Sie sind auf Seite 1von 10

VII INGEPET 2011 (EXPR-2-AC-33E)

RISK REDUCTION IN THE DETERMINATION OF FRACTURES DENSITY AND


ORIENTATION THROUGH AZIMUTHAL PROCESSING AND INTERPRETATION OF
SEISMIC DATA
Alvaro Chaveste (Geokinetics, Inc.)
Abstract
The determination of fracture density and orientation is important in plays in which reservoir
permeability is associated with micro-fractures. In these reservoirs, effective porosity is
dependent upon the density of open fractures and permeability has a directivity component
associated to the fractures orientation. Accurate determination of fracture density and
orientation using surface seismic helps, then, in determining locations of good reservoir
storability, as well as directional drilling orientation for optimum permeability. The anisotropy of
P- and S-wave velocities is commonly associated to magnitude and orientation of stress fields
or open fractures and is; hence, the medium through which factures are characterized using
surface seismic data.
A case study in the Marcellus Shale (NE United States) is presented in which wide azimuth
seismic data, acquired and processed to preserve the amplitude and velocity information of the
source-receiver azimuth, is used to estimate anisotropy attributes. The Marcellus Shale is
known to have two sets of micro-fractures associated to different tectonic events. Particular to
this play is that one set of micro-fractures is perpendicular to the co-located macro-fractures. In
this case although curvature, or other geometric attributes, reveal the strike of macro-fractures,
this does not correspond to that of the micro-fractures. Anisotropy analysis through elliptical
inversion identifies both sets of micro-fractures and verifies their expected position in
relationship to the oroclinal belt and macro-fractures.

Introduction
Conventionally processed seismic data provides a structural image of the sub-surface which
helps, among other things, in defining macro-fractures or faults; nevertheless, the microfractures that define permeability and porosity within a formation, although possibly associated
to macro-fractures, are not necessarily part of the structural picture and, hence, are not
commonly observed in migrated stacks. Micro-cracks can be characterized by considering that
velocities in cracked media vary as a function of micro-fractures orientation and density; the
velocity being faster in the fractures direction or maximum horizontal stress and slowest in a
directional perpendicular to fractures. Velocity anisotropy provides, then, the means of
estimating reservoir properties from seismically derived velocities.
An added benefit of the resultant anisotropic velocity field is that it can be used for anisotropic
NMO correction. In this case the correction applied to each trace is based upon the velocity
ellipse at the traces Common Image Point (CIP) as well as the traces azimuth and offset. The
NMO corrected data results in flatter gathers for subsequent pre-stack amplitude analysis and
higher resolution stacks.

VII INGEPET 2011 (EXPR-2-AC-33E)

The area under analysis is approximately 100 squared miles and is within the Allegheny Plateau
Province of the Appalachian Basin in Pennsylvania, USA. In the study area, the Middle
Devonian Marcellus formation is at a depth of 6000 feet and is 300 feet thick. Stratigraphically it
lies at the basal Hamilton Group and is overlain by the Mahantango Formation, which is a thick
gray-green shale with inter-bedded siltstones and sandstones. The Marcellus Formation is
divided into two organically rich members separated by the Cherry Valley Limestone. The lower
Marcellus shows TOCs larger than 10% which is of economic relevance since a correlation is
seen between organic-rich black shale facies, gas shows and/or production.
Structurally, the Devonian section is influenced by deep-seated structure; which, along with the
salt withdraws and/or decollement of the Upper-Silurian Salina Formation produces shallower
structure decoupled from the deep structure at the salt layer. As a result, the Marcellus is a thinskinned system with shorter wavelength folds than deeper ones. Figure 1 shows the structural
style as well as the seismic character of the stratigraphic column of interest.
(+)

500

600

Tully

700

800

~ Cherry Valley

Mahantango
Marcellus

900

1000

1100

1200

Salina Salt
1300

Onondaga
(-)

Figure 1 Structural style and seismic character of the stratigraphic column of interest.

Methodology
The methodology followed makes use of wide azimuth data to estimate, through elliptical
inversion, anisotropys magnitude and direction. The technique, which fits an ellipse to four or
more velocity measurements, is based on the assumptions that the velocity magnitude changes
following an ellipse in an anisotropic medium (Figure 2).

VII INGEPET 2011 (EXPR-2-AC-33E)

AX

<Or Cr
ien ac k
t
atio
n

f - Azimuth
VMAX
VMIN

M
IN

(VMAX-VMIN)/VMAX
e - Error

Figure 2 - Velocity ellipse. Velocities in a cracked


medium result in different wave velocities for
different propagation directions (anisotropy).
Elliptical attributes (f, VMAX, VMIN an ), obtained
through elliptical inversion, help reduce risk in the
estimation of fractures density and orientation

