Vous êtes sur la page 1sur 9

SPECIAL

T i g h t SECTION:
g a s s a n dT
Rsieg
vh
e rts g
e atsi m
sa
e nmdisg r a t i o n

Downloaded 09/09/14 to 189.225.48.41. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

6HLVPLFSHWURSK\VLFVDQGLVRWURSLFDQLVRWURSLF$92PHWKRGV
IRUXQFRQYHQWLRQDOJDVH[SORUDWLRQ
BILL GOODWAY and MARCO PEREZ, Apache
JOHN VARSEK, Cenovus
CHRISTIAN ABACO, EnCana

xploration and drilling for natural gas in North America


has moved radically away from conventional reservoirs
to focus on unconventional reservoirs such as tight gas sands
and shales. These reservoirs have low porosity and near-zero
permeability with gas stored in natural fractures and within
the matrix porosity. Economic gas production requires
hydraulic fracture stimulation to open connections to
existing natural fractures or matrix porosity, and successful
stimulation depends on the formations geomechanical
brittleness being capable of supporting extensive induced
fractures. However, despite adequate stimulation, significant
variations exist between wells in expected ultimate recovery
(EUR) due to the heterogeneity of these resource plays.
Consequently, predicting natural fractures or fracture-prone
sweet spots is essential to optimize development of such
plays.
This paper describes some rock physics, rock mechanics,
and conventional isotropic AVO methods that can be applied
to map these tight reservoirs and attempts to clarify some
fundamental ambiguities involved in 3D amplitude variation
with azimuth (AVAZ) inversion to correctly detect the orientation and intensity of anisotropy due to stress or fractures.
Mechanical characteristics of tight gas reservoirs from a
seismic perspective
As tight gas sands or shales have low porosity and generally
contain at least some gas, the application of AVO methods
is simplified to primarily extracting lithology and mechanical properties from seismic data. Commercial production of
gas shale is in its infancy in Canada, but is well established
in plays such as the Barnett Shale of Texas. Consequently,
the mechanical characteristics of the Barnett Shale are used
as an analog to help guide AVO techniques in identifying
the resource potential of similar shales. The engineering literature focuses on identifying the optimum Barnett Shale
mechanical properties, high Youngs modulus (E) and low
Poissons ratio (v) as shown by the linear empirical trend in
Figure 1 (Grigg, 2004). In order to better understand the
significance of E and v, these parameters are converted to the
simpler and more seismically intuitive Lam parameters of
lambda, h, (which can be termed incompressibility) and its
orthogonal counterpart mu, + or rigidity (Goodway et al.,
1997, 2001) by using the following relationship:
(1)
An initial observation from Equation 1 is that the inverse
linear relation between increasing E and decreasing v shown
in Figure 1 is simply the result of an increase in rigidity, +.
1500

The Leading Edge

December 2010

Figure 1. Empirical Youngs modulus-to-Poissons ratio relationship for


the Barnett Shale (Grigg, 2004).

Figure 2. Lambda versus Youngs modulus crossplot with curves of


constant mu and lines of constant Poissons ratio.

This empirical trend is easier to understand in terms of rigidity than as a combination of Youngs modulus and Poissons
ratio, because an increase in rigidity enables a shale to sustain
natural or induced fractures.
Equating h and + to E and v in order to compare geomechanical properties on equivalent log crossplots is not so
simple, as E has a nonlinear relationship to h and +. However, a useful reference plot that relates E versus h to curves
of constant + and lines of constant v is shown in Figure 2.
Griggs empirical trend between v and E (shown as the Barnett Trend in Figure 2) can now be seen to have two parts.
One, representing brittle rocks, is perpendicular to increasing
curves of constant + and crosses lines of decreasing v; the
other trend bends toward the origin, and represents ductile

