Probabilistic Fracture Network Models for Preliminary Site Characterization

Abstract

Ontario Power Generation's Deep Geologic Repository Technology Program has developed a geostatistical procedure for creating 3D fracture network models (FNMs) that honour the types of information typically available for preliminary site characterization: detailed information on the locations of surface lineaments from aerial photography and remote sensing; regional tectonic information on stress; geomechanical and structural geology principles; field data gathered from geologically analogous sites. This approach provides a systematic and traceable method that is flexible and that accommodates data from many different sources. The detailed, complex and realistic models of 3D fracture geometry produced by this method can serve as the basis for developing rock property models to be used in flow and transport studies. In addition to being ideally suited to preliminary site characterization, the approach also readily incorporates field data that may become available during subsequent site investigations, including ground reconnaissance, borehole programs and other subsurface studies. The FNMs from this method are probabilistic in the sense that they consist of a family of equally likely renditions of fracture geometry, each one honouring the same surface and subsurface constraints. Such probabilistic models are well suited to studying issues that involve risk assessment and quantification of uncertainty. The geostatistical procedure for simulating FNMs is described, its use in case study examples is presented, and the realism of its fracture geometries is tested using field data collected from the Lagerdorf chalk quarry in northern Germany.