AB

Quantitative Reservoir Characterization

using Seismic Data

Amir Abbas Babasafari

May 2022
Babasafari et al., 2020 1
PRESENTATION OUTLINE

A. Seismic Data Application

B. Seismic Data Limitation

C. Case Studies Selected Based on my Recent Published Research

1. Broadening Seismic Frequency Bandwidth
2. Integrating Rock-Physics Modeling and Stochastic Seismic Inversion
3. Lithology Dependent Anisotropic AVO Correction
4. Petro Elastic Modeling in Carbonate Reservoir
5. Petrophysical Seismic Inversion
6. 4D Seismic Inversion to Monitor Water Saturation and Pore Pressure
7. Machine Learning Application for Lithofacies Classification
8. Machine Learning Application for Fault and Fracture Study

2
A. Seismic Data Application 1) Seismic Data Structural Interpretation
2) Seismic Reservoir Characterization (Qualitative and Quantitative)

• Primary variable • Secondary variable

• Hard data • Soft data
• High resolution vertically • Low resolution vertically
• Limited coverage horizontally • Wide coverage horizontally

Incorporating well and seismic data for reservoir characterization

(An example of Malay Basin)
3
Why using seismic data

Well 1 Well 2 Well 3

In Field Development:

Example Field Study

• Water breakthrough problems

in all 3 wells

• Decision made to inject water

in well 2 to stimulate
After Weber et al., 1995
production in well 3

Grainstone distribution

4
Why using seismic data

Well 1 Well 2 Well 3

Wrong decision because:

• Original correlation based on

lithostratigraphy

• New correlation based on

chronostratigraphy

After Weber et al., 1995

Grainstone distribution

5
B. Seismic Data Limitation

• Acquisition and processing related artifacts (acquisition footprints, surface multiples,

sideswipe, coherent noise, seismic wave attenuation, migration problems etc.)
• Velocity anomalies which may cause pull-ups, push-downs or ray path distortion

6
Seismic Data Limitation

• Seismic Vertical Resolution results in upscaling well data and loss of vertical
resolution
• Non-uniqueness in converting seismic data to reservoir properties between wells
• Seismic anisotropic effects on reservoir characterization

7
Reservoir Modeling Workflow

Spectral
Enhancement

Seismic Anisotropic
Stochastic Bayesian
Pre-stack AVO Uncertainty
Elastic Probability
Amplitude Data Correction Quantification
Inversion Classification

Well logs & Application in

cores Hydrocarbon
Prediction

Petro Elastic Petrophysical

Geological
Modeling Properties
facies
Prediction

Petrophysical
Probability
Evaluation distribution of
different pore
types Static Reservoir
Model & Dynamic Reservoir
Uncertainty Model
Seismic analysis
Interpretation
Horizons and Cross validation by
Faults history matching of Optimizing field
production data development plan

Economic
Evaluation
8
Seismic Data Inversion
Seismic inversion is the technique of quantifying seismic data and
converting it to elastic properties such as acoustic impedance.

Adapted from Hampson-Russell software

Seismic Forward Modelling 9
C. Reservoir Characterization using Seismic Data

10
Babasafari et al.,2020
Seismic vertical resolution

Courtesy of Jason

Courtesy of CGG

11
1. Broadening Seismic Frequency Bandwidth by
Blueing Reflectivity Integration (Qualitative
Interpretation)

BRI seismic section reveals sub seismic geological features compared to original seismic data ; (a), (b) and
(c) locations are representing thin channel sediments and (d) demonstrates fault with minor displacement

12
 Seismic Stochastic Inversion

Seismic inversion is a non-unique problem (Tarantola, 1987)

Geostatistical seismic inversion (Haas and Dubrule, 1994)
Bayesian linearized AVO inversion (Buland and Omre,2003)
Bayesian AVO inversion + SGS (Doyen, 2007)
AVO inversion + MCMC

Schematic illustration of stochastic inversion workflow Pereira, 2017; CGG

(Delbecq and Moyen, 2010)

13
 Ranking analysis and
Uncertainty Quantification

Courtesy of SPE Courtesy of Jason

14
2. Integrating Petro-Elastic Modeling, Stochastic Inversion
and Bayesian Classification

Deterministic Stochastic

15
Litho-facies classification
 11000
Average Acoustic Impedance map
2. Integrating Petro-Elastic Modeling, Stochastic Inversion
and Bayesian Classification

Vp/Vs

AI
6000 Blind well
2.75
Average Vp/Vs map
D
Vp/Vs

7m
C 11 m

AI 16
1.65
2. Integrating Petro-Elastic Modeling, Stochastic Inversion
and Bayesian Classification

Probability cross-section of hydrocarbon distribution; using measured

and predicted elastic logs

17
2. Integrating Petro-Elastic Modeling, Stochastic Inversion
and Bayesian Classification

Hydrocarbon distribution probability map at reservoir interval Extracted probable hydrocarbon

superimposed by structure contour map zones as Geo-Body
18
Seismic anisotropic effect on
reservoir characterization

Seismic anisotropy is velocity dependency on the direction of wave propagation, however the angular
dependence of velocity influences the reflection amplitude changes with offset (AVO).

