1Blackhawk Geometrics, Inc., Golden, CO
2Lynn, Inc., Houston, TX
Abstract: Fractured Tight Gas Reservoir Seismic
Exploration
Optimization
Introduction
Gas yields in the Rocky Mountain region are often controlled by
natural fractures. The Department of Energy has sponsored a series
of investigations to develop cost-effective ways to detect regions
of high fracture density with seismic
and other methods. Here we
present a preliminary synthesis of three of these projects. We
recommend a protocol for fractured-reservoir
seismic
exploration
focusing on all-azimuth, wide-angle P-wave surveying, with
multiazimuth processing aimed at highlighting fracture-zone
anisotropy
.
Background
Fractures cause changes in a rock's stiffness; when regular
fracture sets are present, the additional order may be detected as
anisotropy
of the
seismic
response. The simple and not uncommon
case of a dominant set of parallel subvertical fractures leads to
azimuthal
anisotropy
, that is, a variation of the
seismic
response
with direction. For example, both pressure (P) and shear (S) waves
are slower in the fracture-perpendicular direction than fracture
parallel. Therefore multiazimuth
seismic
processing—dividing
the source-to-receiver raypaths by direction, processing each
direction independently, and examining the differences—can be
used to detect zones of intense fracturing.
S-waves are particularly sensitive to anisotropy
, both in the
overall change in velocity and by splitting into a slow wave
(S2) with transverse particle across fractures and fast
waves with particle motion in the fracture plane (S1).
However, full S-wave recording (shear sources and three-component
geophones) is still costly. A major aspect of our investigations
has been comparison of P- and S-wave responses in the field to test
whether most of the fracture information can be determined using
cheaper P-wave technology alone. Furthermore, P-waves contain
additional information on pore and fracture content (gas vs.
water). P-wave properties of interest include azimuthal variations
in velocity, reflectivity (including Amplitude Variations with
Offset, or AVO), and frequency.
Powder River Basin
A series of orthogonal, intersecting 2D seismic
lines were shot
approximately aligned with the principal fracture trends. Both P-
and full S-wave data were acquired on all lines. Spatial variations
in shear-wave
anisotropy
delineate fracture zones, which were
confirmed by drilling. This study validated S-wave methods for
fracture-zone mapping.
Uinta Basin
Two intersecting 2D seismic
lines were shot at the
Bluebell-Altamont field (Lynn et al., 1996a), approximately
parallel and perpendicular to the principal fracture trend. Both P-
and full S-wave data were acquired on both lines, and S-wave
anisotropy
was calibrated by a 9C VSP (nine-component, vertical
seismic
profile, i.e., three-component receivers and both P- and
S-sources). This study, in a comparatively simple geological
environment (fractured basin), demonstrated that P- and S-wave
responses indeed correlated: the P-wave AVO gradient at the line
intersection was proportional to the S-wave velocity
anisotropy
within corresponding vertical intervals.
Piceance Basin
Aeromagnetic mapping and a 3D P-wave survey were performed at
the Rulison field. Two large basement faults were identified from
the aeromagnetic data, which appear to terminate in seismic
cross-sections as “fault-tip fracture clusters.” These
highly fractured zones correlate with P-wave velocity
anisotropy
and appear to control the “fairway” for gas production in
this moderately complex environment (faulted and fractured
basin).
Wind River Basin
Three seismic
surveys were performed (Lynn et al., 1996b; Grimm
et al., 1998): a large 3D P-wave survey, a 9C VSP, and a smaller 3D
P-to-S survey (the last involves three-component recording of
P-sources, with specialized processing to extract S-waves generated
by reflections). The geology, a faulted, fractured anticline, is
the most complex of the three study areas. The best correlations of
seismic
attributes with gas yield are achieved by considering
azimuthal variations. However, most of the information is contained
in the fracture-parallel component for reflectivity and frequency,
and in the fracture-perpendicular component for velocity. We
interpret these results to indicate that strong scattering in
fracture-perpendicular raypaths degrades the information in
reflectivity and frequency, whereas this orientation is optimal for
mapping relative fracture density from velocity variations: slower
fracture-perpendicular velo-city is associated with higher fracture
density. Decreases in reflectivity correlated with pay are
interpretable as decreases in the impedance contrast of local
gas-charged sand bodies with surrounding shales, whereas the
increase in frequency with pay requires a more complex permeability
anisotropy
model. Using a neural network, the
seismic
azimuthal
variations in reflectivity, frequency, and velocity were combined
with the dominant geological attribute (structural altitude on the
trapping anticline) to map the estimated potential for commercial
gas pay throughout the survey area.
The P-wave velocity anisotropy
, interpreted as proportional to
fracture density, was corroborated by the results of the S-wave
experiments, although some significant spatial differences were
found.
Synthesis
For basins of only moderate structural complexity, such as the
Piceance, there is a strong correlation between fracturing (as can
be inferred from P-wave azimuthal anisotropy
) and gas yield. In the
more complex structure in the Wind River basin, faulting and
fracturing are more ubiquitous: while fractures are still important
in delivering gas to wells, the
seismic
data reveal that mapping
reservoir units in these heterogeneous sediments has a renewed
importance.
For cost-effective seismic
evaluation of fracture density, we
recommend:
(1) Gather sufficient reconnaissance data (field and borehole
geology, in-situ stress, remote sensing, etc) to properly plan
seismic
surveys.
(2) Acquire 3D P-wave surveys with maximum offsets greater than or equal to target depth in all azimuth, using isotropic source and receiver arrays.
(3) Processing in as many azimuths as allowed by cost; 2 or 4 should be sufficient.
(4) Calibrate with limited shear or converted-wave data: a
9C-VSP or live patch of 3C geophones, to document S-wave
anisotropy
.
For refined prediction of gas yield in fractured reservoirs, we recommend:
(1) Exploit higher sensitivity of reflection attributes at large incidence angles (offsets) to better distinguish gas from water. This may require a new paradigm for amplitude-variation-with offset (AVO) modeling (Grimm and Lynn, 1997).
(2) Improve mapping of seismic
attributes to gas yield (e.g., as
in neural network) by fully integrating
seismic
data with well logs
in three dimensions, particularly those new logging tools like
nuclear magnetic resonance that may best distinguish gas from
water.
Acknowledgements
These projects were funded by the Department of Energy Federal Energy Technology Center. Stanford University and Advanced Resources International were the prime contractors for the Powder River basin and Piceance basin studies, respectively.
AAPG Search and Discovery Article #90928©1999 AAPG Annual Convention, San Antonio, Texas