Anda di halaman 1dari 1

FiL íor Purpose Seismic FiL íor Purpose Seismic FiL íor Purpose Seismic FiL íor Purpose

ose Seismic
Reservoir CharacLerisaLion Reservoir CharacLerisaLion Reservoir CharacLerisaLion Reservoir CharacLerisaLion
1roy 1hompson
í
, MaLLhew LamonL
í
, Carlo 8evilacqua
2
, NaLasha Hendrick
í
1roy 1hompson
í
, MaLLhew LamonL
í
, Carlo 8evilacqua
2
, NaLasha Hendrick
í
1roy 1hompson , MaLLhew LamonL , Carlo 8evilacqua , NaLasha Hendrick
í. DownUnder CeoSoluLions PLy LLd í. DownUnder CeoSoluLions PLy LLd
2. Apache Fnergy LimiLed 2. Apache Fnergy LimiLed
1. Introduction 1. Introduction
Quantitative seismic interpretation utilises seismic Quantitative seismic interpretation utilises seismic
amplitude behaviour in conjunction with well log data, amplitude behaviour in conjunction with well log data,
petrophysics and rock physics to make quantitative
predictions about lithology and fluid away from well
locations. Seismic reservoir characterisation in general locations. Seismic reservoir characterisation in general
cannot follow a one-size-fits-all approach ÷ it is critical to cannot follow a one-size-fits-all approach ÷ it is critical to
consider local geological insight. Ìt is also essential to consider local geological insight. Ìt is also essential to
determine the appropriate quantitative interpretation (QÌ) determine the appropriate quantitative interpretation (QÌ)
workflow based on available seismic and well data, and workflow based on available seismic and well data, and
the desired outcome. Together, this will ensure robust the desired outcome. Together, this will ensure robust
and reliable characterisation of the hydrocarbon and reliable characterisation of the hydrocarbon
reservoir is achieved. reservoir is achieved.
Figure 1 illustrates a range of QÌ workflow options
Figure 1: High IeveI road map of range of QI options
Figure 1 illustrates a range of QÌ workflow options
available for reconnaissance, exploration and reservoir-
Figure 1: High IeveI road map of range of QI options
avaiIabIe for seismic reservoir characterisation.
available for reconnaissance, exploration and reservoir-
appraisal applications. This list is by no means
avaiIabIe for seismic reservoir characterisation.
exhaustive, but highlights the range of QÌ work that can
be done to enhance seismic exploration given different inversion-based studies. Rock physics studies should be done to enhance seismic exploration given different
project scenarios. Typically, reconnaissance workflows
inversion-based studies. Rock physics studies should
always be undertaken when well log data are available. project scenarios. Typically, reconnaissance workflows
must be utilised when no wells are available within the
always be undertaken when well log data are available.
must be utilised when no wells are available within the
seismic survey area. Ìn this instance, amplitude This paper gives an overview of the different QÌ seismic survey area. Ìn this instance, amplitude
variation with incident angle (AVA) attributes can be
This paper gives an overview of the different QÌ
workflows that are available and demonstrates how they variation with incident angle (AVA) attributes can be
used to identify exploration targets. As more well control
workflows that are available and demonstrates how they
can add value to a project, with particular emphasis on used to identify exploration targets. As more well control
becomes available, the QÌ workflows can become more
can add value to a project, with particular emphasis on
discussing a new approach to working with AVA becomes available, the QÌ workflows can become more
sophisticated, and move from attribute-type studies to
discussing a new approach to working with AVA
attributes. sophisticated, and move from attribute-type studies to attributes.
3. AVA Stack Rotations 3. AVA Stack Rotations
AVA stack rotations are one powerful attribute that can and fluid properties. Figure 5 shows a synthetic earth AVA stack rotations are one powerful attribute that can
be used to highlight potential hydrocarbon anomalies
and fluid properties. Figure 5 shows a synthetic earth
model and related near- and far-angle stacks. Figure 6 be used to highlight potential hydrocarbon anomalies
when minimal or no well log data are available within the
model and related near- and far-angle stacks. Figure 6
is a plot of the relative elastic impedance as a function of when minimal or no well log data are available within the
seismic survey area. AVA stack rotations are equivalent
is a plot of the relative elastic impedance as a function of
AVA rotation angle, for the three relevant litho-fluid seismic survey area. AVA stack rotations are equivalent
to the well known extended elastic impedance (EEÌ) interfaces in this example. These stochastic modelling
results indicate that there is a broad range of rotation attribute (Whitcombe et al., 2002), and can be thought of
as weighted stacks designed to enhance or suppress
results indicate that there is a broad range of rotation
angles at which the three interfaces will have large and as weighted stacks designed to enhance or suppress
particular AVA responses. EEÌ is typically computed in
angles at which the three interfaces will have large and
different relative impedance contrasts (~ -20 to 90). particular AVA responses. EEÌ is typically computed in
intercept-gradient space, however stack rotations can
different relative impedance contrasts (~ -20 to 90).
Figure 7(a) shows the optimal fluid and lithology stack, intercept-gradient space, however stack rotations can
also be computed using near and far (angle stack)
Figure 7(a) shows the optimal fluid and lithology stack,
using an AVA rotation angle of 52. Note however, that also be computed using near and far (angle stack)
attributes. Near and far AVA attributes have both high
using an AVA rotation angle of 52. Note however, that
generating an AVA rotation stack at this angle, while attributes. Near and far AVA attributes have both high
signal-to-noise ratios and statistical independence
generating an AVA rotation stack at this angle, while
maximising fluid and lithology responses, does not help signal-to-noise ratios and statistical independence
(Herrmann and Cambois, 2001). Figure 4 illustrates the
maximising fluid and lithology responses, does not help
differentiate between the effects of fluids and lithology. (Herrmann and Cambois, 2001). Figure 4 illustrates the
basic concept of AVA stack rotations. Ìn practice, a
differentiate between the effects of fluids and lithology.
An AVA rotation angle of -62 will produce a stack where basic concept of AVA stack rotations. Ìn practice, a
simple scan through rotation angles can be undertaken
An AVA rotation angle of -62 will produce a stack where
the shale and brine sand impedances are almost zero simple scan through rotation angles can be undertaken
to highlight interesting anomalies.
the shale and brine sand impedances are almost zero
compared to a relatively larger gas sand impedance. to highlight interesting anomalies. compared to a relatively larger gas sand impedance.
Figure 7(b) shows this optimal fluid stack ÷ the gas
With access to statistical rock physics and stochastic
Figure 7(b) shows this optimal fluid stack ÷ the gas
effects are very clearly seen in this AVA rotation stack.
forward modelling results, it becomes possible to
calibrate variations in AVA stack rotations to known rock calibrate variations in AVA stack rotations to known rock
Figure 4: (a) Two
AngIe
(c) (d) Figure 4: (a) Two
refIection events with
(c) (d)
refIection events with
(b) two different AVA (b) two different AVA
responses. (c) Two AngIe responses. (c) Two
sampIe weighting
AngIe
sampIe weighting
functions for stacking (a) functions for stacking
resuIting in the
(a)
resuIting in the
stacked traces shown stacked traces shown
in (d). Each weighting
AngIe
function enhances a
particuIar AVA
AngIe
AngIe
particuIar AVA
behaviour.
AngIe
behaviour.
(b) (b)
hard
(a) (c)
Shale A over Shale B
hard
R
e
I
a
t
i
v
e