Two philosophies for elliptical inversion are commonly regarded. One in which the seismic data
is sectored into azimuth bins prior to elliptical inversion (sectored elliptical inversion) and a
second in which all traces are used independently in the analysis (Thompson, 2002). While the
second technique provides higher resolution it does not allow for intermediate QC steps and
results, usually, in noisy estimation of maximum and minimum interval velocities as well as
increasing uncertainty in the estimated azimuth with reflection time as a consequence of the
iterative estimation of this attribute. The noise is usually handled by resampling data to larger
time intervals; which results in decreasing the resolution sought in the first place with the unsectored technique.
In the following paragraphs the methodology and results, using the sectored elliptical inversion,
are presented for an area Northeast USA in which the target formation (Marcellus Shale of the
Devonian) has two sets of fractures related to two different geologic events. The identification
and extent (areal and vertical) of each of these sets is of economic relevance.
Field data analysis and QC
Estimation, from seismic data, of azimuth dependent attributes or rock properties require
uniform data sampling in both, azimuth and offset; as well as a processing sequence that
preserves these for later analysis. The Offset Vector Tile (OVT) acquisition technique (Vermeer,
2005, 2007) allows data processing such that offset and azimuth are preserved through prestack migration.
It should be noted that adequate field design results in optimum use of the data acquired.
Analysis of the azimuth-offset distribution in Figure 3 shows that, for the case presented, data
has a homogeneous offset azimuth distribution from zero to, approximately, 10,000 ft. Traces
with larger offset are not included in the analysis to avoid bias in the estimation of velocities
associated to acquisition design.

VII INGEPET 2011 (EXPR-2-AC-33E)

Offset, ft
-5,000

5,000

10,000 15,000

10,000 15,000

-15,000 -10,000

0
-5,000
-15,000 -10,000

Offset, ft

5,000

Figure 3. Rose diagram of all traces recorded in the


survey. The white circle indicates the range off offsets
for which there is full azimuthal coverage. Traces outside
this circle are not included in the analysis.

(Azimuth-Offset-Fold Distribution from Data)

The pre-stack time migrated (PSTM) offset limited data is sorted into four azimuth bins (0 o to
45o, 45o to 90o, 90o to 135o and 135o to 180o). Post migration processes, such as Radon
filtering, used to increase signal-to-noise ratio are applied independently to each azimuth bin.
Velocity Estimation
The required high-resolution velocities are obtained by automatically updating the handpicked
RMS velocities per azimuth bin using a hybrid approach based on discrete picking in high
semblance events and sample-by-sample picking using AVO attributes (Swan, 2001) in low
semblance areas. This results in the estimation of velocities for every sample of every trace.
Small residuals are expected, as seen in Figure 4, and the use of an automated technique
insures the consistent, un-biased estimation of velocities.

200 -

Tully
Top Marcellus

-20%

DVRMS

20%

400 -

600 -

800 -

1000 -

1200 -

Figure 4 High-density velocity analysis. The Pre-Stack Time Migrated (PSTM) gather in the
central image is the same as the one in the left after residual NMO velocities applied. The magenta
line in the middle panel shows that the residual velocities to apply are in the order of 3%. The
panel on the right shows in color, the incidence angles.

VII INGEPET 2011 (EXPR-2-AC-33E)

The lateral and vertical resolution of the resultant interval velocity field is improved when
compared to the equivalent from the handpicked velocities. In Figure 5 both velocity fields are
compared in vertical section and Figure 6 compares these, along with the amplitudes, in timeslice.
5600
|

5700

5800

5900
|

5600
|

5700
|

5800
|

5900
|

500
600
700
800
900
1000
1100
1200

Well Location

Well Location

1300

Figure 5 Inline 6015. Interval velocities from seismic. The interval velocities computed from the
velocity field estimated through automated velocity analysis (left) results in better spatial and
vertical resolution than the equivalent computed from the handpicked velocities (right).

Figure 6 Time-slice 680 msec. Interval velocities from seismic data. Comparison of interval
velocity fields in time-slice shows the increase in spatial resolution of the velocity field estimated
through the automated velocity analysis (center) as compared to the equivalent computed from
the handpicked velocities (right). Note that velocity structure in the central panel has a good
correlation with the structure as seen in the reflection amplitude time slice (left).

Elliptical Inversion

VII INGEPET 2011 (EXPR-2-AC-33E)

The interval velocities estimated for each azimuth sector are elliptically inverted to generate,
sample-by-sample, volumes of maximum velocity (VMAX), minimum velocity (VMIN), the azimuth
to the fast velocity ( ) and the error in fitting the ellipse ( ). Figure 7 shows VMAX and VMIN
extracted at the Lower Marcellus (Top Marcellus + 40 msec.).
The magnitude of anisotropy ( ) computed as
= (VMAX VMIN) / VMAX

Eq. 1

is displayed in Figure 8 along with the azimuth ( ).

Figure 7 Maximum and minimum velocities (VMAX and VMIN) extracted at the Lower Marcellus (top
Marcellus + 40 msec). VMAX and VMIN are used to estimate anisotropy (A in equation 1).

Figure 8 Azimuth ( ) and anisotropy ( ) extracted at the lower Marcellus (top Marcellus + 40
msec.). The map in the left is the azimuth, in degrees from North, of the fast velocity (VMAX); which
is related to that of the maximum horizontal stress ( H) or fractures direction. The image to the
right corresponds to the magnitude of anisotropy commonly associated to fracture intensity.

