Semi-Automated Fracture Detection for Three-Dimensional Stochastic Fracture Network Generation and Analysis

Stephen Ahlgren1, Robert Smallshire1, Paul Griffiths1, James Holmlund2, and Joe Nicoli2. (1) Midland Valley Exploration Ltd, 14 Park Circus, Glasgow, G3 6AX, United Kingdom, phone: +44 (0) 141 332 2681, steve@mve.com, (2) Geo-Map, Inc, 3323 N. Campbell Ave, Suite 1, Tucson, AZ 85719

Population of reservoir models with fracture patterns having similar geometric and topological characteristics as natural fracture systems is a formidable challenge for geoscientists and petroleum engineers.

We present a method to integrate observed fracture data into a three-dimensional structural model to generate realistic predictive fracture networks. We use a new technique for acquiring fracture data from analog outcrops using a long-range three-dimensional laser scanning system. Sub-planar regions within the laser scan data are identified using automated and semi-automated feature extraction algorithms. Descriptive population and clustering statistics are automatically computed from the geometry of detected fractures.

These fracture statistics are used to stochastically populate a three-dimensional volume with synthetic fractures. The three-dimensional synthetic fracture network may be analyzed topologically to define fracture connectivity and used as a resource for well planning or upscaling for simulation, thus providing an important link between field data and synthetic discrete fracture networks.

These new techniques have been applied to blasted outcrop faces in southern Arizona to evaluate fracture systems and population statistics. The fracture attributes were used to generate a simulated, three-dimensional fractured reservoir model with a synthetic fracture network.

AAPG: Fractured Reservoirs and Multi- Phase Fluid Flow
AAPG Annual Meeting 2002: Our Heritage-Key to Global Discovery Technical Program
AAPG Annual Meeting 2002: Our Heritage-Key to Global Discovery Main Page to: Registration and Exhibit Information