An MPFA-based Dual Continuum – Discrete Feature Model for Simulation of Flow in Fractured Reservoirs

Abstract

Discrete feature models (DFMs) are widely used to represent complex geological systems involving fracture networks and faults. Because DFMs explicitly incorporate these features, the complexity of the simulation model increases with the number of features included. The explicit resolution of very large numbers of features is therefore not tenable, which motivates the development of effective-property models for such systems. Previous work has incorporated full-tensor treatments for the matrix into DFMs through use of multipoint flux approximation (MPFA). This enables the matrix to include the permeability effects of subgrid fractures, though it does not model the dual-continuum behavior that will result for well-connected fracture networks. In this work, we introduce a new model, called the Dual Continuum – Discrete Feature Model (DC-DFM), that accounts for the dual-continuum nature of matrix–subgrid fracture interactions. In the DC-DFM methodology, the matrix continuum and the subgrid fracture continuum are both connected to the explicitly represented discrete features. To account for the full-tensor nature of the effective matrix and fracture permeability, the MPFA-O discretization scheme is applied to each continuum separately. Mass transfer between subgrid fractures and matrix rock is modeled using a transfer function. Numerical results for single-phase gas and two-phase oil-water systems are presented. High degrees of accuracy using DC-DFM, relative to reference fine-grid simulations, are achieved for the gas production examples. DC-DFM accuracy is not as high, though it is still reasonable, for the two-phase flow case. Speedups of a factor of 150 or more relative to fine-grid simulations are achieved using DC-DFM.