The presence of multi-scale heterogeneities, including stimulated fractures (hydraulically induced or open), natural fractures of various sizes embedded in unconventional low permeability reservoirs, increases the complexity of the reservoir simulation. This paper proposes a methodology to address this challenge, taking into account reservoir key parameters such as fractures locations, orientation, anisotropy and reservoir low permeability in a unique model as simple as possible.
The method may be included in the Discrete Fracture Model (DFM) family as it discretizes complex fracture networks and the surrounding matrix. Actually, some DFMs rely on unstructured grids to conform the fracture geometry and location, where all types of fractures are explicitly discretized, leading to a complicated and often non tractable numerical system to solve. To overcome these limitations, Embedded Discrete Fracture Models (EDFM) propose a hierarchical method to easily deal with this multi-scales problem. However, the matrix-fracture interaction is not properly handled with the EDFM due to the very low matrix permeability. In this paper, we will present a DFM based on MINC (Multiple INteracting Continua) approach to improve the EDFM. The MINC proximity function is computed by taking into account all discrete fractures, and the fractures are considered within a triple porosity model, where the propped fractures are explicitly discretized and other fractures are homogenized. In order to improve the flow exchange between the matrix and fracture media, the matrix gridblock is subdivided according to a MINC proximity function based on the distance to all discrete fractures, by using randomly sampled points. This approach is particularly useful for multi-phase flow simulations. For example, in a tight-oil reservoir, when the fracture pressure drops below the bubble point, gas starts to appear in the matrix formation near the fracture faces. The MINC method is suitable to simulate this kind of phenomena, which cannot be handled with a standard approach. This improved DFM based on a MINC proximity function was tested for a single-phase flow case (gas only) and a two-phase flow case with gas liberation from a tight-oil formation. The results are accurate comparing to an explicit model set as the reference solution. An application to a tight-oil case is presented and the three phase (water, oil and gas) flow is simulated with the improved DFM.
Usually, shale gas formations are naturally fractured and fractures are irregularly distributed through the reservoirs. Such characteristics from shale formation increase the heterogeneity and the complexity of the reservoir simulation and make flow modelling for such reservoirs quite difficult (see Fig. 1). To model a realistic reservoir fracture network, a new type of model called discrete fracture models (DFMs) has received a great attention. These kinds of models, which discretize explicitly complex fracture networks (hydraulic, reactivated, induced, micro-fractures, etc…), involve too many unknowns and often non tractable numerical system to solve.