Downloaded 09/09/14 to 189.225.48.41. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Tight gas sands

rocks (e.g., coals) with very low + values (< 4 GPa) that approach a constant
Poissons ratio of 0.34. The Barnett
Shale ranges shown in Figure 1 (E =
45 MMpsi and v = 0.20.3) translate
to cutoff ranges of moderately high rigidity (+ = 1016 GPa) and h/+ ratios
between 0.78 and 1.08 as shown in
Figure 2.
The transformations above guide
the petrophysics and AVO used to interpret the results of prestack inversion
with the goal of identifying similar
gas shales in the context of a standard
LambdaRho (h) and MuRho (+)
crossplot template shown in Figure 3
(Goodway, 2009). This template compares various shale and carbonate lithologies from Western Canada to two
published Barnett Shale values whose Figure 3. LambdaRho versus MuRho crossplot comparing various shales and carbonates from
relative EURs (Simon, 2005) are indi- Western Canada to the Barnett Shale (with background pure mineral points and lines of
cated on the grid of mineral reference constant Poissons ratio and P-impedance).
points. The arrows on this template
also indicate the directions for lithology and increasing porosAll these effects have an unambiguous shift to low hl vality. Both Western Canadian (red and green) and the Barnett ues, making this the best attribute to map the most prospec(yellow) gas shales occupy a space away from ductile shales tive zones.
due to their higher quartz content which results in an increase
in rigidity mu (or MuRho) and decrease in Poissons ratio Minimum horizontal closure stress
that is consistent with Griggs relation. However, MuRho and A different geomechanical perspective, of engineering oriPoissons ratio alone do not clearly distinguish gas shales from gin, describes a rocks ability to support natural or induced
calcareous shales (light blue) that are barriers to hydrofrac fractures as being governed by the minimum horizontal clostimulation, as they share a similar narrow range of values.
sure stress shown in Equation 2. This stress is defined as the
Leon Thomsen, in his 2002 Distinguished Instructor minimum pressure required to open a pre-existing fracture
Short Course, discounted the incompressibility parameter, h. or plane of weakness and is a function of E and v, giving rise
He wrote, in the companion book to the course, that "h does to the emphasis on these properties for the Barnett Shale.
not have a common name, since it is not useful for much in
geophysics (despite what you might have heard!). However,
(2)
despite Thomsens claim, we feel that h, or better still hl,
merits consideration as the best discriminator of gas shales
The closure stress equation can be written in terms of the
from ductile shales, unfracable calcareous shales, and carbon- more practical seismic moduli h and + in a similar manner as
ates. This is clearly seen by the horizontal shift to low Lamb- used above to convert E and v:
daRho values with a near doubling in relative EUR for the
3
1
2
Barnett (yellow points 3053) in Figure 3. A similar decrease
in LambdaRho shown by the yellow arrow also separates the
(3)
red from green clusters of Western Canadian gas shales. This
can be explained as follows:
were = horizontal (minimum closure) stress, = overburden stress, yy = yy = maximum horizontal stress, exx, eyy =
1) A gas-filled porosity or microfracture effect similar to a strains in x and y, PP = pore pressure, BV and BH = vertical and
conventional reservoir sand where the product of lambda horizontal poro-elastic constants.
(incompressibility) and density (porosity and fluid) enIn Equation 3, the effective stress (box 2) and maximum
hances the hl response.
(box 3 within the square brackets)
horizontal stress =
2) A pore-pressure effect seen in logs with overpressured are rotated into the minimum horizontal stress direction
zones.
by the ratio h/(h + 2+) (also see Figure 4a). Now the most
3) A geomechanical reason due to a reduction in the mini- fracable zones occupying low h and midrange +p in the
mum closure stress (i.e., a mix of intrinsic brittleness and crossplots can be understood as also having the lowest closure
anisotropy that will be considered next).
stress. Unfortunately, as with all ratios including Poissons ra-

December 2010

The Leading Edge

1501

Downloaded 09/09/14 to 189.225.48.41. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Tight gas sands

Figure 4. (a) Idealized vertical fracture for aligned confining stresses


and strains related to the minimum horizontal closure stress. (b) Plan
view of idealized vertical plane of weakness with h and + parallel (>)
and perpendicular ().

