uGeneral statement

uFigures

uData-processing

uInterpretation

uFractures

uConclusion

uAcknowledgments

uGeneral statement

uFigures

uData-processing

uInterpretation

uFractures

uConclusion

uAcknowledgments

uGeneral statement

uFigures

uData-processing

uInterpretation

uFractures

uConclusion

uAcknowledgments

uGeneral statement

uFigures

uData-processing

uInterpretation

uFractures

uConclusion

uAcknowledgments

uGeneral statement

uFigures

uData-processing

uInterpretation

uFractures

uConclusion

uAcknowledgments

uGeneral statement

uFigures

uData-processing

uInterpretation

uFractures

uConclusion

uAcknowledgments

uGeneral statement

uFigures

uData-processing

uInterpretation

uFractures

uConclusion

uAcknowledgments

uGeneral statement

uFigures

uData-processing

uInterpretation

uFractures

uConclusion

uAcknowledgments

Data-Processing Workflow

The data-processing workflow that was implemented resulted in high-resolution C-wave images. An important byproduct of this workflow was information related to fracture orientation and to fracture intensity. The first step in the data processing was to rotate horizontal components from their recorded field orientations to their source-receiver azimuths, or “radial,” directions. If seismic data are azimuthally isotropic, all C-wave reflection energy would be concentrated on the radial component, and data acquired by the transverse sensor could be discarded. However, after rotating the XinChang data to radial/transverse coordinates, a significant amount of C-wave energy remained on the transverse component.

This data behavior confirmed the presence of shear-wave splitting, which occurs when a shear wave encounters an azimuthally anisotropic layer such as one of the XinChang fractured reservoir units. In S-wave splitting, the C-wave polarizes into two new waves, a phenomenon known as “birefringence.” One of the new split waves is polarized parallel to, and the second orthogonal to, the fracture orientation. The velocities of the two waves differ – the faster wave being polarized parallel to the fractures, and the slower wave polarized orthogonal to the fractures (Figure 1). In addition, each of the new waves splits again when it encounters a new anisotropic layer, resulting in a complicated mix of waves arriving at each sensor. Shear-wave splitting can yield valuable information regarding fractures; however, unless addressed correctly, wave-splitting reduces the bandwidth of stacked or migrated C-wave images. Thus, our data-processing workflow was modified to capitalize on the shear-splitting that was detected.

After sensor rotation to radial/transverse coordinates, the main components of the workflow included surface wave attenuation, resolution of shear-wave statics, surface-consistent signal processing and Q compensation. These steps were performed independently on the radial and transverse components. Next, the data for each component were subdivided into 36 10-degree, azimuth sectors. Each azimuth sector was migrated separately via a prestack C-wave time migration. This migration step required a velocity model for both P and S wavefields. Because it is difficult to derive shear-wave velocities from converted-wave data, we developed a novel scheme in which P and S velocity fields were re-parameterized into new variables that could be estimated from the C-wave data.

Following migration, each sector volume was subjected to residual move-out correction, muting and stacking. The azimuth volumes were then re-assembled into azimuth-sector gathers for each migration bin. For any migrated output location, a C-wave reflection has a characteristic signature on the radial and transverse components as a result of interference patterns between the polarized fast and slow waves. Typically, radial data have a sinusoidal type of behavior with azimuth, while transverse data exhibit polarity reversals every 90 degrees of azimuth (Figure 2). If these two C-wave responses were stacked as is, the result would be a low-resolution image (Figure 3a).

The most important step in the data processing was our layer-stripping anisotropic correction. This procedure removed the effects of shear-wave spitting at each anisotropic boundary by simulating the effect of an isotropic medium. A single anisotropic layer was stripped to form an isotropic layer in the following manner:

Step 1. Knowing that azimuths corresponding to polarity reversals observed on the transverse component define fracture orientation, these azimuth angles were used to rotate the data from radial/transverse to fast/slow directions.

Step 2. A cross-correlation between fast and slow data determined the time lag between these two wave modes; a static correction was then done to time align slow and fast data.

Step 3. An additional rotation back to radial and transverse coordinates concentrated all of the energy onto the radial component and produced azimuthally isotropic data. These adjusted data were stacked to form a high-resolution C-wave image (Figure 3b).

An important byproduct of step one is fracture orientation. For any particular layer, maps of fracture orientation throughout the entire data volume were generated using the azimuth angles determined in this data-processing step. Second, the amount of fracturing is related to travel-time differences between fast and slow shear-waves, and the cross-correlation in step two yielded time-difference information that was used to infer fracture intensity.

Post-Processing/Interpretation

Upon completion of the 3D3C seismic data processing, we initiated post-processing and interpretation activities. This portion of the workflow involved integrating available well, outcrop, and core data with the processed 3D3C seismic data. The post-processing included:

• Acoustic, shear and elastic inversions.
• Generation of seismic attributes.
• Shear-wave splitting analysis to help define lithology, stratigraphy and fracturing.