(a) Schematic illustration of VTI media in presence of shale and sand layers
and (b) Graphical representation of wave front, Anisotropy mode in red and Isotropy mode in blue (after Thomsen, 1986 )

19
 Slide 20

Seismic anisotropic effect on

reservoir characterization
Seismic anisotropy effect on amplitude

𝛥𝛿 𝛥𝜀
𝑅𝐵 𝜃 = 𝐴 + (𝐵 + ) sin2 𝜃 + ( 𝐶 + ) sin2 𝜃 tan2 𝜃
2 2

𝑅𝐴 𝜃 = 𝐴 + 𝐵 + 𝐶

𝛥𝛿 𝛥𝜀
𝑅𝐵 𝜃 = 𝑅𝐴 𝜃 + sin2 𝜃 + sin2 𝜃 tan2 𝜃
2 2
 R B θ = P-P reflection coefficient derived from original seismic data
𝛥𝛿 = 𝛿2 − 𝛿1
 Value and sign of anisotropic parameters𝛥𝜀
differences
= 𝜀 −(Δδ𝜀 and Δε) at reflection interfaces between two layers
2 1

Seismic anisotropy effect on velocity  Corresponding incidence angle θ for each angle stack (near, middle and far)
1 𝛥𝑉𝑝0 𝛥𝜌
3. Lithology Dependent Anisotropic
AVO Correction

Shear Impedance log of well D

(blind well) in seismic equivalent
resolution superimposed on Shear
Impedance results before (left) and
after (right) amplitude correction

Overlay comparison of S-Impedance logs at well location D (blind well),

filtered measured (green), before (red) and after (blue) anisotropy amplitude
correction; S-Impedance is corrected under influence of anisotropy more
than P-Impedance, Black box represents reservoir interval

21
Petro Elastic Modeling (PEM)

Input:
Output:
PEM
• Mineral volumes
• Porosity • Compressional velocity
• Fluid saturation and fluid • Shear velocity
density • Density

Minerals

• Solid Rock Modulus (Bulk and Shear) calculation

• Computing the Dry Rock Modulus (Bulk and Shear)
Pore
• Gassmann Fluid Substitution to obtain Saturated Rock Modulus
• Elastic properties prediction (Density, P-wave velocity, S-wave velocity)

Fluid

22
Petro Elastic Modeling (PEM)

The elastic properties of a rock mainly depends on:

• Rock composition (minerals and fluids)

• Rock texture (spatial distribution of minerals and pores)

Petro-Elastic Models:

• Empirical models includes Gardner and Greenberg-Castagna models

• Effective Medium models consists of Granular model and Inclusion model

• Hybrid models

Granular model
in Clastic rocks

23
 4. Petro Elastic Modeling in Carbonate Reservoir

(a) Graphical definition of pore aspect ratio. (b) Crossplot of P-wave velocity
versus porosity showing the influence of pore aspect ratio in carbonate
sediments. (Pore aspect ratio contours range between the Reuss and Voigt
averages.) Increase in the aspect ratio of pores (A, B, and C) increases P-wave Normalized dry rock bulk modulus versus porosity labeling with contours of constant pore space stiffness
velocity for a given porosity (Russell and Smith, 2007). and color coded by volume fraction of Pore-types, a) Stiff pore, b) Interparticle pore and c) Micro
24 pore
together with corresponding photomicrograph and distribution histogram of each pore-type
4. Petro Elastic Modeling in Carbonate Reservoir

a) A cross section passing through Wells A and B representing posterior probability of getting stiff pore-
type with 25% of total porosity volume overlaid by water saturation logs for reservoir interval
identification. b) The corresponding average probability map of getting stiff pore-type with 25% of total
porosity volume at reservoir interval.
25
 Reservoir properties prediction

• Empirical equations
• Geostatistical methods
• Petrophysical seismic inversion
• Co-simulation after stochastic inversion
• Machine learning and deep learning
• Hybrid techniques

