Due to their initial low permeability, unconventional plays can be economical only through hydraulic fracturing. This process, in order to be controlled needs to rely on a solid representation of the natural fracture geometry, an accurate stimulation model which considers the interaction with natural lineaments, and a physical reservoir model which can account for the different flow regimes occurring during production. The stimulated volume drainage can be evaluated using either Decline Curves Analysis/Rate Transient Analysis (DCA/RTA) techniques or reservoir simulation. In both cases, the geometry of the final Discrete Fracture Network (DFN) issued from the natural characterization and the stimulation, is very important, and for practical purposes is either overly idealized (Warren & Root approach) or oversimplified (Bi-wing). The models have shown their limitations when confronted with measurements in the field, opening up ways to use DFN geometries within integrated reservoir studies. The present work addresses some of the issues above, developing a hierarchical Discrete Fracture Model (DFM) based on the "filtering" of a stimulated DFN, realistically obtained by the characterization step and the stimulation process. This leads to a triple-continuum representation, consisting of: (1) the matrix media, (2) a high conductive stimulated fracture network and (3) a low conductive stimulated fracture network. The method consists in homogenizing low conductive networks, keeping a user defined backbone of high conductive fractures as the main "reservoir" DFN. One of the main advantages of this DFM relies on the way we compute the well-known Multiple Interacting Continua (MINC) approach, using a "proximity function" formalism, able to simulate transient effects. Using practical examples, this paper demonstrates applicability capacities of this method, enabling the integration of more complex geometries within a "quick" simulation framework.
