ABSTRACT: Rock fracture characterization is essential for several applications, which include geothermal energy extraction, hydraulic fracturing and radioactive waste disposal. Unfortunately, observing rock fractures is not possible, therefore, methods were developed to use site-specific observable data such as borehole logs and exposed outcrops to predict fracture networks. In this work, we present EllipFrac, a model and software developed for modeling natural rock fracture networks stochastically. The main input parameters of the model are observable/ measurable parameters, more precisely, the accumulated fracture area per unit volume, ?? and the mean and standard deviation of fracture area, ??[??] and ?? respectively. EllipFrac assumes that fractures are degenerate ellipsoids, whose geometric characteristics are stochastically generated from probability density functions (PDFs) developed in this work. In EllipFrac, fractures’ elongation is directly controlled by the user and the Monte Carlo method is used to quantify uncertainty. EllipFrac is also equipped with a novel analytic method that can be used to determine fractures intersections and can heuristically estimate the number of connected paths between any set of two points (fractures) within the reservoir. Moreover, EllipFrac is capable of deciding whether any set of two points in the reservoir are connected. The latter two functionalities take advantage of graph theory related concepts. EllipFrac is implemented in MATLAB® and can run in serial or in parallel mode for computational efficiency.

This content is only available via PDF.
You can access this article if you purchase or spend a download.