These steps built better structural and stratigraphic models, mapped fracture patterns and intensity, and provided an improved understanding of the region’s geologic and tectonic history.

The Sichuan Basin underwent dramatic subsidence rates during the early Mesozoic. Burial of the reservoir to depths of 20,000 feet or more occurred soon after deposition and before gas was generated in surrounding and underlying source rocks. The resulting compaction reduced reservoir porosities to less than 4 percent, causing reservoir rocks to be almost impermeable. As a result, production in XinChang Field is fracture-dependent.

The fracture network:

• Made gas charge possible.
• Created a major part of the gas storage capacity.
• Is the mechanism by which gas stored in matrix porosity can be accessed during production.

Interbedded sand-shale sequences are the best exploration targets – these thinner-bedded, brittle layers fracture more easily and with higher density than do their thicker counterparts. The integration of geological history with production data resulted in a model showing that storage capacity in the reservoir depends on interconnection of fractures in fault-damage zones and on the connections that these damage zones make with naturally fractured sandstone reservoir beds.

The densely sampled, full-azimuth P-wave data acquired in this study supplied higher frequencies than existed in legacy seismic data, resulting in an improved structural picture with excellent fault resolution. However, because sands and shales had similar acoustic impedances, lithology could not be determined from P-wave data alone. Fortunately, converted-shear (C-wave) data were valuable for discriminating lithology in these rocks and delivered vital insights into the stratigraphic architecture of targeted reservoirs. The C-wave data provided a means to define interbedded, sand-shale sequences that were areas of optimal fracture intensity and enabled delineation of source rock (Figure 4).

Identification of Fractures

Because fractures dominate storage and movement of gas in XinChang Field, developing a tool to identify and map the best fractured zones was a high priority. The C-wave dataset proved to be that needed tool, because the data provided azimuthal definitions of S-wave velocity differences that could be used to map variations in fracture orientation and intensity.

As illustrated in Figure 5, seismic wave propagation is minimally affected parallel to the dominant fracture trend in rocks that have a simple one-directional fracture system (the “fast direction,” V_fast), but a maximum velocity reduction (the “slow direction,” V_slow) is aligned perpendicular to the oriented fractures. With multiple sets of fractures, such as the third orthorhombic case in Figure 5, velocity is reduced in all directions, and V_fast approaches V_slow, resulting in this type of fractured volume appearing to be an isotropic medium. This model implies multidirectional, interconnected fracture sets should be located in areas where there are smaller amounts of anisotropy and also reduced C-wave velocity. This information can be utilized to search for well locations that will penetrate multi-directional fracture zones.

As described above, fracture orientation was determined by deriving the azimuths of V_fast from shear-wave splitting analyses performed in a layered-Earth approach. In the XinChang survey area, regional geology, borehole breakout and FMI log results all indicate that the current maximum horizontal stress is oriented along azimuths of 80-110 degrees. This information validated the shear-wave splitting results for orientation in the uppermost Earth layer, which has an average orientation of 95 degrees (Figure 6).

It also is important to note that fractures oriented close to this principal stress direction are more likely to be open, though the extreme overpressure in this area keeps other fractures open as well. Knowing fracture orientation in zones of higher anisotropy, where a single set of parallel fractures is more likely to exist, can help in designing directional or horizontal wells that will intersect more fractures, yielding higher production in these areas.

Fracture density was determined from the analysis of the shear-wave splitting, specifically by measuring the time difference between reflections observed in V_fast and V_slow shear volumes using Transform software. Figure 7 shows a display of the V_fast – V_slow time difference data (in color) overlain on a similarity plot. Note that the light green areas, where the best producing wells are located, show less anisotropy than do areas with less productive wells and also correlate with the slower V_fast velocity zones (as predicted above). FMI logs available in some of these wells confirm multi-azimuthal fractures, not a single set of oriented fractures, are present in the better producers and support the model of less anisotropic behavior with dual direction fractures.

Conclusion

Insights from similarity processing, curvature attributes and shear-wave splitting analysis provided three independent fracture density measures that were integrated into discrete fracture network (DFN) models and fracture maps. The integration of interpretations from all disciplines – outcrop analysis, seismic, core, well log and well production data – enabled the interpretation team to select 19 new well locations – three of which have been drilled and completed as producers. One of these new wells is the most productive well in the area. The Xin-2 and Xin-3 wells were mentioned in the January 2008 AAPG Explorer as two of the most significant wells in the Far East, producing 18 mmcfgd and 8.2 mmcfgd, respectively, from a 500-foot gas column.

Acknowledgments

The authors thank the management of Sinopec and Southwest Petroleum Branch for granting permission to present this paper – especially Xu Xiangrong, president of Southwest Petroleum Branch Company, for his leadership and commitment to cutting-edge technologies. The authors also thank the other contributors to the interpretation project, including AAPG member Roger Palomino and Doug Allinson, Felix Diaz, Reinaldo Nossa, Santi Randazzo and Jim Simmons.

Return to top.