Characterizing fracture networks: an effective approach using seismic anisotropy attributes

The development of naturally fractured reservoirs is significantly influenced by the characteristics of the fracture network since these control the volume and flow direction of the fluid through the reservoir rocks. The presence of fractures can be very beneficial since knowledge of their characteristics allows the design of well paths that intersect a larger number of permeable fractures, thus increasing production. It also enables optimized placement of injectors for improved sweep efficiency and better control of the reservoir pressure. However, fractures can also be damaging for the economic potential of a field. They can create preferential flow paths which may lead to premature water breakthroughs or, conversely, act as barriers to impede production. A good understanding of the fracture network in terms of intensity, orientation, and spatial distribution is therefore essential for improved reservoir development. Some fracture information is available from core observations and image log interpretation, but these data are only valid in the vicinity of the borehole, and when extrapolated beyond this, can lead to erroneous prediction of the overall reservoir dynamics. Even though geostatistical methods can help to reduce the uncertainty associated with spatial predictions by taking into account the geological heterogeneities, a true 3D attribute is necessary in order to accurately characterize a fractured reservoir. Following the observations of Crampin (1985a, 1985b), Crampin et al. (1986), Thomsen (1988), and others, it is now widely recognized that fracture systems are often found to be aligned in a preferential direction. This induces a directional (or azimuthal) dependence of seismic properties such as traveltime, velocities and reflection amplitudes. This directional dependence, also referred to as anisotropy, can cause seismic shear waves to split in preferential directions related to the alignment of the fractures: the fast shear waves being polarized parallel to the fractures and the slow shear waves polarized perpendicular to the fractures. It can also affect the amplitude of compressional waves depending on the azimuthal direction of propagation. An analysis of the anisotropy effects observed in 3D seismic data can therefore provide insight into the fracture characteristics (Thomsen, 1995; Lynn et al., 1995, 1996). Methods based on shear-wave splitting analysis are well established (Crampin, 2000), but unfortunately shear wave data are relatively expensive to acquire and process. As a result, in the last few years there has been a growing interest in P-wave azimuthal amplitude variation. Rüger and Tsvanskin (1997) demonstrated that reliable estimates of the anisotropy parameters could be obtained from the P-wave amplitude. Later, the approximations of the P-wave reflection coefficient presented by Rüger (1998) were extended into a linearized form by Jenner (2002) and used by many authors for inferring fracture properties. Angerer et al. (2003a) then went on to propose an integrated workflow for seismic fracture characterization from wide-azimuth large offset P-wave and PS data. This approach (Figure 1) includes a state-ofthe- art geostatistical decomposition technique combined with azimuthal anisotropy analysis and yields quantitative estimates of fracture intensity and direction. It can be applied either to horizon-based attributes, such as interval velocity maps and RMS amplitude maps, or to 3D attribute volumes such as seismic amplitude and AVO attributes. Here we present the results of a recent fracture characterization study based on P-wave azimuthal anisotropy from a wide-azimuth 3D land dataset. The goal of the study was to characterize the fracture distribution in a carbonate reservoir using several attributes such as interval velocities, RMS amplitude and seismic amplitude. In addition, to confirm the validity of the estimated anisotropy attributes, this study also included an analysis of the fractures using FMI/FMS log data from six wells. Several comparisons of anisotropic and conventional processing illustrate the benefits of our approach at different stages of the processing. Finally, we review the accuracy and efficiency of each seismic attributes in terms of characterizing the fractures orientation, magnitude and distribution.