Sie sind auf Seite 1von 53

Seismic Detection of Fractures

810
811811811
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
811
Why Worry About Fractures?

They dominate permeability:


Can make tight gas economical
They can harm good reservoirs by creating
thief zones

812812812
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
812
What do we need to know?

Where to find Sweet Spots for drilling


Fracture location
Fracture intensity
Fracture orientation
Gas, oil, or water?
Permeability

813813813
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
813
Fractures Can Have Many Seismic Attributes

Low P- and S-wave velocities


Anomalous reflectivity (Impedance)
Low Q (high seismic attenuation)
Low Poissons Ratio (Vp/Vs ratio)
Anomalous AVO
Azimuthal variation in velocity, AVO

the optimum choice will vary from site to site

814814814
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
814
Site-to-site variations in the optimum
attribute result from:
Rock type
Fracture geometries
Fluids
Acquisition geometries
Business objective/constraints

815815815
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
815
Velocity Indicator of Fractures
6
Fractured Limestone
Bedford Limestone
Water-saturated
Sat.
5
V
P
Dry

(km/s)
Velocity(km/s)
4

More fractures
Velocity
3 Dry

Sat. V
S
Saturated
2

1
0 100 200 300
Effective
Effective Pressure
Pressure (bars)
(bars)

Adding Fractures:
Lowers Seismic Velocities
Change Seismic Vp/Vs ratio
816816816
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
816
Velocity Indicator of Fractures

in-line

Good Producers
- lower Velocity

X-line NMO Vel

Low stacking velocity correlates with


fracture-related production 817817817
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
817
Seismic waves tend to see fractures when:
their direction of propagation or
their direction of polarization
is perpendicular (or nearly perpendicular) to the fracture faces
P fast

Fracture normal
P slow

S fast
S slow

S slow

818818818
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
818
819819819
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
819
820820820
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
820
821821821
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
821
822822822
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
822
Shear Wave Splitting
When fractures are vertical, a vertically propagating shear wave will
split into two modes, one slowed by the fractures, the other propagating
at approximately the speed of the unfractured rock.
S fast
S slow

input wavelet

split output signal

time
823823823
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
823
Alford Rotation
If we generate shear waves polarized in one of the principal directions, then only
that wave will propagate. But usually the survey will be at some arbitrary
orientation relative to the fractures. Then crossline shear waves will split into
slow and fast modes, and inline shear waves will split into slow and fast modes.
We could repeat the survey by rotating the vibrators and geophones in the field
to align with the principal directions. Or, we can simply rotate the souces and
receivers in the computer after the survey is complete. This is called the Alford
Rotation.
crossline
source, rcvr.

inline
survey line source, rcvr

824824824
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
824
825825825
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
825
826826826
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
826
Azimuthal Attributes

827827827
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
827
Azimuthal Seismic Attributes

3D Surveys may yield N-S


azimuthal variations in:
NMO E-W
Amplitude
Frequency/Q

Anisotropy might be
visible if fractures are
azimuthally aligned.

828828828
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
828
829829829
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
829
830830830
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
830
831831831
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
831
832832832
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
832
Seismic yields outcrop-consistent
fracture directions
North

P-wave Traveltime

S-wave Traveltime
Outcrop

West

Dipole Sonic Log P-wave Amp

South 833833833
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
833
Geologic Controls
Constraints on Fracture Occurrence
Bed curvature (geometry)
Bed thickness
Lithofacies (brittleness)
Nearness to faults
Stress
Why are these important?
Additional constraints/lower risk
Linking strategy between wells and seismic
Information below seismic resolution

834834834
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
834
Fractures Often Prefer Certain Lithofacies
Joint Sheared Joint

Brittle facies often


accommodate strain
by fracturing, while
shales might
accommodate strain
ductily.

J-M. Florez, 2005 835835835


Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
835
Role of Lithofacies
In this case, log data indicate inverse relation
between shaliness and fracture density.

836836836
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
836
837837837
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
837
838838838
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
838
839839839
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
839
840840840
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
840
Impact of the acquisition footprint

top: amplitude map at 1200 ms fold map

bottom: amplitude map at 1300 ms acquisition geometry

841841841
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
841
Impact of the acquisition footprint

top: amplitude map at 1200 ms fold map

bottom: amplitude map at 1300 ms acquisition geometry

842842842
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
842
Rock Physics Workflow
Our strategy is an extension of the model-based analysis
that we use for more conventional reservoir
characterization. Typical steps:

GWLA for well log quality control and editing.


Determine local geologic fracture models (likely
orientations, facies, pore fluids).
Rock Physics models to numerically place fractures into
the background model derived from logs.
Rock Physics models to generate anisotropic pseudo-
logs.
Compute synthetic seismic traces, attributes to identify
possible fracture signatures, as well as pitfalls.

