This paper presents a comparison between natural and stochastic fracture systems in terms of their geomechanical response to in-situ stresses. An analogue fracture network (AFN) is extracted from the geological map of a limestone outcrop. A corresponding discrete fracture network (DFN) is generated using the statistics obtained from the analogue pattern, to ensure the two networks share the same statistical characteristics. The geomechanical response is modelled using the combined finite-discrete element method (FEMDEM). A series of numerical experiments is designed with far-field stresses applied at a range of angles to the rock domain. A comparison between the natural fracture system and its DFN equivalent is made based on phenomena such as fracture-dependent stress heterogeneity, re-activation of pre-existing fractures, new crack propagation and variability of aperture distribution. This study addressed the validity of using the DFN approach for geomechanical modelling of fractured rock masses and also has implications for flow simulations.
Naturally fractured reservoirs contain numerous discontinuities which greatly influence or even dominate the hydraulic behaviour of the host media. The formulation of computational fracture networks, therefore, is of great importance if fluid flow is to be accurately modelled and oil production reliably predicted.
An analogue fracture network (AFN) is often used to realistically characterise the geometrical attributes of a naturally fractured rock mass. It is a fracture pattern of actual rock outcrops involving complicated intersections, bends and segmentations. Many studies have been done based on mapped analogues to investigate the mechanisms of fluid transport in carbonate reservoirs (Belayneh et al. 2006). A uniform aperture distribution or a simple linear correlation between fracture length and aperture is always assumed in conventional AFN-based flow simulations. This hypothesis oversimplifies the geologically-induced variability of natural fracture apertures. For instance, in a compressive geological circumstance, only a few fractures are hydraulically active and serve as paths for fluid migration, whereas others may contribute little. Obviously, an accurate quantification of the flow behaviour of a porous rock medium requires a natural fracture network incorporating a realistic aperture distribution as well as new crack propagation on account of potential impacts on connectivity (Latham et al. 2013).