VII INGEPET 2011 (EXPR-2-AC-33E)

Interpretation of Results
In unconventional shale-gas plays drilling is often designed to cut across open joints, if present.
When additional stimulation is necessary, common practice is to drill normal to the maximum
horizontal stress (SH) of the contemporary stress field. The identification of the magnitude and
orientation of open fractures and/or the maximum horizontal stress plays then an important role
in reservoir management.
According to Engelder (2008) Three joint sets are common in the Appalachian Basin,
designated as J1, J2, and J3. Where present in outcrop, J1 maintains its ENE orientation
regardless of location relative to the oroclinal bends of the Central and Southern Appalachian
Mountains (i.e. Engelder and Whitaker, 2006). In the Valley and Ridge, J2 is found normal to
fold axes, qualifying this set as a systematic cross-fold joint set.
In the area of analysis the oroclinal bends trend East-West (Figure 9) and the orientation of the
J2 joints associated to the valley and ridges is North-South as determined by the estimated
azimuth (Figures 8 and 9). Away from the oroclinal bends, the estimated azimuths are,
approximately, 90 degrees corresponding to the J1 set in agreement with Engelders
observations.

Ridge

Valley

Figure 9 The top Marcellus time structure map at left shows the valley and ridges of the oroclinal
belt. The fracture orientation in these is North-South as observed in the right image
corresponding to an extraction of the estimated azimuth ( ) at the lower Marcellus (Top Marcellus
+ 40 msec.).

The cross-plot of anisotropy versus azimuth of Figure 10 shows clearly both sets of fractures.

VII INGEPET 2011 (EXPR-2-AC-33E)

Azimuth, deg

J2

Figure 10 The cross-plot of


anisotropy and azimuth extracted
at the lower Marcellus (Figure 8)
shows two clusters associated to J1
and J2 joint sets.

J1

J2
Anisotropy

Analysis of the estimated azimuth and anisotropy in vertical section (Figure 11) provides
information about seal integrity. Note that in the valley and ridges the estimated anisotropy
associated to the J2 set goes beyond the top Marcellus suggesting that, in this location, the seal
is non-existent. Away from the valley and ridges, where the J1 set prevails, the seal appears to
exist as no fractures go across the top Marcellus into the Mahantango Shale. In the study area
the locations where the J1 set exists and shows large anisotropy magnitude may prove to be
good locations for gas entrapment.

500 -

600 -

700 -

Valley
800 -

900 -

Top Marcellus

Top Marcellus

1000

1100

1200

Ridges

Figure 11 Anisotropy azimuth (left) and magnitude (right). Fractures orientation in the valley
and ridges is North-South as expected for the J2 set. Away from the valley and ridges, the
fractures orientation is East-West corresponding to that of the J1 set.

An added benefit of the anisotropic velocity field ( and ) is that it can be used to NMO correct
the traces as a function of both, offset and azimuth. This would result, in an anisotropic
environment, in higher resolution stacks than those obtained with conventional velocity
estimation techniques (Figures 12 and 13).

VII INGEPET 2011 (EXPR-2-AC-33E)

9
(+)

(-)

Figure 12 PSTM stack. Stack with handpicked NMO velocities

(+)

(-)

Figure 13 PSTM stack with high-resolution anisotropic (HTI) NMO correction. Note the
improvement in resolution, continuity and fault definition when compared to the stack in Figure
12.

Conclusions
The methodology presented estimates anisotropys magnitude and orientation that match the
expected fractures density and orientation. The definition of the fractures set (either J1 or J2)
and its areal and vertical extent is of importance as they determine fractures conditions directly
associated to, seal integrity, orientation and intensity of open fractures; all of which help in
placing wells and defining orientation for optimum performance.

VII INGEPET 2011 (EXPR-2-AC-33E)

10

Acknowledgements
The author thanks Geophysical Pursuit Inc. (GPI) and Geokinetics Inc., for permission to publish
the data. My appreciation goes to Alex Villasana who contributed processing the data and to,
Mary Edrich and Tony Rebec for their input on the stratigraphy and structural style.

References
Engelder, T., 2008, Structural geology of the Marcellus and other Devonian gas shales:
Geological conundrums involving joints, layer-parallel shortening strain, and the contemporary
tectonic stress field: Pittsburg Association of Petroleum Geologists Field Trip (Sept. 12-13,
2008).
Engelder, T., and Whitaker, A., 2006, Early jointing in coal and black shale: Evidence for an
Appalachian-wide stress field as a prelude to the Alleghanian orogeny: Geology, v. 34, p. 581584.
Swan, H. W., 2001, Velocities from AVO Analysis: Geophysics, 66, 1735-1743.
Thompson, L., 2002, Distinguished Instructor Short Course, SEG (2002).
Vermeer, J. O. Gijs, 2005, Processing orthogonal geometry what is missing? SEG Expanded
Abstracts, 24, 2201-2204.
Vermeer, J. O. Gijs, 2007, Reciprocal offset-vector tiles in various acquisition geometries. SEG
Expanded Abstracts, 26, 61-65.

Das könnte Ihnen auch gefallen