Connectivity Assessment and Permeability Upscaling of Fracture Network Using Grid Independent Particle Tracking Algorithm

Abstract

Abstract Discrete Fracture Network (DFN) models are now becoming an industry practice to model fractures in naturally fractured reservoirs. However, incorporating these fractures in simulation model has always been a challenge. In all previous approaches, explicit representation of fractures has been either in 2D (assuming vertical fractures) or in simple models within a small domain. So far it has been impossible to represent discrete fracture networks in realistic reservoir models mainly because of two reasons. Firstly because DFN comprises of extremely large number of fractures for full field reservoir model, it requires significant computational capability to represent them. Secondly the extreme aspect ratio between fracture aperture and fracture length makes it impossible to represent DFN on the simulation grid. Effective permeability of a fracture network depends upon the connectivity of the fractures to form connected flow paths. Many different techniques have been proposed to find the upscaled permeability of fracture network. But most of them are not rigorous enough and lack detailed characterization. These techniques either require the fracture network to be represented on a gridded system or make overly simplifying assumptions. Therefore, there is need for new methods which are accurate, comprehensive and computationally fast. In this paper a novel modelling approach is adopted. An implicit fracture network has been used to represent discrete fracture network. It doesn't requires explicit representation of fractures on the simulation grid and is computationally very fast. Then a particle tracking algorithm is used to find the percolation characteristics of the network. The particles can percolate from one end of the network to the other end if a connected path exists. Similarly the proportion of percolating particles represents the number of connected paths in the fracture network and therefore is a measure of effective permeability of the system. The particle statistics are calibrated against high resolution flow simulation for some simple fracture network representations. The calibration enables us to get upscaled permeability of a complex fracture network if the statistics of the random particles is known. The technique is computationally inexpensive and fast. Uncertainty assessment of fracture network permeability is a vital part of fracture characterization. It is important because there is very little information available about the fracture network away from the wellbore. In this work multiple realizations of fracture network has been generated to characterize the uncertainty in the upscaled permeability. All these realizations are conditioned to the same input data and are therefore equi-probable. The present work includes the following Generate multiple realizations of implicit fracture network models all conditioned to the same input data.Obtain the effective/upscaled permeability of the implicit fracture network using particle tracking algorithm.Deduce the percolation threshold for the fracture network measured as threshold fracture intensity.Characterize permeability anisotropy due to preferential orientation of fractures.