Reservoir properties estimation by solving an inverse problem

(Ghosh and Babasafari, 2019)

26
 Slide 27

Petrophysical Seismic Inversion

Coléou et al., 2005

 Slide 28

5. Petrophysical Seismic Inversion

3D Seismic Data Initial Petrophysical Model

(Porosity)

Seismic Stochastic inversion

Elastic Model from seismic Elastic Model from PEM

(Inverted Acoustic (Modeled Acoustic
Impedance) Impedance)
Petro-Elastic Modeling
per each litho-facies class

Matching
Process
No Minimizing misfit
Yes
function
Final Petrophysical
Model
(Porosity)
5. Petrophysical Seismic Inversion

Porosity

a) Litho-facies classification. b) predicted porosity using PEI method and c) Well log porosity
superimposed on predicted porosity
29
5. Petrophysical Seismic Inversion

Porosity

Clay Volume

Water Saturation30
PETROLEUM GEOSCIENCE DEPARTMENT
 6. 4D Seismic Inversion to Monitor Water
Saturation and Pore Pressure

Pay zone probability cross-section in NW-SE direction at reservoir interval (a)

baseline survey and (b) monitor survey; it shows that the pay zone probability
is reduced in the injected water zone. (c) Baseline survey and (d) monitor
survey; it shows the pay zone probability increased in an unswept oil area.
31
6. 4D Seismic Inversion to Monitor Water
Saturation and Pore Pressure

Difference map of predicted water saturation between baseline and monitor at reservoir interval (a) increased
water saturation (positive values) representing injected water zones and (b) decreased water saturation (negative
32
values) representing bypassed oil zones.
 7. Machine Learning Application for
Lithofacies Classification

Lithofacies log in seismic equivalent resolution (target log) together with seismic attributes
extracted at well A used as features for training in a machine learning classifier

Cross-plots of two selected features, seismic amplitude versus inverted acoustic impedance,
at well locations color-coded with identified lithofacies classes showing the efficiency level
of classification for multiple machine learning classifiers 33
 7. Machine Learning Application for
Lithofacies Classification

(a) The classification accuracy calculated using actual and predicted lithofacies classes
within testing points for various machine learning algorithms, (b) Confusion matrix, (c)
Evaluation metrics (precision) for multi-class classification based on random forest
classifier

(a) Seismic amplitude overlaid with synthetic seismogram logs, (b) Inverted acoustic
impedance overlaid with measured acoustic impedance logs, and (c) Predicted lithofacies
classes overlaid with identified lithofacies logs 34
 8. Machine Learning Application for Fault
and Fracture Study

Interpreted fractures dip and azimuth together with the conventional well logs at well W6; (a)
Dip angle of faults and natural fractures, (b) Rose diagram of the fault and natural fractures
showing strike azimuth, (c) Dip angle of breakout and drilling induced fractures, (d) Rose
(a) The selected seismic attributes as input data for training in the supervised classification
diagram of breakout and drilling induced fractures indicating strike azimuth, (e) Gamma-ray log,
technique, (b) Supervised discontinuity attribute map extracted at reservoir top showing 2
(f) Density log, (g) P-sonic log, (h) Acoustic image log.
classes; non-fracture (0) and fault/fracture (1), (c) Semblance attribute section, (d) Supervised
discontinuity attribute section.
35
 8. Machine Learning Application for Fault
and Fracture Study

Advance method:
Amplitude variation with angle and azimuth study (AVAZ)
Shear wave splitting analysis using multi-component

Fracture density map (P20) at reservoir top (a) Interpreted borehole image logs, (b) Supervised discontinuity attribute, (c) Integrating a and b
through the geostatistical approach, (d) the thickness map between, (e) Fracture density map (P20) at reservoir top superimposed on the 36
structure contour map to show structural high and low in the study area
The book is appropriate for researchers and students those who are interested in geophysics and
petroleum geoscience. 'Geological reservoir modeling and seismic reservoir monitoring' is the
title of Chapter 5, which is specifically prepared for reservoir modeling.
37
38
Thanks for your kind attention

Amir Abbas Babasafari (AB)

https://www.linkedin.com/in/amir-abbas-babasafari-41a34741/
https://www.researchgate.net/profile/Amir_Babasafari3
https://scholar.google.com/citations?user=vSMX3h4AAAAJ&hl=en
https://orcid.org/0000-0001-9904-9434
https://www.scopus.com/authid/detail.uri?authorId=56369064100
https://github.com/AB-Blue
https://www.youtube.com/channel/UCG4H4VJljE4SZwbxb9Lqo2w

E-mail: [email protected]; [email protected]

39