La perméabilité des roches est fortement influencée par la présence de fractures car ces dernières constituent un chemin préférentiel pour l’écoulement des fluides. Ainsi la présence de fractures naturelles est un facteur déterminant pour la productivité d’un réservoir. Dans le cas de roches à faible conductivité, des techniques de stimulation telle que la fracturation hydraulique sont utilisées pour en augmenter la perméabilité et rendre le réservoir exploitable d’un point de vue économique. A l’inverse, dans le cas du stockage géologique, la présence de fractures dans la roche représente un danger dans la mesure où elle facilite le transport et la migration des espèces disséminées dans la roche. Pour le stockage de CO2, les fuites par les fractures présentes dans les couvertures du réservoir et la réactivation des failles constituent un risque majeur. Et en ce qui concerne le stockage géologique de déchets radioactifs, la circulation de fluide dans des réseaux de fractures nouvelles ou réactivées au voisinage de la zone de stockage peut aboutir à la migration de matériaux nocifs. Il est donc important de prévoir les effets de la présence de fractures dans un réservoir. Le but de cette thèse est le développement d’un outil numérique pour la simulation d’un réseau de fractures et de son évolution sous sollicitation hydro-mécanique. Grâce à sa commodité, la méthode des éléments finis étendue (XFEM) sera retenue et associée à un modèle de zone cohésive. La méthode XFEM permet en effet l’introduction de fissures dans le modèle sans nécessairement remailler en cas de propagation des fissures. L’écoulement du fluide dans la fissure et les échanges de fluide entre la fissure et le milieu poreux seront pris en compte via un couplage hydro-mécanique. Le modèle est validé avec une solution analytique asymptotique pour la propagation d’une fracture hydraulique plane dans un milieu poroélastique en 2D comme en 3D. Puis, nous étudions la propagation de fractures hydrauliques sur trajets inconnus. Les fissures sont initialement introduites comme des surfaces de fissuration potentielles étendues. Le modèle de zone cohésive sépare naturellement les domaines adhérents et ouverts. Les surfaces potentielles de fissuration sont alors actualisées de manière implicite par un post-traitement de l’état cohésif. Divers exemples de réorientation de fissures hydrauliques et de compétition entre fissures voisines sont analysés. Enfin, nous présentons l’extension du modèle aux jonctions de fractures hydrauliques / The permeability of rocks is widely affected by the presence of fractures as it establishes prevailing paths for the fluid flow. Natural cracks are then a critical factor for a reservoir productiveness. For low permeability rocks, stimulation techniques such as hydrofracturing have been experienced to enhance the permeability, so that the reservoir becomes profitable. In the opposite, when it comes to geological storage, the presence of cracks constitutes a major issue since it encourages the leak and migration of the material spread in the rock. In the case of CO2 storage, the scenario of leakage across the reservoir seal through cracks or revived faults is a matter of great concern. And as for nuclear waste storage, the fluid circulation in a fracture network around the storage cavity can obviously lead to the migration of toxic materials. It is then crucial to predict the effects of the presence of cracks in a reservoir. The main purpose of this work is the design of a numerical tool to simulate a crack network and its evolution under hydromechanical loading. To achieve this goal we chose the eXtended Finite Element Method (XFEM) for its convenience, and a cohesive zone model to handle the crack tip area. The XFEM is a meshfree method that allows us to introduce cracks in the model without necessarily remeshing in case of crack propagation. The fluid flow in the crack as well as the exchanges between the porous rock and the crack are accounted for through an hydro-mechanical coupling. The model is validated with an analytical asymptotic solution for the propagation of a plane hydraulic fracture in a poroelastic media, in 2D as well as in 3D. Then we study the propagation of hydraulic fractures on non predefined paths. The cracks are initially introduced as large potential crack surfaces so that the cohesive law will naturally separate adherent and debonding zones. The potential crack surfaces are then updated based on a directional criterion appealing to cohesive integrals only. Several examples of crack reorientation and competition between nearby cracks are presented. Finally, we extend our model to account for the presence of fracture junctions