E
I
a
s
t
i
c

I
m
p
e
d
a
n
c
e
Shale A over Shale B
Shale A over Gas Sand
Shale A over Brine Sand
R
e
I
a
t
i
v
e

E
I
a
s
t
i
c

I
m
p
e
d
a
n
c
e
Shale A over Brine Sand
R
e
I
a
t
i
v
e

E
I
a
s
t
i
c

I
m
p
e
d
a
n
c
e
R
e
I
a
t
i
v
e

E
I
a
s
t
i
c

I
m
p
e
d
a
n
c
e
soft
R
e
I
a
t
i
v
e

E
I
a
s
t
i
c

I
m
p
e
d
a
n
c
e
soft
R
e
I
a
t
i
v
e

E
I
a
s
t
i
c

I
m
p
e
d
a
n
c
e
high
(b) (d)
R
e
I
a
t
i
v
e

E
I
a
s
t
i
c

I
m
p
e
d
a
n
c
e
high
(b) (d)
R
e
I
a
t
i
v
e

E
I
a
s
t
i
c

I
m
p
e
d
a
n
c
e
R
e
I
a
t
i
v
e

E
I
a
s
t
i
c

I
m
p
e
d
a
n
c
e
AVA Rotation AngIe
Figure 6: AVA rotation angIe versus reIative
Iow
Gas sand wedge
Figure 6: AVA rotation angIe versus reIative
eIastic impedance for three Iitho-fIuid Iow
Gas sand wedge
between two shaIes eIastic impedance for three Iitho-fIuid
interfaces. 52   is the angIe at which the interfaces. 52   is the angIe at which the
difference between each interface is
Figure 5: (a) AI and (b) Vp/Vs of a synthetic earth modeI using
difference between each interface is
maximised (at the same time, minimising
Figure 5: (a) AI and (b) Vp/Vs of a synthetic earth modeI using
properties from a reaI fieId exampIe. (c) Near- and (d) far-angIe
  
