CHARACTERIZATION OF IN-SITU STRESS AND PERMEABILITY IN FRACTURED RESERVOIRS

Expanded details and additional results are presented on two methods for estimating fracture orientation and density in subsurface reservoirs from scattered seismic wavefield signals. In the first, fracture density is estimated from the wavenumber spectra of the integrated amplitudes of the scattered waves as a function of offset in pre-stack data. Spectral peaks correctly identified the 50m, 35m, and 25m fracture spacings from numerical model data using a 40Hz source wavelet. The second method, referred to as the Transfer Function-Scattering Index Method, is based upon observations from 3D finite difference modeling that regularly spaced, discrete vertical fractures impart a ringing coda-type signature to any seismic energy that is transmitted through or reflected off of them. This coda energy is greatest when the acquisition direction is parallel to the fractures, the seismic wavelengths are tuned to the fracture spacing, and when the fractures have low stiffness. The method uses surface seismic reflection traces to derive a transfer function, which quantifies the change in an apparent source wavelet propagating through a fractured interval. The transfer function for an interval with low scattering will be more spike-like and temporally compact. The transfer function for an interval with high scattering will ring and be less temporally compact. A Scattering Index is developed based on a time lag weighting of the transfer function. When a 3D survey is acquired with a full range of azimuths, the Scattering Index allows the identification of subsurface areas with high fracturing and the orientation (or strike) of those fractures. The method was calibrated with model data and then applied to field data from a fractured reservoir giving results that agree with known field measurements. As an aid to understanding the scattered wavefield seen in finite difference models, a series of simple point scatterers was used to create synthetic seismic shot records collected over regular, discrete, vertical fracture systems. The model contains a series of point scatterers delineating the top tip and bottom tip of each vertical fracture. When the shot record is located in the middle of the fractured zone and oriented normal to the direction of fracturing, a complicated series of beating is observed in the back scattered energy. When the shot record is oriented parallel to the fracturing, ringing wavetrains are observed with moveouts similar to reflections from many horizontal layers. These results are consistent with the full 3D elastic modeling results. An AVOA analysis method was refined and applied to a field data set. An iterative, nonlinear least squares inversion that uses the Gauss-Newton method and analyzes the full range of azimuths simultaneously was employed. Resulting fracture location and strike orientation estimates are consistent with other fracture information from the area. Two modeling approaches for estimating permeability values from seismically derived fracture parameters have been investigated. The first is a statistical method that calculates the permeability tensor for a given distribution of fractures. A possible workflow using this method was tested on fracture distributions obtained from the Transfer Function-Scattering Index analysis method. Fracture aperture and length estimates are needed for this method. The second method is a direct flow model of discrete fractures and fracture networks using a computational fluid dynamics code. This tool provides a means of visualizing flow in fracture networks and comparing expressions for equivalent fracture aperture flow to the actual flow. A series of two dimensional models of fractures and fracture networks, as well as a 3-D model of a single rough fracture, were tested