Papers »

Coupling fluid flow and earthquake modeling in 2d and 3d discrete fracture networks

Many computational approaches to induced seismicity have been continuum-based. However, earthquakes generally occur on discrete faults, and the stresses induced by deformation are highly heterogeneous, anisotropic, and dependent on the fault location and orientation. Therefore, discrete fracture network (DFN) modeling has strong potential advantages. However, DFN modeling is technically challenging, especially in fracture networks with complex geometry. I will demonstrate a computational approach that allows efficient, coupled, simulation of fluid flow, deformation, and friction evolution in large, complex 2D and 3D fracture networks. Dynamic effects are handled with a radiation damping term. Stresses are calculated with the boundary element method, and flow in the matrix is handled with a nonconforming mesh. Because these techniques are used, it is not necessary to create a conforming mesh of the area or volume surrounding the fractures, significantly simplifying implementation. Friction can be handled with a rate/state formulation or with a static/dynamic formulation, the latter being less realistic but more efficient. I will discuss numerical issues that arise when fluid pressure is so high that fractures experience low effective normal stress, or even opening. These computational challenges are not merely numerical- it does not appear that constitutive laws have been developed that are capable of fully describing these conditions. Simulations suggest that low effective normal stress conditions may be responsible some forms of seismicity, especially microseismicity observed during hydraulic fracturing of shale formations.

Author(s):

Mark McClure    
The University of Texas at Austin
United States