Jason Robert York
Published: 2018
Total Pages: 189
Get eBook
In the effort to create new technology to enhance our ability to retrieve hydrocarbons, the technique of hydraulic fracturing has shown to be extremely beneficial. This involves pumping fluids at high pressures and high rates to induce and propagate fractures near the wellbore to stimulate production in otherwise low permeability reservoirs. To better understand the physical processes involved, several models have been proposed for numeric simulation. This work expands on an existing hydraulic fracturing model based on the nonlocal theory of peridynamics, detailed in Ouchi et al. [3]. Peridynamics is a relatively new reformulation of continuum mechanics, applicable even when discontinuities such as fractures are introduced. To incorporate the influence of inelasticity in the established model, which may be significant for several geologic materials, a multi-surface yield model is proposed. This yield model builds on a Drucker-Prager related yield model formulated for peridynamics by Lammi et al. [46], adding a tension cut-off surface as well as a cap to include hardening effects associated with inelastic compaction. The formulation of these additional surfaces in the peridynamic framework will be detailed and numerically demonstrated in this dissertation. As the peridynamic based hydraulic fracture model continues to develop complex capabilities, such as inelasticity, computational expense continues to be an ever-growing concern. Although the peridynamic formulation has demonstrated the capability of modeling complex fracture behavior, the computational expense is noted to be quite expensive relative to classic local models. Recently, methods have been introduced for coupling nonlocal bond based peridynamic grids with local finite element meshes, detailed in Galvanetoo et al. [74]. These coupling methods have demonstrated applicability to static equilibrium mechanics problems, while introducing negligible errors. In this work, the coupling method is implemented with the nonlocal hydraulic fracturing model, using peridynamics near existent and propagating fractures, as well as a standard finite element formulation far from the influence of such features. To further increase computational efficiency, a dynamically adaptive mesh coarsened away from the peridynamic region is implemented with the capability of converting finite element nodes to peridynamic nodes. This novel method of coupling peridynamics with a highly efficient mesh in the hydraulic fracture model will be fully detailed in this dissertation. In addition, 2D and 3D results will be provided using this method, demonstrating the capability of the coupled model to simulate complex fracture behavior, as well as discuss its impact on simulation capabilities and performance.