No description available.

Different rock formations may appear within the same mass, or even within the same formation there may exist layers of different materials, creating interfaces between layers (an interface may be defined, in more general terms, as a frictional contact that separates two similar or dissimilar materials). Currently, there is no well-established experimental work that investigates the influence of frictional interfaces, interface orientation and flaw geometries on crack behavior (i.e. initiation, propagation and coalescence) in brittle specimens under compressive loading. A series of experiments on homogeneous gypsum specimens, used as a rock-model material, containing two pre-existing open flaws and a frictional interface has been performed under uniaxial compression. The experiments investigate how cracks interact with interfaces and how different variables (i.e. flaw geometry, interface inclination angle and interface roughness) affect crack behavior in homogeneous materials separated by an interface. The specimens are 203.2mm high, 101.6mm wide, and 25.4mm thick. The two flaws, with 0.1mm aperture and 12.7mm length (2a), are created through the thickness of the specimen. The spacing (S) between flaws, continuity (C), and inclination angle, measured from the horizontal, (β) define the geometry of the flaws. Three flaw geometries are tested: S=0, C= -2a= -12.7mm, β= 30° (i.e. a left-stepping geometry); S= 2a= 12.7 mm, C=a=6.35 mm, β= 30° (i.e. an overlapping geometry) and S= 3a= 19.05mm, C=0, β= 30° (i.e. a right-stepping geometry). Smooth and rough unbonded interfaces are created by casting the specimen in two parts. The first half of the specimen is cast against a PVC block with an inclined face (i.e. 90°, 80° or 70°) with respect to the vertical axis of the specimen. The second half is then cast against the first one. Sandpaper may be attached to the PVC block to provide different roughness to the interface; a debonding agent applied to the interface ensures a cohesionless contact. In the experiments, digital image correlation (DIC) is used to monitor crack propagation on the specimen surface. The experiments indicate that the interface itself is an important contributor to new cracks and its presence in the specimens reduce crack initiation stress. Furthermore, the increase in roughness and inclination of the interface (i.e. from horizontal to 70° from the vertical) causes crack initiation stress to decrease. It was also observed that the angle between the incident crack plane and the interface affects whether an incident crack will penetrate an interface or be arrested: Tensile cracks that meet the interface at 30° to 60° angle get arrested, while those at or above 70° cross the interface with an offset of 0 – 1.2 mm. While shear cracks that meet the interface at 20° to 63° angles get arrested at the interface, while those at or above 70° cross the interface with an offset in the range of 0 – 1.76 mm. Another relevant finding is the fact that changes in interface roughness or inclination angle did not affect the angles at which cracks initiate or reinitiate at the interface.<div>A numerical study was conducted using the Extended Finite Element Method (XFEM) capability in ABAQUS, to further investigate the fracture behavior observed in the experiments and, more specifically, the influence of the different types of interfaces. An extensive investigation of the stress fields around the tips of the flaws and of the new cracks, as well as along the interface in the specimens, was conducted to determine the relationship between stresses and crack initiation and propagation (i.e. type and direction of cracks). The stress-based approach yields predictions of tensile and shear cracks location and initiation direction that are in good agreement with experimental results. The numerical investigation indicated that rougher horizontal interfaces induced slightly higher tensile stresses around the interior and exterior flaw tips than smoother interfaces, which may explain why tensile cracks at these locations initiated earlier in specimens with a rough interface. Moreover, inclined interfaces induced higher tensile stresses around the interior and exterior flaw tips than horizontal interfaces, which may justify that, in the experiments, inclined interfaces promoted crack initiation earlier than horizontal interfaces.</div>

La fracturation hydraulique est au cœur d'un certain nombre de phénomènes naturels et induits et est cruciale pour un développement durable de la production de ressources énergétiques. Compte tenu de son rôle crucial, ce phénomène a été pris en compte au cours des trois dernières décennies par le monde académique. Néanmoins, un certain nombre d'aspects très importants de ce processus ont été systématiquement négligés par la communauté. Deux des plus remarquables sont l'incapacité de la grande majorité des modèles existants à aborder la propagation des fractures hydrauliques dans les massifs rocheux fracturés où l'injection de fluide peut à la fois conduire à la fracturation de la roche intacte et à la réactivation de fractures préexistantes. Un autre aspect essentiel de ce processus est qu'il est intrinsèquement tridimensionnel, ce qui est souvent négligé par les modèles actuellement disponibles. Pour aborder ce problème essentiel, un modèle hydro-mécanique couplé basé sur la méthode des éléments discrets a été développé. La masse rocheuse est ici représentée par un ensemble d'éléments discrets interagissant à travers des lois de contact cohésifs qui peuvent se casser pour former des fissures à l'intérieur du milieu simulé. Ces fissures peuvent se coalescer pour former des fractures. Une méthode de volume fini est utilisée pour simuler l'écoulement de fluide entre les éléments discrets. L'écoulement est calculé en fonction de la déformation de l'espace poreux dans le milieu intact et de l'ouverture des fissures dans les fractures. De plus, les