Embedded Discrete Fracture Modelling or how to put a needle in a haystack

Any endeavour to accurately model flow through fractured porous media at the field-scale must overcome two important challenges. First, the discretized representation of the medium needs to accommodate the complex geometry of small-scale intersecting fractures with various lengths, apertures and orientations. Second, the model formulation must ensure that the conductivity of these fractures, which can be orders of magnitude higher or lower than that of the host rock, is properly taken into account when computing the pressure map. The Embedded Discrete Fracture Model (EDFM) [Lee et al, SPEJ 2000] is well known in the literature for its flexibility in representing fractures. It was proven effective in capturing the flow behaviour through porous media containing highly conductive fractures [Hajibeygi et al, JCP 2011; Moinfar et al, SPEJ 2014; Tene et al, JCP 2016]. However, the EDFM formulation fails to capture the effect of low-permeable features, such as embedded flow barriers. This presentation explores two different prototype models which aim to overcome the limitations of EDFM, while maintaining its flexbility. More specifically, these models construct independent grids for the fracture and matrix domains, in a manner similar to EDFM. However, the transmissibilities of the matrix interfaces are adjusted to account for the conductivity of neighbouring fracture networks. The results of numerical experiments, investigating the sensitivity to grid resolution and fracture-matrix permeability ratio, show that the new models are a step forward from EDFM towards field-scale simulation of flow through fractured porous media. Possible paths to further increase the generality of embedded discrete fracture models are outlined at the end of the talk.