A system of statistics-based rules for extending surface fracture models

Abstract

A fracture network model represents interconnected discontinuities within a solid medium. These can be used for simulating hydraulic flow through fractured rock, which is a critical concern in selecting nuclear waste repository sites. In the early phases of site selection, only an incomplete model of surface features is available, often insufficient for a meaningful hydraulic analysis. In cases such as these, extending the known data into larger plausible stochastic models can satisfy the needs of hydraulic simulation to provide meaningful insight. This thesis outlines a procedure that uses analysis of an incomplete dataset of known surface features, and uses its statistical clues as conditioning constraints for producing surface fracture models that are larger, and more detailed. The models are evaluated in how well they honour the geostatistics-inspired constraints and characteristics derived from the source data, and the results are successful in reflecting many of these statistical inputs.