tio, it is not clear whether the requisite low ratio is due to low
values in the numerator, high values in the denominator, or a
combination. However, the empirical observation from Figure 3 reveals that h in the numerator is the dominating factor
of the minimum closure stress. It is worth noting that, from
Griggs empirical relation shown in Figure 1, a higher Youngs
modulus suggests better fracture-prone zones in the Barnett;
but, in Equation 2, this would lead to a higher minimum
closure stress from an enhancement of the last tectonic term
in the equation.
Equation 2, for the minimum closure stress, is not obviously anisotropic. However, in its seismic formulation (Equation 3), the tectonic term (box 3) is clearly a function of
fractional horizontal elastic strain anisotropy due to tectonic
compression. Furthermore, the separate effects of h and +
on minimum closure stress can be understood from a simple
HTI anisotropy fracture model shown in Figure 4a. (HTI =
horizontal symmetry axis and transverse isotropy.) The model
is represented by four terms in h and +, both parallel (>) and
perpendicular () to the maximum horizontal stress direction . These four Lam terms are a logical extension of the
three-parameter ANNIE shale model introduced by Schoenberg and Sayers (1996) that is described by a single isotropic
h term and two + terms, +> (parallel) and + (perpendicular).
The model is captured in Figure 4b, where rigidity, +,
determines the rocks resistance to transverse shear failure or
fractional weakness T, while h is the rocks resistance to
fracture dilation that relates to pore pressure or normal fractional weakness N (Schoenberg and Sayers, 1995). These
concepts can be best described through the stiffness tensor
matrix relating stress to strain for a simple HTI model from
which the minimum closure stress equation is derived (Goodway et al., 2006).
Box 3 in Equation 3, derived from the last term in Equation 2, can be thought of as the tectonic confining stress expressed as the product of maximum horizontal shear stress
( ) and an anisotropic elastic strain energy or potential (e2yy
e2xx) / e2yy (Sayers, 2010). Figure 4a gives a physical sense
of this anisotropy, as being equivalent to the aspect ratio of a
fracture caused by accumulated horizontal tectonic shortening in a compressional geologic setting. While it is difficult to
1502

The Leading Edge

December 2010

Figure 5. LambdaRho, MuRho (LMR) log crossplot from well in 3D


study area showing relation between tight gas sands and shales as well
as background ductile shales and carbonates.

relate a seismic attribute for the tectonic term, a formulation


in terms of measurable shear-wave splitting (Equation 4) can
be made as follows. Assume for polarized S-waves that the fast
shear velocity VS1 myy (e.g., the fast velocity component in
the S-wave splitting parameter from a dipole log). Then as:
is the % of shear wave splitting where VS1
= fast shear velocity, VS2 = slow shear velocity

as both ratios are dimensionless % anisotropy. So the tectonic


term can be represented as:
(4)
Isotropic AVO inversion for geomechanical rock
properties
The log analysis motivation and the seismic relationships for
mapping minimum closure stress in tight gas formations
lead to a methodology using standard AVO inversion on a
3D data set to extract the geomechanical or elastic seismic
attributes described previously. The inversion method estimates fractional P-impedance contrast, Ip/Ip, and fractional S-impedance contrast, Is/Is, as 3D volumes through
weighted stacking of CMP angle (e) gathers (Fatti et al.,
1994), based on a rearrangement of the 1980 Aki and Richards equation.
A data set, chosen to demonstrate the method of minimum closure stress mapping described here, was optimally ac-

Downloaded 09/09/14 to 189.225.48.41. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Tight gas sands

quired with a 3D survey having the required dense offset-azimuth sampling.


