We apply a new method for birefringence analysis at depth to a tight gas reservoir in Rulison Field, Colorado. The new method is based on a multicomponent version of the virtual source method in which VSP receivers are turned into virtual shear sources in the zone of interest. This allows accurate detection of even small amounts of azimuthal anisotropy under complex overburden where traditional methods fail. We used the new method to measure shear wave splitting of less than 1% in a reservoir under significantly anisotropic overburden. That was enough to infer fracture orientation, which turned out to be close to the orientation interpreted from FMI logs.
Rulison field in the Piceance basin, Colorado (Figure 1), produces gas from a low-permeability interbedded sequence of sand and shales in the Williams Fork formation of the Mesaverde group. Production is thought to be controlled by interconnected natural vertical fractures (Lynn, 1999). Hence, to optimize field development, it is necessary to map fracture distribution and azimuth. This can be facilitated by measuring azimuthal anisotropy in the reservoir from VSP data in key wells. In particular, we would like to study shear wave splitting and polarization in the reservoir because these quantities can be related to fracture density and orientation. The problem is that the overburden at Rulison is complex and exhibits azimuthal anisotropy stronger than that in the reservoir. Its influence on shear waves would need to be removed before studying the reservoir. Traditionally that would be done through layer-stripping of 2C x 2C VSP data (Winterstein and Meadows, 1991; Thomsen et al., 1999). However, if the overburden symmetry axes vary with depth, as they presumably do at Rulison, we would need to instrument the well all the way from the surface to the reservoir for proper layer stripping. Given that the reservoir is at more than 4500 ft depth, that is unfeasible.
To circumvent this problem, a new approach to studying shear wave splitting at depth was proposed by Bakulin and Mateeva (2008). It involves turning the horizontal components of VSP receivers into orthogonal shear virtual sources in the borehole – or, essentially, redatuming the 2C x 2C VSP to a 2C x 2C virtual data set with sources and receivers in the borehole. Then, signals from these new shear sources recorded at the horizontal components of other receivers can be used to study shear wave splitting in the medium between and below the receivers. The most remarkable feature of this method is that we do not need to know anything about the overburden to create virtual sources in the borehole. Therefore, we can use it to probe a layer of interest below any heterogeneous and anisotropic overburden.
Let us start by a brief explanation of how multicomponent virtual shear sources are created in the borehole. For more on the fundamentals of the virtual source method see Bakulin and Calvert (2004, 2005), and for more details on this particular application see Bakulin and Mateeva (2008) or their companion abstract in this book.