Quantitative data integration for fracture characterization using statistical rock physics

The goal of this dissertation is to design a methodology for quantitative integration of geological information with seismic data using rock physics theories, in the framework of an inverse problem. I illustrate this method with fracture characterization of hydrocarbon reservoirs. Fractures are of great importance in oil industry, because significant amounts of hydrocarbons are trapped in tight reservoirs. In such formations, natural fractures are the main factors controlling fluid flow.; There are different types of information that can be used to study fractures, such as geologic, seismic, and well-log, and each one of them contributes in a different way to fracture characterization. Thus, by using these different sources of information, we can better constrain the predictions on the fracture parameters, such as fracture density and orientation. The various types of data from geology and seismic can be combined quantitatively if we translate them into the common language of probability theory. This probabilistic approach allows us to integrate quantitatively the various types of information and to estimate the uncertainty in our predictions.; I demonstrate this methodology with a fractured carbonate reservoir in eastern Texas. The fracture orientation is determined from seismic data, based on rock physics theories. I use a bootstrap method to estimate the uncertainty in the fractures' strike, due to seismic measurement errors. The fracture density is determined using rock physics theories, by integrating quantitatively the prior information, obtained from the geological interpretation of a fault at the top of the reservoir, with reflectivity attributes derived from a 3D seismic data set. I emphasize the uncertainty in the fracture density, and the relative impact of the prior geologic information in comparison with the seismic information on the final results for fracture density distribution.; This integration methodology provides a framework for quantitatively combining diverse data into one consistent result for subsurface rock properties, with an estimate of uncertainty associated with it. Therefore, this method facilitates an informed decision-making process for reservoir management.