The AVO flow included the extraction
of Ip/Ip and Is/Is followed by inversion to LambdaRho and MuRho
volumes. Target zones in this area are
tight gas sands that overlie gas shales,
thereby demonstrating the relationship
between both types of unconventional
reservoirs. An LMR (Lambda Mu Rho)
log crossplot from a well within the 3D
(Figure 5) shows that tight gas sands
have higher values of MuRho (yelloworange polygons) than gas shales (light
grey polygon) due to an increase in rigidity, but share a similar low range for
LambdaRho. The preceding discussion
about the impact of low lambda values on minimum closure stress would
imply that the most prospective gas Figure 6. Interpretation of LMR crossplot polygon zones from the inverted seismic equivalent to
sands and shales are differentiated by the log data in Figure 5 projected onto a 3D line as a color-coded overlay.
this key geomechanical attribute from
background ductile shales (dark grey
polygon) and carbonates (blue polygon). However, a difference between
the best sands and shales is seen in the
opposed MuRho directions of the red
and white arrows in the crossplot. For
tight gas sands, the most fracable zones
have a reduction in both LambdaRho
and MuRho in the direction of increasing porosity similar to conventional
sands (Figure 3), while the best gas
shale zones have an increase in rigidity
MuRho due to brittleness. In the limit,
the best gas shale zones share the same
LMR crossplot space as the best tight
gas sands.
Based on this log analysis, various
polygons identifying lithologies and
zones with the best geomechanical and
minimum closure stress attributes were
selected to help map the inverted 3D
data using the equivalent crossplot to
the log data. The data points enclosed
by these seismic crossplot polygons Figure 7. Offset-azimuth cube and 1800-m offset slice from a 3D CMP over a fractured
were projected back onto a line from carbonate reservoir (Weyburn CO2/water flood, Canada). AVAZ variation: perpendicular is
strong (+ve gradient) and parallel is weak (flat gradient).
the 3D as shown in Figure 6.
The 3D line in Figure 6 can now be
interpreted using the color-coded polygon overlays from Fig- after stimulation far exceeded those of the surrounding wells
ure 5. Tight gas sands with higher porosity are identified in drilled without the benefit of this LMR-based interpretation.
a yellow zone sandwiched between the overlying carbonates Horizontal wells are currently being located in the underlying
in blue and the basal gas shales in dark and light gray. These light gray gas shales that are predicted to have the lowest closands were successfully drilled in their thickest position (as sure stress from the crossplot template and preceding analysis.
evidenced by the drop in gamma and density logs at the top
of the formation between the upper blue and pink horizons) Examples of azimuthal AVO (AVAZ)
at the vertical well location. Production rates and pressures The following two sections investigate the practical applicaDecember 2010

The Leading Edge

1503

Downloaded 09/09/14 to 189.225.48.41. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Tight gas sands

Figure 8. Fractured carbonate reservoir offset-azimuth panels from a


3D CMP. Azimuth panels of single offsets increasing right to left with
arrows indicating fast (parallel) and slow (perpendicular) directions
from residual NMO (Todorovic-Marinic et al., 2005). AVAZ
variation: parallel is strong (+ve gradient) and perpendicular is weak
(flat gradient).
Figure 10. Amplitude variation in HTI medium as advertised
by contractor industry. Note that unlike Figure 9, the relative
stack response from these azimuthally anisotropic gradients would
be stronger than the isotropic equivalent. (Polar azimuth plots of
gradients modeled from Rgers 1997 AVAZ equation.)

Figure 9. AVAZ variation with stack response expectation: parallel is


strong (flat to +ve gradient) and perpendicular is weak (-ve gradient).

tion of azimuthal AVO used to estimate the stress anisotropy


in the tectonic term in the closure stress equation (Equation
3, box 3) and Equation 4. The literature contains numerous examples and observations of AVAZ, some of which are
shown here in order to better understand the expectation
and potential in pursuing this type of analysis (Lynn et al.,
2004). Offset-azimuth cube and panel displays of 3D CMPs
(Figures 7 and 8) clearly link significant azimuth variation
in AVO gradient with azimuthal NMO time delays due to
a known symmetry plane direction. However, there is an
ambiguous relationship of AVAZ gradient to both orientation and intensity of the anisotropy; this is unlike azimuthal
NMO, where residual arrival times are clearly fast or slow.
This ambiguity is seen by comparing the fractured carbonate example in Figure 7, that exhibits an increasing (strong)
gradient perpendicular to fractures and a decreasing (weak)
or flat gradient parallel to fractures, to the opposite case for
another fractured carbonate in Figure 8 (Todorovic-Marinic
et al., 2005). At issue is that without geological or log information indicating the presence of fractures, interpretation
of the seismic AVAZ observations alone could be wrong for
both orientation and intensity, even in a relative sense.
However, AVAZ prediction for fracture orientation and
intensity can be established through modeling as shown in
Figures 9 and 10, but even here there is a conflict with no
single and obvious guiding principle. The Williams and Jen1504