fractures naturelles sont modélisées explicitement de sorte qu'elles peuvent présentées des comportements mécanique et hydraulique différents de ceux de la matrice rocheuse intacte. La simulation des processus de fracturation hydraulique dans un milieu initialement intact en considérant plusieurs points d'injection plus ou moins espacés a permis de mettre en évidence l'évolution spatio-temporelle des fractures hydrauliques et de quantifier l'impact des différentes stratégies d'injection sur des indices représentatifs du volume fracturé, de l'intensité et de la densité des fractures ou encore sur la pression de fluide au niveau du puits. De plus, l'injection dans une fente de perforation non alignée sur le plan de contrainte minimum a génère des fractures hydrauliques non planaires percolantes si la connectivité est faible, ce qui peut être gênant pour la mise en place du proppant. En outre, des interactions fortes prennent place entre des fractures hydrauliques étroitement espacées ont été mises en évidence grâce au le suivi de la orientation de contrainte principale locale et ont révélé l'importance des effets d'ombre de contrainte. Des solutions sont proposées pour optimiser les traitements multiples à partir d'un puits de forage non parfaitement aligné. Enfin, l'interaction entre une seule fracture hydraulique et une seule fracture naturelle de propriétés et d'orientations variables a été étudiée à l'aide du modèle proposé. L'évolution de la fracture hydraulique et la réponse globale de l'échantillon ont été enregistrées d'une manière comparable aux données expérimentales existantes pour établir un pont entre les résultats expérimentaux et numériques. Les fractures naturelles persistantes semblent être des barrières pour la fracture hydraulique si leur conductance est élevée par apport a celle de la matrice ou si leur raideur est faible par rapport a la rigidité du milieu environnant. D'autre part, une faible rigidité dans les discontinuités non persistantes pourrait provoquer une bifurcation de la fracture hydraulique principale. De plus, des angles d'approche élevés et des contraintes différentielles fortes semblent favoriser le croisement de la fracture naturelle alors que des angles faibles engendrent plutôt un glissement ou une dilatation par cisaillement de la partie du plan qui n'est pas affectée par la perturbation de la contrainte. / Hydraulic fracturing is at the core of a number of naturally occurring and induced phenomena and crucial for a sustainable development of energy resource production. Given its crucial role this process has been given increasing attention in the last three decades from the academic world. Nonetheless a number of very significant aspects of this process have been systematically overlooked by the community. Two of the most notable ones are the inability of the vast majority of existing models to tackle at once the propagation of hydraulic fractures in realistic, fractured rocks-masses where hydraulic fracturing is a competing dipole mechanism between fracturing of the intact rock and re-activation of exiting fracture networks. Another essential aspect of this process is that it is intrinsically three-dimensional which is neglected by most models. To tackle this vital problem taking into account these pivotal aspects, a fully coupled hydro-mechanical model based on the discrete element method has been developed. The rock mass is here represented by a set of discrete elements interacting through elastic-brittle bonds that can break to form cracks inside the simulated medium. Theses cracks can coalesce to form fractures. A finite volume scheme is used to simulate the fluid flow in between these discrete elements. The flow is computed as a function of the pore space deformation in the intact medium and of the cracks' aperture in the fractures. Furthermore, the natural fractures are modelled explicitly and present mechanical and hydraulic properties different from the rock matrix. Employing this model in an intact numerical specimen, single fluid injection and multiple closely spaced sequential injections, enabled the description the full spatio-temporal evolution of HF propagation and its impact on quantitative indexes used in description of hydraulic fracturing treatments, such as fractured volume, fracture intensity and down-the-hole pressure for different control parameters and in-situ stress-fields. Moreover, injections from perforation slots which are not well aligned to the minimum stress plane showed possible creation of percolating non-planar hydraulic fractures of low connectivity, which can be troublesome for proppant placement. Also, strong interactions between closely spaced HF were highlighted by tracking the local principal stress rotation around the injection zones, emphasizing the importance of stress shadow effects. Optimization solutions are proposed for multiple treatments from a non-perfectly aligned wellbore. Finally, interaction between a single hydraulic fracture and a single natural fracture of varying properties and orientations was studied using the proposed model. The evolution of the hydraulic fracture and the global response of the specimen were recorded in a way comparable to existing experimental data to bridge the experimental and numerical findings. Persistent natural fractures appeared to be barriers for the hydraulic fracture if their conductance is high compared to the matrix conductivity or if their stiffness is significantly low compared to the rock matrix rigidity. Low stiffness in non-persistent defects might also cause a bifurcation of the main hydraulic fracture due to the local stress field perturbation around the defect and ahead of the hydraulic fracture tip. Furthermore, high approach angles and differential stresses seemed to favour crossing of the natural fracture while low angles enable shear slippage or dilation on the part of the plane which is not affected by the local stress perturbation.