Stochastic propagation of discrete fracture networks.

in: Proc. 29th Gocad Meeting, Nancy

Abstract

Faults and fractures are known to significantly alter the circulation of fluids in reservoirs, either acting as seals or conduits. Characterization and modeling of the spatial distribution of fractures, as well as their properties (e.g. geometry, connectivity, aperture, etc.) can therefore have a great impact for flow study and recovery prediction of hydrocarbons. Among available methods, objectbased simulation with marked point process emerges as a peerless technique to generate models of discrete fracture networks (DFN). However, providing consistent structural models of DFN remains difficult for the approach relies on simplistic assumptions, such as planar and independent fractures. In this paper, we report on our investigation to overcomes these limitations. A new method for stochastic simulation of fracture networks has been developed. The proposed method integrates principles of fracture mechanic in the simulation process. The developed algorithm is a hybrid object/variogram based geostatistical approach where fracture are represented as a set of connected points rather than parametric objects. An object-based simulation is used to generate initial fracture polygons. Each fracture tip is then propagated to simulate the fracture growth. The direction of propagation is determined by kriging and account for mechanical interaction with surrounding fractures. As a result the method mimics the initiation, propagation, interaction and termination of fractures, generating more realistic fracture patterns and connections. The different steps of this method as well as the principles that leads to its design are presented in this paper

Download / Links

BibTeX Reference

@INPROCEEDINGS{HenrionGM2009,
    author = { Henrion, Vincent and Caumon, Guillaume },
     title = { Stochastic propagation of discrete fracture networks. },
 booktitle = { Proc. 29th Gocad Meeting, Nancy },
      year = { 2009 },
  abstract = { Faults and fractures are known to significantly alter the circulation of fluids in reservoirs, either acting as seals or conduits. Characterization and modeling of the spatial distribution of fractures, as well as their properties (e.g. geometry, connectivity, aperture, etc.) can therefore have a great impact for flow study and recovery prediction of hydrocarbons. Among available methods, objectbased simulation with marked point process emerges as a peerless technique to generate models of discrete fracture networks (DFN). However, providing consistent structural models of DFN remains difficult for the approach relies on simplistic assumptions, such as planar and independent fractures. In this paper, we report on our investigation to overcomes these limitations. A new method for stochastic simulation of fracture networks has been developed. The proposed method integrates principles of fracture mechanic in the simulation process. The developed algorithm is a hybrid object/variogram based geostatistical approach where fracture are represented as a set of connected points rather than parametric objects. An object-based simulation is used to generate initial fracture polygons. Each fracture tip is then propagated to simulate the fracture growth. The direction of propagation is determined by kriging and account for mechanical interaction with surrounding fractures. As a result the method mimics the initiation, propagation, interaction and termination of fractures, generating more realistic fracture patterns and connections. The different steps of this method as well as the principles that leads to its design are presented in this paper }
}