Transcript Seismic reservoir characterization constrained by well log rock

Combining statistical rock physics and
sedimentology to reduce uncertainty in
seismic reservoir characterization
Per Åge Avseth
Norsk Hydro Research Centre
FORCE seminar, June 5 - 2001, Stavanger
Seismic reservoir characterization Problem statements
 Where are the good quality sands?
 Where is the oil?
 What are the uncertainties?
Outline
 Link rock physics to sedimentary facies
Well log interpretation, rock physics modeling, cross-plot analysis
 Quantify the uncertainty in seismic response related to
facies and fluid variability
Facies classification, MC-simulation, seismic modeling
 Predict most likely facies and fluids and their occurrence
probabilities from seismic amplitudes
3-D AVO inversion, seismic reservoir mapping
Why facies?
Well
Vertical stratigraphy
Depositional system
Clean sand
Shaly sand
Shale
 Facies = characteristic sedimentary units that occur in
predictable patterns/geometries within a depositional system.
 Walther’s law: Facies that are conformably stacked on top
of each other, can also be found laterally to each other.
FACIES I:
Gravels and conglomerates
I
FACIES II:
II
Thick-bedded sandstone
FACIES III:
Interbedded sandstone-shale
III
FACIES IV:
Silty shale and
silt-laminated shale
FACIES V:
Pure shale
FACIES VI:
Chaotic deposits
IV
V
VI
Identification of seismic lithofacies
from well logs (= training data)
V
V
2100
IV
II b
2150
Depth (m)
Depth (m)
2100
IV
II b
2150
II c
III
2200
40
60
80 100 120 140
Gamma Ray (API)
II c
III
2200
1.5
2
2.5
3
Vp (km/s)
3.5
4
Seismic properties
3
V
6.5
IIc
IIb
2.5
III
Vp/Vs
Acoustic Impedance
7.5
IV
5.5
4.5
40
IV
III
2
V
60
80
100
120
Gamma Ray (API)
IIb
140
1.5
40
IIc
60
80
100
120
Gamma Ray (API)
140
Seismic lithofacies classification
(quadratic discriminant analysis)
Well
#1
Well #1
Well
#2 2
Well#
Well
#3 3
Well #
2100
V
IV
III
Dept h ( m)
2200
IIc
IIb
IIa
2300
M  x  μ i   i
2
T
1
x  μ i 
Validation analysis of training data
Success-rate (%)
100
50
82 % Vp and GR
60 % Vp only
45 % GR only
0
IIa
IIb
IIc
III
IV
V
Cum. Frequency
Cumulative distribution functions
of seismic properties (brine facies)
1.0
IIb
V
0.5
IIa
IIc
00
Cum. Frequency
III
IV
5.0
5.5
6.0
6.5
Acoustic Impedance
7.0
1.0
IIa
IIb
0.5
IIc
III
IV
V
0
1.8
2.0
2.2
2.4
Vp/Vs-ratio
2.6
2.8
1.0
IIb
0.5
0.5
Cum. Frequency
Cum. Frequency
Cumulative distribution functions:
Expanding data base to include oil facies
00
IIb-oil
4.0
4.5
IIc
IIc-oil
IIa
IIa-oil
6.0
6.5
5.0
5.5
Acoustic Impedance
7.0
1.0
IIa-oil
IIb
0.5
0
IIa
IIb-oil
1.6
IIc
IIc-oil
1.8
2.0
2.2
Vp/Vs-ratio
2.4
2.6
Monte-Carlo simulation and
AVO probabilities
R ()  R (0)  G  sin 2 
1  V  

R (0)   P 
2  VP
 
2
VS  
V 
1 VP
G
 2 2 
 2 S 
2 VP
VS 
VP  
Bivariate pdfs: R(0) vs. G
Bivariate pdfs: R(0) vs. G
(Grouped facies)
3-D seismic topography of
top reservoir horizon (travel-time)
Turbidite system outline
Feeder-channel
Lobe-structure
100ms
1km
R(0) at Top Reservoir horizon
(3D seismic topography)
Relatively high R(0)
(blue)
Relatively low R(0)
(yellow)
G at Top Reservoir horizon
(3D seismic topography)
Relatively large negative
gradient G (yellow)
Relatively small
gradient (blue)
Seismic lithofacies prediction
(3D seismic topography)
Interbedded
sand-shales
Oil sands
Brine sands
Shale
Seismic lithofacies prediction
(map-view)
Facies probability maps
Limitations of methodology
 Tuning and thin-bed effects
 Noise and processing effects
 Lateral velocity trends in overburden
 Cap-rock assumptions
 Representative statistics?
 Seismic interpretation of top reservoir
Conclusions
• ROCK PHYSICS INNOVATIONS: The link between rock physics and
sedimentology gives better understanding of depositional systems
from seismic amplitudes.
• INTEGRATED METHODOLOGY: Based on statistical rock physics
and facies classification we can create probability maps of facies and
pore fluids from seismic inversion results.
• BUSINESS IMPLICATIONS: Facies probability maps can be used as
inputs for various decision and risk analyses in hydrocarbon
exploration and reservoir development.
The road ahead: Extended integration
• GEOSTATISTICS: Expand on one-point uncertainty analysis and
include spatial statistics (Eidsvik et al., 2001)
• RESERVOIR MODELING: Apply the methodology to constrain
reservoir modeling, flow simulation and production forecasting
(Caers et al., 2001).
• SEISMIC IMAGING: Include other uncertainties related to
seismic processing, tuning effects, overburden, anisotropy, etc.
End quote
“The language of probability allows us to
speak quantitatively about some situation
which may be highly variable, but which does
have some consistent average behavior....
Our most precise description of nature must
be in terms of probabilities.”
- RICHARD FEYNMAN