843843843
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
843
Typical Rock Physics Approach
Logs to derive
site-specific
rock properties

Rock Physics models to


explore likely fracture
scenarios
Fracture densities
Fracture orientations
Fluids
Lithology Compute
seismic
Candidate fractured
intervals signatures

Modified depth-
dependent elastic rock 844844844
properties Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
844
Rock Physics Fracture Models
The role of elastic fracture models is to help provide a link between seismic
observables such as elastic constants, anisotropy, shear wave splitting, and
directions of symmetry planes and the physical rock and fracture properties:
fracture length, orientation, aperture, geometry, connectedness, permeability.
The physical mechanism of the elastic fracture effect is very simple: fractures,
whether wet or dry, soften the rock in a way that depends on the fracture
shape, the number of fractures, the fracture filling material, and fracture
direction. Normal and shear stresses acting on the plane of the crack induce
more overall strain than would be observed in the host rock without the
fractures.

There are two general approaches to quantifying


the effect of cracks and fractures :

penny shaped crack models

planes of compliance or discontinuous slip


845845845
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
845
Hudsons Penny-Shaped Crack Model
Hudsons model is based on a scattering theory analysis of an elastic solid with thin penny-
shaped ellipsoidal cracks or inclusions. The effective moduli ceffij are given as
eff 0 1 2
c ij = c ij + c ij + c ij
1 2
where c 0ij are the isotropic background moduli and c ij, c ij are the first and second
order corrections respectively. (Note that Hudson uses a slightly different definition for
the shortened 2-subscript elastic notation, and there is an extra factor of 2 in his c 44 ,
c 55 , and c 66 .. This makes the equations given in his paper for c 144 and c 244
slightly different from those given below which are consistent with the more standard
notation)
For a single crack set with crack normals aligned along the 3-axis,
the cracked media shows transverse isotropic symmetry and the
corrections are
2
1 ( + 2 ) 2
c 111 = U3 c 33 = U 3
( + 2 ) c 1
= U 1
c 113 = U3 44
1
c 66 = 0
2 3
,+ 28
where q = 15 2 + 28 = Na3 = = crack density
V 4
The isotropic background elastic moduli are and while a and are the crack radius
and aspect ratio respectively. U1 and U3 depend on the crack conditions, and are given on
846846846
the next page. Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
846
Hudsons Penny-Shaped Crack Model

U1 and U3 depend on the crack conditions. For dry cracks

16( + 2) 4( + 2)
U1 = U3 =
3(3 + 4) 3( + )

For "weak" inclusions (i.e. when /[K' + (4/3)'] is of the order 1 and is not small
enough to be neglected),

16( + 2) 1 4( + 2) 1
U1 = U3 =
3(3 + 4) (1 + M) 3( + ) (1 + )

4 ( + 2 ) [K + (4/3)]( + 2)
where M = =
(3 + 4 ) ( + )

with K' and ' the bulk and shear modulus of the inclusion material. Dry cavities can be
modeled by setting the inclusion moduli to zero. Fluid-saturated cavities are simulated
by setting the inclusion shear modulus to zero.

847847847
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
847
Hudsons Penny-Shaped Crack Model

The first order changes 1 and 1 in the isotropic elastic moduli and of a material
containing randomly oriented inclusions are given by

2
1 = (3U 1 + 2U 3)
15
(3 + 2) 2
3 1 + 2 1 = U3
3

For two or more crack sets aligned in different directions, corrections for each crack set
are calculated separately in a crack-local coordinate system with the 3-axis normal to the
eff
crack plane, then rotated or transformed back into the coordinates of c ij and finally the
results are added to get the overall correction.

848848848
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
848
849849849
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
849
850850850
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
850
851851851
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
851
852852852
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
852
Top of James Lime
Anomalous Stack Amplitude at Fracture clusters
3D Seismic VSP

853853853
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
853
Quantifying Amplitude Anomalies

unfractured

fractured

Quantitative Rock Physics model, tied to well.


854854854
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
854
Distinguishing AVO signatures of
fractures vs. shaley facies

Increasing
fractures
Increasing
shalines
D.Sava,
2004

At this site, gas-filled fractures can be distinguished


855855855
from unfractured shaley facies
Stanford Rock Physics using AVO
Laboratory - Gary Mavko & Tapan Mukerji
855
Quantifying AVO of fractures, fluids

gas

At this site, gas-filled fractures can be


distinguished with AVO,Stanford
butRock
not water-filled.
Physics
856856856
Laboratory - Gary Mavko & Tapan Mukerji
856
Uniformly Fractured
1 set, 30 degrees offset

857857857
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
857
Uniformly Fractured
2 sets, 30 degrees offset

858858858
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
858
Uniformly Fractured
1 set, 40 degrees offset

2700

2800

2900 b
c

e
f
g

3100 h

859859859
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
859
NonUniformly Fractured
1 set, 40 degrees offset

2700

2800

2900

3100

860860860
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
860
AVO
Computed Anomaly: Gas-filled Fractures
Offset Offset

Fractured Unfractured

Fracture indicator
861861861
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
861
Conclusions
No Silver Bullet for fracture detection;
Anisotropy alone might not be the answer
Fracture mapping has many pitfalls
Rock physics can help reduce risk, by:
Quantifying signatures of fracture scenarios
Exploring effects of rock types, fluids
Integrating well log and seismic constraints
Finding optimum attribute for the site and the
acquisition

862862862
Stanford Rock Physics Laboratory - Gary Mavko & Tapan Mukerji
862

Das könnte Ihnen auch gefallen