maximised (at the same time, minimising
popuIation overIap). -62   is the angIe at
stacks generated using a 30 Hz Ricker waveIet with -90   phase.
popuIation overIap). -62   is the angIe at
which gas sand has a reIativeIy Iarger which gas sand has a reIativeIy Iarger
impedance compared to the near zero impedance compared to the near zero
impedance for the brine sand and shaIe. impedance for the brine sand and shaIe.
(a) (b) (a) (b) (a) (b) (a) (b)
Figure 7: (a) OptimaI IithoIogy and fIuid stack; (b) optimaI Figure 7: (a) OptimaI IithoIogy and fIuid stack; (b) optimaI
gas sand stack.
Figure 8: (a) OptimaI IithoIogy and fIuid impedance; (b)
optimaI gas sand impedance. gas sand stack. optimaI gas sand impedance.
AcknowIedgements: The authors wish to thank DownUnder GeoSolutions Pty Ltd, Apache Energy Ltd and AcknowIedgements: The authors wish to thank DownUnder GeoSolutions Pty Ltd, Apache Energy Ltd and
KUFPEC for their support of the research behind this presentation. KUFPEC for their support of the research behind this presentation.
Ceophysics PosLer P9 Ceophysics PosLer P9
2. Rock Physics ModeIIing 2. Rock Physics ModeIIing
V
s
d
e
p
t
h
Rock physics is the critical link between seismic data
(and their derivatives) and geology. Statistical rock
d
e
p
t
h
(and their derivatives) and geology. Statistical rock
physics work should be undertaken to quantify the physics work should be undertaken to quantify the
geophysical signatures of all rock and fluid types of geophysical signatures of all rock and fluid types of
interest any time well log data are available.
Vp Vp
interest any time well log data are available.
Vp Vp
Ìdeally, a statistical rock physics model is constructed
r
h
o
p
h
i
Ìdeally, a statistical rock physics model is constructed
via a detailed petrophysical analysis that defines depth-
r
h
o
p
h
i
via a detailed petrophysical analysis that defines depth-
dependent lithological end-member trends. An end dependent lithological end-member trends. An end
member is defined as the cleanest example of a member is defined as the cleanest example of a
lithology, and is based on interpretation (not automated
Vp Vp
lithology, and is based on interpretation (not automated
cut-offs) of available log data. Figure 2 shows the rock
Vp Vp
physics model for two different lithologies. Once the
elastic properties of end members are known, the elastic elastic properties of end members are known, the elastic
behaviour of any rock composed of different proportions
Figure 2: StatisticaI rock physics modeI for an end-
member sand (red) and shaIe (green). behaviour of any rock composed of different proportions
of these end members can be determined.
member sand (red) and shaIe (green).
of these end members can be determined.
Subsequent stochastic forward modelling facilitates the
V
p

/

V
s
Figure 3: Stochastic Subsequent stochastic forward modelling facilitates the
understanding of how seismic responses will change as
V
p

/

V
s
Figure 3: Stochastic
forward modeIIing understanding of how seismic responses will change as
a function of key variables such as depth, fluid content,
V
p

