracture to cross thick natural fractures filled with materials softer than the host rock and to be diverted by thick natural fractures with harder filling materials. The induced hydraulic fracture also tends to cross hard natural fractures when the natural fractures are relatively thin. In addition, the induced hydraulic fracture from the injection port is shown to be diverted by a thin, hard natural fracture that is placed relatively close to the injection port but crosses the same natural fracture when placed farther away from the injection port. Using our in-house numerical simulator that is based on the phase field approach, we model these laboratory experiments to gain insights into the observed fracture behaviors. Our results provide clear evidence of the impact of natural fracture filling material, natural fracture width, and the induced hydraulic fracture length on the outcome of hydraulic fracture interaction with natural fractures. The small-scale, 2-D nature, and well-characterized properties of our laboratory specimens are also valuable for validating numerical hydraulic fracturing simulators that are capable of modeling the effect of pre-existing natural fractures on hydraulic fracture propagation.