The Leading Edge

December 2010

ner model in Figure 9 logically explains the expected weaker


3D stack response to fractures due to a decreasing gradient
perpendicular to strike, while the isotropic or parallel case has
a flat gradient and hence a strong stack response (i.e., no fractures). This model is opposite to the polar plot of azimuthal gradients in Figure 10, generated by the familiar AVAZ
equation developed by Rger (1997). The relative stack response from the azimuthally anisotropic gradients in Figure
10 would be counter-intuitively stronger than the isotropic
equivalent, unlike the Williams and Jenner model.
The literature lacks any description of an obvious rule
of thumb to deduce the AVAZ response, with one exception from Thomsen suggesting that the direction of the lesser
AVAZ gradient is parallel to fracture strike (Thomsen, 1995).
Consequently, the following discussion develops a classification of a more general AVAZ response through an analysis of
Rgers AVAZ equation. Figure 11 shows the AVO gradient
variation with azimuth for two models with different anisotropic parameters from Rger and Tsvankins 1997 TLE paper. Some initial conclusions from these models are:
1) The magnitude of azimuthal gradient variation is much
smaller than the basic isotropic AVO gradient despite a
choice of realistic values for anisotropy between 8 and
15%.
2) The magnitude of azimuthal gradient variation with azimuth in both the data and model examples in Figures
710 are significantly larger than predicted by Rgers
models, even reversing polarity in Figure 10.
3) The relation of decreasing or increasing azimuthal gradients with respect to the fracture symmetry plane are reversed between the two models (compare red arrows between panels in Figure 11) due to the interaction of the
HTI anisotropic parameters a, b(v), (v) (Thomsen 1986,

Tight gas sands

Downloaded 09/09/14 to 189.225.48.41. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

1995).
4) The zero-offset reflection has no
azimuth variation but is reduced for
gas-filled or dry fractures.
This lack of generality is unlike
the isotropic Rutherford and Williams
AVO gradient classification, indicating changes in VP/VS, and may explain
why conventional AVO is in more
common use.
Anisotropic AVO theory and models from Rgers equation
Inversion for anisotropy using P-waves
is based on conventional AVO follow- Figure 11. Models based on Rgers linearized three-term azimuthal AVO gradient equation.
ing Rgers reformulation of the Aki
and Richards equation for reflectivity
with incidence angle. This reformulation describes a simple model of vertically aligned fractures (HTI anisotropy) as shown in Figure 12. The HTI
model is identical to the more familiar transverse isotropy (VTI) model
for horizontal layers (e.g., shales), the
only difference being a 90 rotation
of the symmetry axis from vertical to
horizontal. Just as isotropic AVO is a
consequence of P-wave conversion to
12. Model of HTI media used in Rgers 1997 AVAZ P-wave reflection equation
vertically polarized Sv shear waves at Figure
showing anisotropic conversion to shear waves for principal and nonprincipal planes of
a reflecting boundary, so the azimuth- symmetry.
ally anisotropic equivalent converts to
a fast velocity Sv (S1) shear wave in the principal isotropic incidence angles. This is equivalent to the familiar VTI model
plane and a slow velocity Sv (S2) shear wave in the aniso- where the difference between vertical and horizontal P-wave
tropic symmetry axis plane (Figure 12). This polarization is propagation is described by Thomsens (epsilon) parameter.
termed shear-wave splitting and is defined as a percentage of By analogy P-wave propagation between the principal planes
shear-wave anisotropy by Thomsens a parameter (Equation for VTI and HTI is described by Thomsens b(delta) parameter for nonnormal incident angles at 45 to the symmetry
4).
The corresponding HTI to VTI Thomsen parameters can planes (Thomsen, 1986). However, for the HTI model, both
and b undergo a transformation to (v) and b(v), where (v)
be compared as follows:
denotes a vertical axis reference due to the 90 symmetry
axis rotation from their VTI equivalents as given by Equaand
HTI relationship to VTI:
tion 5. These rotated HTI parameters, (v) and b(v) along with
where
and = P-wave velocity,
-wave veloc- a, appear in Rgers AVAZ equation for AVO variation with
azimuth angles q between the principal symmetry axis and
ity and for completeness as used later in Equation 8
isotropy plane as shown in Equation 6.
(5)
following

where:

(Alkhalifah

and Tsvankin 1995).


Note that (v) and b(v) are always negative while the S-wave
splitting parameter a remains positive as it is referenced to the
HTI model for polarized vertical shear-wave S1, S2 propagation. P-wave propagation velocity decreases to a minimum
for propagation in the HTI symmetry axis plane at increasing

(6)

December 2010

The Leading Edge

1505

Downloaded 09/09/14 to 189.225.48.41. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Tight gas sands

Vp (m/s)

Vs (m/s)

Rho (gm/cc)

Vp/Vs

avg. Vp/Vs

2462

1100

2.312

2.24

2.10

2760
1400
2.305
1.97
Table 1. Log data from Colorado gas shale case study area. Top layer
background Colorado shale (isotropic or VTI layer). High impedance
top Colorado B zone (HTI layer)

Figure 13. 3D displays for (a) three-term (Rger) versus (b) two-term
(Shuey) equations of azimuthal AVO curves from model values in
Table 1.