/

V
s
forward modeIIing
resuIts for four Iitho- a function of key variables such as depth, fluid content,
and reservoir quality. Ìn addition to enabling the
resuIts for four Iitho-
fIuid combinations at and reservoir quality. Ìn addition to enabling the
prediction of the most likely seismic response, statistical
fIuid combinations at
a singIe depth. The prediction of the most likely seismic response, statistical
rock physics enables the prediction of the range of
a singIe depth. The
eIIipses represent the rock physics enables the prediction of the range of
possible responses. Capturing this population
eIIipses represent the
P95 contours of the
behaviour results in a more realistic evaluation of any
seismic anomalies.
probabiIity density
functions (PDFs). seismic anomalies.
AI
functions (PDFs).
AI
4. Adding VaIue with Seismic Inversion 4. Adding VaIue with Seismic Inversion
Seismic inversion-based QÌ workflows are possible with When multiple wells have penetrated a known Seismic inversion-based QÌ workflows are possible with
access to well log data in the seismic survey area.
When multiple wells have penetrated a known
reservoir, more focused reservoir characterisation is access to well log data in the seismic survey area.
Seismic-to-well ties can be made, and relative AVA
reservoir, more focused reservoir characterisation is
possible, and stochastic inversion can be utilised to Seismic-to-well ties can be made, and relative AVA
inversion products can be generated. AVA stack rotation
possible, and stochastic inversion can be utilised to
understand key characteristics and uncertainties in
can be applied following simultaneous relative inversion
÷ this adds value through largely removing the effect of
important reservoir parameters such as net sand,
porosity, fluid saturation etc (Glinsky et al., 2005). ÷ this adds value through largely removing the effect of
the seismic wavelet. Figure 8 shows the same AVA
porosity, fluid saturation etc (Glinsky et al., 2005).
the seismic wavelet. Figure 8 shows the same AVA
stack rotations as seen in Figure 7, but applied after Figure 10 shows an example of how stochastic stack rotations as seen in Figure 7, but applied after
simultaneous relative inversion. The individual sand and
Figure 10 shows an example of how stochastic
inversion was used to add value to a project by simultaneous relative inversion. The individual sand and
shale units are now obvious in Figure 8(a), and the gas-
inversion was used to add value to a project by
clarifying why a gas-sand probability volume was shale units are now obvious in Figure 8(a), and the gas-
saturated sand wedge can be seen in Figure 8(b). Note
clarifying why a gas-sand probability volume was
showing a GWC below the known fluid contact depth saturated sand wedge can be seen in Figure 8(b). Note
that, it is important to appreciate that stack rotation
showing a GWC below the known fluid contact depth
(derived via drilling). Model-based stochastic that, it is important to appreciate that stack rotation
angles are depth dependent. For example, AVA
(derived via drilling). Model-based stochastic
inversion was used to specifically invert for the fluid angles are depth dependent. For example, AVA
responses can change with depth as a function of
inversion was used to specifically invert for the fluid
contact depth. The prior model was built with the responses can change with depth as a function of
different compaction gradients between sands and
contact depth. The prior model was built with the
known contact depth, however uncertainty was placed different compaction gradients between sands and
shales. Ìt is necessary to take this into consideration
known contact depth, however uncertainty was placed
on the contact to allow the inversion to move the shales. Ìt is necessary to take this into consideration
when doing reservoir characterisation over large time
on the contact to allow the inversion to move the
contact if required to fit the seismic data. The most when doing reservoir characterisation over large time
windows.
contact if required to fit the seismic data. The most
likely post inversion model is shown in Figure 10(a) ÷
Where well control is sufficient, it becomes possible to
the inversion did move the contact deeper to fit the
seismic observations. Figure 10(b) shows the Where well control is sufficient, it becomes possible to
construct low-frequency models and generate absolute
seismic observations. Figure 10(b) shows the
posterior uncertainty of the contact is minimal where construct low-frequency models and generate absolute
rock property derivatives. These inverted rock properties
posterior uncertainty of the contact is minimal where
the contact was moved deeper. Subsequent depth rock property derivatives. These inverted rock properties
can then be integrated with depth-dependent, stochastic
the contact was moved deeper. Subsequent depth
conversion work demonstrated that both the can then be integrated with depth-dependent, stochastic
rock physics models using a Bayesian classification
conversion work demonstrated that both the
simultaneous inversion and stochastic inversion were rock physics models using a Bayesian classification
scheme to make quantitative predictions about lithology
simultaneous inversion and stochastic inversion were
correct ÷ there was gas in the sand under question; scheme to make quantitative predictions about lithology
and fluids away from well locations (Lamont, et al.,
correct ÷ there was gas in the sand under question;
