Simulating and predicting the flow behaviour of fractured reservoirs start from a characterization of fractures and the construction of a calibrated representative Discrete Fracture Network. There are many fracture scales and their explicit modelling is impossible for fluid flow characterization at the reservoir scale. Therefore, at the field scale, for multiphase flow simulations, an equivalent dual media upscaled model is generally used. Nevertheless, upscaling of the fractures equivalent properties for any reservoir entails the following difficulties:
• accounting for the physics of flow transfers at the fracture scale;
• applying the proper hypothesis to approximate the solutions;
• being efficient enough in terms of computational time.
Many methods can be used to estimate the equivalent permeability of a diffuse fracture network. We currently use a fast analytical method or a more accurate but slower numerical method. Both methods take into account the medium heterogeneities in terms of porosities and permeabilities. When the numerical one is sometimes too CPU intensive the analytical one is not always applicable in particular for partially connected DFN.
The basic idea of this paper is to show, on a representative field case, how we can employ the analytical model when it is acceptable and use the numerical method for specifics situations in order to be efficient both in terms of computational time and of results accuracy. To achieve this aim, we suggest employing a Connectivity Index criterion. Results show that we can reduce the CPU time of upscaling procedure while conserving consistent results.
A better characterization of the properties needed for the reservoir modelling is allowed by recent improvements of the data acquisition. For naturally fractured reservoirs more specifically, two types of media can be distinguished (matrix and fracture) with discontinuous flow properties and heterogeneous storage capacities.
The main fluid conducting medium is the fractures network due to its high permeability, while the storage capacity is characterized by matrix porosity. The flow simulation can be based on both the petrophysical data and the geostatistical fracture descriptions. There are many fracture scales and their explicit modelling is impossible. The reservoir engineer used to discretize explicitly the reservoir scale objects like faults and to use upscaling methodologies for the smaller heterogeneities. Consequently, various parameters have to be considered and the determination of equivalent properties at the fluid flow simulation scale is a crucial step to obtain reliable production forecasts.
Typical workflows for simulating and predicting the flow behaviour of fractured reservoirs start from a characterization of fractures and the construction of a representative Discrete Fracture Network (DFN). For multiphase flow simulations at the field scale, an equivalent dual media upscaled model is generally used (Warren&Root, 1963 double permeability is the most frequently cited). Nevertheless, upscaling of the fractures equivalent properties for any reservoir entails the following difficulties:
• accounting for the physics of flow transfers at the fracture scale;
• applying the proper hypothesis to approach the solutions;
• being efficient enough in terms of computational time.