Incidence angle , , are averaged across the reflection


interface between HTI and overlying layer HTI. Thomsen
and
are differences in anisotropy
parameters
from HTI to overlying layer. = azimuth angle with respect
to symmetery-axis plane where = 0.
For the isotropic plane, Equation 6 reduces to a version of
the Aki and Richards AVO equation (Ganguly and Goodway,
1998; Wang 1999) as a function of reflectivity contrasts in
density (/), P-wave velocity (_/_), and rigidity (+/+).
However, Rger chose to follow the approach of Shuey (1985)
by gathering the _/_, (v), and b(v) terms with significant
contrast (i.e., CisoCaniso) into the third higher-order incidence
angle term. Consequently, similar to the isotropic case, both
1506

The Leading Edge

December 2010

AVO and AVAZ equations do not have the critical curvature


discrimination when used in industry practice as a two-term
approximation (compare Figure 13a to 13b). However, the
relative contribution of the second versus third term is a far
worse problem in the AVAZ case, leading to fundamental ambiguities and hence errors involved in AVAZ inversion. This
also explains the wide variation and confusion observed in
the data and model examples shown in the previous section.
Log values given in Table 1 from a gas shale in the shallow Colorado group of Alberta are used to demonstrate this
problem by modeling the isotropic/HTI shale layers AVAZ
response. Maximum values of a = 0.08 for shear-wave splitting were established from dipole log measurements for the
Colorado gas shale. Other anisotropy parameters for lowporosity shales obtained from the literature suggest relative
values of a/ = 0.8 for a < 0.2 (Sayers, 2004, for shale values
published by Jones and Wang, 1981; Hornby, 1994; Johnston and Christensen, 1995; and Wang 2002). A gas-filled
fracture model is further characterized by elliptical anisotropy
where (v) = b(v) with typical values of 0.1 to 0.15 (Rger,
1997; Bakulin et al., 2000).
Figures 13a and 13b show the AVAZ variation with incidence angle from parallel (isotropic plane) to perpendicular (in symmetry axis plane) for this Colorado shale model.
These figures enable a comparison between the two-term
Shuey-type approximation used in practice to the full threeterm Rger formulation (Equation 6). The following observations can be made:
1) The full three-term curves are similar to those shown by
Rger for gas-filled fractures with similar HTI parameters
(right panel of Figure 11) where little azimuthal separation
can be seen between the gradients for incidence angles <
35. At higher angles, discrimination between azimuths is
possible due to the curvature in the three-term equation.
The curvature diminishes with decreasing azimuth (q)
from parallel to across fractures.
2) The two-term approximation shows a large and opposite
separation from the full three-term equation for azimuthal
AVO gradients at incidence angles of e > 30 and is unable to match the critically diagnostic three-term curvature
beyond 35.
3) It is not possible to determine whether the anisotropy is
elliptical (i.e., * = 0, Equation 5) or anelliptical from the
two-term industry approximation.
4) The most startling observation is that using a two-term
approximation to fit the actual three-term measurement
would produce a result that showed no azimuthal anisotropy for angles < 35 and the wrong 90-rotated fracture
azimuth for angles > 35.
These conclusions seriously undermine the ability of the
AVAZ method, as used in practice, to detect the orientation
and degree of anisotropy from a fracture or stress-induced
HTI model.
The reason for these inconsistencies is that the contribution of the second term in Rgers equation is reduced or

Downloaded 09/09/14 to 189.225.48.41. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Tight gas sands

removed below 35 of incidence angle, as a result of Baniso


having anisotropic parameters a and b(v) with opposite signs
(Equation 5). Consequently, the high-incidence-angle third
term (Ciso Caniso) in Equation 6 has significant impact on the
azimuthal gradient and cannot be ignored.
In a yet more ambiguous way cross-over angle occurs at
33.9 where for e < 33.9 the parallel azimuth gradient is below that of the perpendicular (across or in the symmetry axis
plane) curve and reverses for e > 33.9 with a greater, more
visible separation (Figure 13a). For the gas-filled fracture
case being considered here, the elliptical anisotropy terms in
Rgers equation can be isolated to obtain this cross-over
incident angle (e) with zero azimuthal (q) amplitude variation (Figure 13a) whose value can be found from:
(7)
where for values used in the Colorado gas shale model for gas
filled fractures:

