This paper presents field case studies of a coupled geomechanics and fluid flow computational algorithm for hydraulic fracture simulation in the Marcellus and Barnett formations. This computational algorithm is also coupled with tensile and shear-failure models to help describe pre-existing natural fracture evolution and its interaction with approaching hydraulically induced fractures. The numerical simulations are performed using field conditions and known parameters. The goal of the case studies presented in this work is to demonstrate the capabilities of the tightly coupled numerical algorithm for hydraulic fracturing analysis, especially for analysis of a complex fracture network. Emphasis is placed on characteristics of pre-existing natural fractures because the characteristics of pre-existing natural fractures directly result in different extents of complexity for created fracture networks. The presence of natural fractures contributes not only fracture interactions, but also well communications. The case studies presented can be used to provide insight for refining hydraulic fracturing treatment designs in similar reservoirs.
In recent years, cost reduction has become a focal point for operators considering hydraulic fracturing treatments in horizontal wells for unconventional formations. An important aspect of reducing cost is gaining a better understanding of the fracturing process and its influence on fracturing design and production, especially in the case of complex fracture mechanisms. Pre-existing natural fractures in unconventional reservoirs contribute uncertainties during the fracturing process in terms of propagation of hydraulic fractures and the interaction between natural fractures and hydraulic fractures. The interaction between hydraulic fractures and natural fractures could arrest or divert propagation of hydraulically induced fractures and generate a complex fracture network (CFN) other than planar fracture geometries. Because of such complexity, the classic planar model is too overly simplified to be useful for CFN simulations.
Multiple numerical simulation models have been developed for the hydraulic fracturing process for planar fracture geometries and complex fracture networks. These fracture models can be divided into two main categories. One category is continuum-mechanics based models, the other is discrete fracture models. Continuum-mechanics based models, such as the Continuum Damage Model (CDM), considers the displacement field and damage of material as continuous state variables. There is no explicit description of fracture surface geometry and no displacement discontinuity across fracture surfaces. Continuum-mechanics based models are available in commercial software, such as ABAQUS, and have been used in fracturing simulations, such as the studies by Shen, 2012. These models are facing many difficulties, such as coupling with fluid flow, proppant transportation/settling, and describing the physical interaction behaviors between hydraulically induced fractures and pre-existing natural fractures. Much more effort has been put into discrete fracture models that explicitly describe fracture geometries and displacement discontinuities across fracture surfaces. These models deal with different levels of geometry complexity of fracture propagation and interactions, from bi-wing planar fractures to complex fracture networks in 2-dimensional and 3-dimensional space. To fully describe the physics of the fracturing process, these models are also coupled with rock mechanics and fluid dynamics to model complex fracture propagation, stresses induced by fractures, complex fluid rheology inside fractures, proppant transport, and proppant settling. For fracture propagation and stress analysis in rock formations, many numerical methods have been used, including the Finite Element Method (FEM) coupled with the Cohesive Zone Model (CZM) (Chen, 2009), the extended Finite element Method (XFEM) developed by Moës et al., 1999, the Distinct Element Model (DEM) (Nagel et al., 2011), and the Displacement Discontinuity Method (DDM) by Olson, 2004. Each method has its own advantages and disadvantages in terms of computational efficiency and input complexity, such as heterogeneities of rock properties and distribution of pre-existing natural fractures. FEM based models, FEM with CZM, and XFEM, inherit the benefit of traditional FEM. They are flexible to incorporate linear or nonlinear constitutive laws, and heterogeneities of rock properties can also be handled rather easily, but these models might be computationally expensive. DDM has its main advantage in meshing fracture geometry; it is only necessary to mesh the fractures themselves, not the volume around the fractures. So, DDM has high computational efficiency, but it is not easy to include rock heterogeneities in the model.