[pt] Este trabalho faz uma análise da variação dos parâmetros que têm importância na propagação de fraturas hidráulicas e da influência desses parâmetros na conten-ção do fraturamento. Os experimentos numéricos foram feitos em um modelo 2D utilizando um simulador de elementos finitos com acoplamento sequencial hidro-dinâmico, tendo como premissa o comportamento dos processos envolvidos em es-tado estacionário. Inicialmente foram feitos testes de validação das soluções numé-ricas empregadas neste trabalho a partir de casos cujas soluções são bem conheci-das. Então, efeitos de variações de poropressão, de estado de tensões, propriedades das rochas, intervalos de início da fratura hidráulica, efeitos térmicos e o dano à permeabilidade da formação permoporosa foram utilizados para avaliar a contenção da fratura hidráulica. Primeiramente os efeitos foram avaliados separadamente e, em seguida, foram combinados aos pares, por meio de sorteio, e então avaliados. Os estudos levaram à conclusão de que o fator de maior influência para o início da propagação da fratura hidráulica na rocha capeadora (primeiros metros) é o valor da tensão mínima de confinamento do reservatório e a continuidade da propagação vertical na rocha selante é dominada pelo contraste de tensões entre rochas reserva-tório e capeadora. Entretanto, os demais parâmetros exercem influência na conten-ção do fraturamento hidráulico e devem ser levados em consideração neste tipo de estudo, principalmente os que servirão de insumo para a tomada de decisões. / [en] This master thesis analyzes the parameter s variation on the hydraulic frac-ture s propagation importance and the influence of these parameters on fracture containment. The numerical experiments were performed in a 2D model using a finite element simulator with sequential hydrodynamic coupling, having the sta-tionary behavior of the processes involved as premise. Validation tests were ini-tially performed for the numerical solutions used in this thesis from cases which solutions are well known. Then, effects of pore pressure variations, stress state, rock properties, hydraulic fracture opening intervals, thermal effects, and damage to the permoporous formation were used to evaluate the hydraulic fracture containment. At first, the effects were evaluated separately, then sorted for pair combinations, so they could be analyzed. These analyzes led to the conclusion that the most influen-tial factor for the hydraulic fracture initial propagation in the cap rock (first meters) is the reservoir’s minimum confinement stress value, and the vertical propagation continuity in the sealing rock is dominated by the stress contrast between reservoir and cap rocks. However, other parameters influence the hydraulic fracturing con-tainment and should be considered for this type of study, especially those that will serve as input for decision-making.

Both joint sets and fault-related fractures serve as important conduits for fluid flow. In the former case, they can strongly influence both permeability and permeability anisotropy, with implications for production of water, hydrocarbons and contaminant transport. The latter can affect issues of fluid flow, such as whether a given fault seals or leaks, and fault mechanics. These fractures are commonly interpreted as Natural Hydraulic Fractures (NHFs), i.e., mode 1 fractures produced when pore fluid pressure exceeds the tensile strength of the rock. Various mathematical models have been a rich source of hypotheses to explain the formation and propagation of NHFs, but have provided only limited information and nothing about processes of fracture initiation in originally intact rock. Recent laboratory experiments of French et al. (2012) have advanced our understanding of mechanical controls on fracture initiation and spacing. Here, detailed analysis of both through-going fracture surfaces, non-through-going fractures, in experimentally deformed samples provide a deeper understanding of NHF processes and resulting geometric features in porous siliciclastic sedimentary rocks. Observations indicate that both fracture planarity and microcrack damage (which has not previously been reported for opening mode fractures) vary significantly depending on the degree of mechanical heterogeneity and anisotropy of the host rock. Variations reflect mechanical controls on fracture initiation and propagation, suggesting that fracture spacing may in part reflect the distribution of mechanical heterogeneities. These data indicate that the more homogeneous the rock, the greater the microcrack damage surrounding a given NHF, increasing expected fracture-associated permeability for a given fracture aperture.