Despite the confusion in the relation of the azimuth


gradient to incident angle, there is the potential to use this
cross-over angle to establish relations between the various
anisotropic parameters. If such a cross-over angle exists (i.e.,
where all azimuths have the same reflectivity), then the HTI
anisotropy is likely to be near elliptical (v) b(v) and the a/(v)
ratio can be estimated from tan2 e as shown in Equation 7.
Such a cross-over incidence angle is visible in the offsetazimuth cube display shown in Figure 7 (strong band of red
across all azimuths marked as cross-over incident angle).
Given the importance of the third term in Rgers equation, a better approach would be to rewrite the equation in
three terms of equal significance. The result, shown in Equation 8, can now be related to the equivalent isotropic AVO
equation. For the elliptical case where * 0, the underlying
physical connection of the HTI anisotropic parameters b(v)
(ve sign) and a(+ve sign) can be seen as respectively reducing the isotropic AVO gradient terms of _/_ and +/+ as
these are the appropriate parameters associated with the Pwave HTI phase velocity and shear-wave splitting.

(8)
where (from Equation 5)
and Tsvankin, 1995).

(Alkhalifah

AVAZ classes and rules of thumb for azimuthal AVO


gradients
The following AVAZ classes are defined for the tight gas case
of elliptical anisotropy described in Equation 8, when * = 0

(i.e., b(v) = (v)).


Class A: For ` _ a v , the parallel azimuth
gradient is always less than the perpendicular gradient (confirming Thomsen, 1995).
Class B: For ` _ a v >, there exists an incident cross-over angle ecross where for e < ecross, the parallel
azimuth gradient is greater than the perpendicular gradient
and this relation reverses for e > ecross. If such a cross-over
angle exists (i.e., where all azimuths have the same reflectivity), then a/(v) for the anisotropic layer can be estimated from
tan2 ecross as shown in Equation 7 (Perez, 2010).
Summary
The use of conventional AVO to identify optimum geomechanical properties for the successful exploitation of tight gas
sands and shales was described, based on comparisons to the
Barnett Shale. Optimum gas-shale properties have relatively
low hl (incompressibility) and high +l (rigidity) that give
rise to geomechanical brittleness capable of supporting extensive induced fractures. Furthermore, these properties also
produce the lowest closure stresses (i.e., largest fractures) due
to the h/(h + 2+) term in the closure stress equation. Seismic 3D AVAZ, used to detect anisotropy due to fractures or
stress, offers the only opportunity to identify fracture-prone
zones prior to committing to significant horizontal drilling
costs. However, fundamental theoretical and practical ambiguities in standard industry AVAZ practice used to detect
the orientation and intensity of anisotropy exist. This led to
the development of a new set of equations that reduce these
ambiguities by providing an AVAZ classification or rules of
thumb similar to that for AVO and thereby improve the
ability of the technology to establish the presence of fractureprone zones and hence optimum gas recovery.
References
Aki K. and P. G. Richards, 1980, Quantitative seismology: W. H.
Freeman.
Alkhalifah, T. and I. Tsvankin, 1995, Velocity analysis for transversely
isotropic media: Geophysics, 60, no. 5, 15501566.
Bakulin, A., V. Grechka, and I. Tsvankin, 2000, Estimation of fracture parameters from reflection seismic data. Part I: HTI model
due to a single fracture set: Geophysics, 65, no. 6, 17881802.
Fatti J. L., G. C. Smith, P. J. Vail, P. J. Strauss, and P. R. Levitt, 1994,
Detection of gas in sandstone reservoirs using AVO analysis: A 3D
seismic case history using the Geostack technique: Geophysics, 59,
no. 9, 13621376.
Gidlow P. M., G. C. Smith, and P. J. Vail, 1992, Hydrocarbon detection using fluid factor traces; A case history: SEG/EAEG Summer
Research Workshop Abstracts, 7879.
Goodway W., T. Chen, and J. Downton, 1997 Improved AVO fluid
detection and lithology discrimination using Lam parameters;
hl, +l, and h/+ fluid stack from P and S inversions: CSEG National Convention Expanded Abstracts.
Ganguly N. and W. Goodway, 1998 Iterative extraction of h, + and l
from AVO: M.Sc. co-op report University of Victoria.
Goodway W., 2001, AVO and Lam constants for rock parameterization and fluid detection: CSEG Recorder, 26, no. 6, 3960.
Goodway W., J. Varsek, and C. Abaco, 2006, Practical applications of
P-wave AVO for unconventional gas resource plays. Part 2: DetecDecember 2010