but an erroneous depth conversion model was and fluids away from well locations (Lamont, et al.,
2008). Figure 9 shows a gas-sand probability volume
but an erroneous depth conversion model was
pushing the anomaly below the physical fluid contact. 2008). Figure 9 shows a gas-sand probability volume
derived using this methodology. Ìn this example,
pushing the anomaly below the physical fluid contact.
derived using this methodology. Ìn this example,
seismic inversion added significant value to the field as Gas
(a)
seismic inversion added significant value to the field as
a result of the QÌ results providing confidence to drill
Gas
Saturation
high
(a)
a result of the QÌ results providing confidence to drill
down-dip of a dry hole to make a gas discovery. This
SHALE
high
was the first of around a dozen successful new
discovery and appraisal wells drilled on this seismic D
e
p
t
h
discovery and appraisal wells drilled on this seismic
inversion product.
SAND D
e
p
t
h
inversion product.
PRIOR
CONTACT Iow
Discovery WeII Dry WeII
CONTACT Iow
high high
Posterior depth Posterior depth
uncertainty of
fIuid contact
High gas statistics based on High gas
probability in red
statistics based on
1000 posterior
iterations
Iow (b)
iterations
Figure 9: Gas-sand probabiIity voIume derived from
Iow (b)
Figure 9: Gas-sand probabiIity voIume derived from
simuItaneous absoIute inversion and prediction of
Figure 10: (a) Most IikeIy posterior reservoir modeI
simuItaneous absoIute inversion and prediction of
fIuids using the depth-dependent, stochastic rock
derived via stochastic inversion, showing gas saturated
sand; and (b) posterior uncertainty in the depth of the
fIuids using the depth-dependent, stochastic rock
physics modeI and a Bayesian cIassification scheme.
sand; and (b) posterior uncertainty in the depth of the
fIuid contact.
physics modeI and a Bayesian cIassification scheme.
fIuid contact.
5. Summary
can be generated, dependent on well control, and can
5. Summary
can be generated, dependent on well control, and can
be integrated with rock physics models to predict be integrated with rock physics models to predict
lithology and fluids away from well locations
The most effective integrated QÌ studies for reservoir
lithology and fluids away from well locations
delineation and characterisation are fit-for-purpose.
Appropriate QÌ workflows should be designed on a case- 6. References Appropriate QÌ workflows should be designed on a case-
by-case basis, according to the geological and
6. References
by-case basis, according to the geological and
geophysical information that is available. Glinsky, M.E., Asher, B., Hill, R., Flynn, M., Stanley, M., geophysical information that is available. Glinsky, M.E., Asher, B., Hill, R., Flynn, M., Stanley, M.,
Gunning, J., Thompson, T., Kalifa, J., Mallat, S., White,
Statistical rock physics models derived from well log
Gunning, J., Thompson, T., Kalifa, J., Mallat, S., White,
C., and Renard, D., 2005. Ìntegration of uncertain Statistical rock physics models derived from well log
data are important for understanding seismic responses
C., and Renard, D., 2005. Ìntegration of uncertain
subsurface information into multiple reservoir simulation data are important for understanding seismic responses
and their uncertainty, and should always be derived as
subsurface information into multiple reservoir simulation
models. The Leading Edge. 24(10): 990-999. and their uncertainty, and should always be derived as
part of a QÌ workflow whenever log data are available.
models. The Leading Edge. 24(10): 990-999.
part of a QÌ workflow whenever log data are available.
Herrmann, P., and Cambois, G., 2001. Statistically
AVA attributes are a powerful tool for gaining insight into
Herrmann, P., and Cambois, G., 2001. Statistically
uncorrelated AVO attributes. EAGE 63
rd
Conference AVA attributes are a powerful tool for gaining insight into
the hydrocarbon reservoir even when wells are not
uncorrelated AVO attributes. EAGE 63 Conference
and Technical Exhibition. P187. the hydrocarbon reservoir even when wells are not
present in the seismic survey area. A modified approach
and Technical Exhibition. P187.
present in the seismic survey area. A modified approach
to AVA stack rotations presented here works in the near
versus far space, and can take advantage of any
Lamont, M.G., Thompson, T.A., and Bevilacqua, C.,
versus far space, and can take advantage of any
statistical rock physics information to optimise the
Lamont, M.G., Thompson, T.A., and Bevilacqua, C.,
2008. Drilling success as a result of probabilistic
statistical rock physics information to optimise the
discrimination between fluid- and lithology-related
lithology and fluid prediction: A case study in the
Carnarvon Basin, W.A.. APPEA Journal 48(1). discrimination between fluid- and lithology-related
seismic amplitude anomalies.
Carnarvon Basin, W.A.. APPEA Journal 48(1).
seismic amplitude anomalies.
Whitcombe, D.N., Connolly, P.A., Reagan, R.L., and
When well log data are available, seismic inversion can
Whitcombe, D.N., Connolly, P.A., Reagan, R.L., and
Redshaw, T.C.,2002. Extended elastic impedance for When well log data are available, seismic inversion can
be performed. Relative or absolute inversion products
Redshaw, T.C.,2002. Extended elastic impedance for
fluid and lithology prediction. Geophysics. 67(1): 63-67. be performed. Relative or absolute inversion products fluid and lithology prediction. Geophysics. 67(1): 63-67.

Anda mungkin juga menyukai