SPE Reservoir Evaluation & Engineering
Volume 11, Number 4, August 2008, pp. 719-725

Summary

Seismic anisotropy is sensitive to the intensity and orientation of fractures aligned by tectonic or local stress fields. Azimuthal anisotropy of amplitude vs. offset (AzAVO) is one common measure of seismic anisotropy for fracture detection. For this study, we calculated P-wave AzAVO attributes for a Lower Cretaceous limestone reservoir in east Texas. Our study tests AzAVO inversion for the relatively low-fold, fair data quality typical of some land surveys. We developed a workflow to integrate the AzAVO inversion with rock physics, forward seismic modeling, seismic-scale fault interpretation, image logs, and cores.

Our AzAVO inversion method was based on Rüger’s equation, which describes the observed reflectivity (R) as a function of incidence angle (θ_i ) and acquisition azimuth (∅i) to equal a function of normal incidence reflectivity (A₀), amplitude vs. offset (AVO) slope (B₀), anisotropy magnitude (B₁), orientation of symmetry of anisotropy (∅₀), and noise (n_i ):

R(θ_i , ∅_i ) ≈ A₀ + [B₀ + B₁ cos² (θ_i - ∅₀ )] sin²θ_i + n_i

Inversion of Rüger’s equation yielded four attribute volumes (corresponding to A₀ , B₀ , B₁, and ∅₀ ). Additionally, we proposed an additional data-quality parameter that describes the relative amount of the total reflection energy predicted by the model. Preinversion processing carefully preserved signal via random noise attenuation, superbinning, bandwidth balancing, and phase alignment.

In general, AzAVO inversion displayed geologically interpretable high-anisotropy anomalies. Noise, however, limited our confidence for quantitative fracture prediction. On the flanks of the anticline, data quality was high (greater than 50% fit of observed to predicted amplitude). Here the magnitude of the anisotropy (B₁) volume showed strong lineations parallel to northeast-southwest trending faults. Over the crest of the anticline, the inversion quality was degraded by an overburden effect, coincident with poor data quality (less than 50% fit). AzAVO orientation maps were noisy and in some places obscured by an acquisition footprint. Locally, however, AzAVO orientations were parallel to east-northeast fracture orientations from cores, maximum horizontal stress from image logs, and fast-velocity direction from a dipole-sonic log.

Introduction

A network of open, connected fractures increases the permeability of many reservoirs, especially carbonate reservoirs, worldwide. Understanding the distribution of fractures helps us to predict the spatial variations in production rates and flow anisotropy. Geologic fracture characterization tends to be limited by the number of cores or image logs, as well as sampling biases inherent to measuring vertical fractures in vertical wells. Seismic anisotropy is a fieldwide fracture detection technique most sensitive to the open fractures that are important for flow.

We define seismic anisotropy as the directional response of any seismic attribute, such as travel time or amplitude. Theoretical models show that both travel time and amplitude vary with azimuth in rocks with vertically aligned fractures (Rüger 1996, 1998; Tsvankin 2001; Grechka et al. 1999). Azimuthal normal moveout (AzNMO) detects anisotropy by use of the normal moveout travel time attribute of a prestack common midpoint (CMP) gather. We focus on AzAVO, which detects anisotropy from the amplitude variation of a particular event, such as Primary-wave (P-wave) or converted shear-wave (S-wave). Previous seismic models have predicted that shear-wave data may have a clearer anisotropic response than P-wave data (Lynn and Thomsen 1990; Li 1997). This study, however, considered only P-wave anisotropy. Because of the wide availability of P-wave seismic data, we expect the development of P-wave anisotropic techniques to have a broad business impact.

There are several known limitations of azimuthal anisotropy for fracture detection. Seismic waves respond to open fractures, so they are relatively insensitive to fractures closed by stress or cementation. In a similar manner, if fracture density is low, fracture-induced anisotropy is too weak to detect. Fracture sets in multiple orientations may diminish the magnitude of the azimuthal anisotropy. Azimuthal anisotropy is enhanced for vertical fractures and reduced for more shallow dipping fractures. Steeply dipping beds may induce an azimuthal anisotropy. Anisotropy at a reservoir level may be influenced by the anisotropy of rocks that overlie it. Finally, like any seismic attribute, azimuthal attributes are sensitive to seismic data quality. Interpreting azimuthal attributes requires understanding the geologic sources of anisotropy, the effects of azimuthal and offset distribution, the assumptions of the data processing, and the effects of noise.

For this study, our objectives were the following:

1. Develop AzAVO algorithms and workflows for fracture detection.
   
2. Apply the workflows for a fractured carbonate reservoir.
   
3. Test the workflows on a seismic survey with fair land P-wave seismic acquisition and data quality.
   
4. Integrate the seismic anisotropy results with mapped structures, stacked data attribute volumes, dipole sonic logs, fractures in cores and image logs, and production data.

Insight gained from this study may apply to other fractured carbonate reservoirs worldwide.

© 2008. Society of Petroleum Engineers

View full textPDF ( 2,122 KB )

History

• Original manuscript received: 11 September 2006
• Meeting paper published: 5 November 2006
• Revised manuscript received: 1 November 2007
• Manuscript approved: 20 March 2008
• Version of record: 20 August 2008