The Leading Edge

1507

Downloaded 09/09/14 to 189.225.48.41. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Tight gas sands

tion of fracture prone zones with azimuthal AVO and coherence


discontinuity: CSEG Recorder, 31, no. 4, 5365.
Goodway W., 2009, The Magic of Lam: SEG North American Honorary Lecture (available online at www.seg.org).
Grigg M., 2004, Emphasis on mineralogy and basin stress for gas shale
exploration: SPE meeting on Gas Shale Technology Exchange.
Lynn H., 2004, The winds of change: Anisotropic rocks-their preferred direction of fluid flow and their associated seismic signaturesPart 1: The Leading Edge, 23, no. 11, 11561162.
Lynn H., 2004, SEG/AAPG Distinguished Lecture.
Perez M., Beyond isotropy in AVO and LMR: CSEG Recorder (in
press).
Thomsen L., 1986, Weak elastic anisotropy: Geophysics, 51, no. 10,
1954-1966.
Thomsen L., 1995, Elastic anisotropy due to aligned cracks in porous
rocks: Geophysical Prospecting, 43, 805829.
Thomsen L., 2002, Understanding seismic anisotropy in exploration
and exploitation: SEG Distinguished Instructor Short Course.
Todorovic-Marinic D., B. Mattocks, R. Bale, D. Gray, and S. Roche,
2005, More powerful fracture detection: Integrating P-wave, converted-wave, FMI and everything: CSEG National Convention
Expanded Abstracts.
Rger, A. and I. Tsvankin, 1997, Using AVO for fracture detection:
Analytic basis and practical solutions: The Leading Edge, 16, no.
10, 14291434.
Rger A., 1997, P-wave reflection coefficients for transversely isotropic
models with vertical and horizontal axis of symmetry: Geophysics,
62, no. 3, 7137822.
Rutherford S. R. and R. H. Williams, 1989, Amplitude-versus-offset

1508

The Leading Edge

December 2010

variations in gas sands: Geophysics, 54, no. 6, 680-688.


Sayers C., 2010, Geophysics under stress: Geomechanical applications of seismic and borehole acoustic waves: SEG Distinguished
Instructor Short Course.
Schoenberg, M. and C. Sayers, 1995, Seismic anisotropy of fractured
rock: Geophysics, 60, no.1, 204211.
Schoenberg M., F. Muir, and C. Sayers, 1996, Introducing ANNIE: A
simple three-parameter anisotropic velocity model for shales: Journal of Seismic Exploration, 5, 3549.
Shuey R. T., 1985, A simplification of the Zoeppritz equations: Geophysics, 50, no. 4, 609615.
Wang Y., 1999, Approximation of the Zoeppritz equations and their
use in AVO analysis: Geophysics, 64, no. 6, 19201937.
Williams M. and E. Jenner, 2002, Interpreting seismic data in the
presence of azimuthal anisotropy: or azimuthal anisotropy in the
presence of the seismic interpretation: The Leading Edge, 21, no.
8, 771774.

Acknowledgments: We acknowledge the significant contributions of


Doug Anderson, Eric Keyser, Daria Pusic (EnCana), Brian Hoffe
(Shell), Dave Gray (Nexen), and Jon Downton (CGGVeritas). In
addition, I (Bill Goodway) would like to thank my TLE co-editors
Colin Sayers, Chris Liner and Tad Smith for their encouragement,
assistance and patience in the publication of this article. Finally
Tads expert editing of this article has improved its readability well
beyond the quality of the original submission.
Corresponding author: bill.goodway@apachecorp.com

Vous aimerez peut-être aussi